{
  "best_metric": 0.160739004611969,
  "best_model_checkpoint": "./results_t5_mixed_wiki_cv_arhiv_large/checkpoint-19699",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 59097,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00025381999086248035,
      "grad_norm": 28009.150390625,
      "learning_rate": 1.9998307866727585e-05,
      "loss": 16.9957,
      "step": 5
    },
    {
      "epoch": 0.0005076399817249607,
      "grad_norm": 8431.59375,
      "learning_rate": 1.999661573345517e-05,
      "loss": 15.242,
      "step": 10
    },
    {
      "epoch": 0.000761459972587441,
      "grad_norm": 13960.4951171875,
      "learning_rate": 1.9994923600182752e-05,
      "loss": 14.3643,
      "step": 15
    },
    {
      "epoch": 0.0010152799634499214,
      "grad_norm": 9187.435546875,
      "learning_rate": 1.9993231466910336e-05,
      "loss": 13.9964,
      "step": 20
    },
    {
      "epoch": 0.0012690999543124015,
      "grad_norm": 11192.2958984375,
      "learning_rate": 1.999153933363792e-05,
      "loss": 13.0181,
      "step": 25
    },
    {
      "epoch": 0.001522919945174882,
      "grad_norm": 10066.05859375,
      "learning_rate": 1.9989847200365503e-05,
      "loss": 12.9025,
      "step": 30
    },
    {
      "epoch": 0.0017767399360373623,
      "grad_norm": 7277.57373046875,
      "learning_rate": 1.9988155067093086e-05,
      "loss": 12.5301,
      "step": 35
    },
    {
      "epoch": 0.002030559926899843,
      "grad_norm": 157497.796875,
      "learning_rate": 1.998646293382067e-05,
      "loss": 11.712,
      "step": 40
    },
    {
      "epoch": 0.0022843799177623227,
      "grad_norm": 2509.368408203125,
      "learning_rate": 1.9984770800548254e-05,
      "loss": 11.4352,
      "step": 45
    },
    {
      "epoch": 0.002538199908624803,
      "grad_norm": 9022.3505859375,
      "learning_rate": 1.9983078667275837e-05,
      "loss": 10.8293,
      "step": 50
    },
    {
      "epoch": 0.0027920198994872835,
      "grad_norm": 2620.181640625,
      "learning_rate": 1.998138653400342e-05,
      "loss": 10.5957,
      "step": 55
    },
    {
      "epoch": 0.003045839890349764,
      "grad_norm": 4354.4189453125,
      "learning_rate": 1.9979694400731e-05,
      "loss": 10.201,
      "step": 60
    },
    {
      "epoch": 0.003299659881212244,
      "grad_norm": 2636.816162109375,
      "learning_rate": 1.9978002267458588e-05,
      "loss": 9.8294,
      "step": 65
    },
    {
      "epoch": 0.0035534798720747245,
      "grad_norm": 2717.7568359375,
      "learning_rate": 1.997631013418617e-05,
      "loss": 9.7655,
      "step": 70
    },
    {
      "epoch": 0.003807299862937205,
      "grad_norm": 1577.282958984375,
      "learning_rate": 1.9974618000913755e-05,
      "loss": 9.2661,
      "step": 75
    },
    {
      "epoch": 0.004061119853799686,
      "grad_norm": 1305.2647705078125,
      "learning_rate": 1.997292586764134e-05,
      "loss": 8.5601,
      "step": 80
    },
    {
      "epoch": 0.004314939844662166,
      "grad_norm": 1029.8795166015625,
      "learning_rate": 1.997123373436892e-05,
      "loss": 8.3479,
      "step": 85
    },
    {
      "epoch": 0.0045687598355246455,
      "grad_norm": 488.6922302246094,
      "learning_rate": 1.9969541601096505e-05,
      "loss": 7.9367,
      "step": 90
    },
    {
      "epoch": 0.004822579826387126,
      "grad_norm": 1308.7169189453125,
      "learning_rate": 1.996784946782409e-05,
      "loss": 7.6314,
      "step": 95
    },
    {
      "epoch": 0.005076399817249606,
      "grad_norm": 1560.9044189453125,
      "learning_rate": 1.996615733455167e-05,
      "loss": 7.2683,
      "step": 100
    },
    {
      "epoch": 0.0053302198081120866,
      "grad_norm": 1124.25390625,
      "learning_rate": 1.9964465201279256e-05,
      "loss": 6.7153,
      "step": 105
    },
    {
      "epoch": 0.005584039798974567,
      "grad_norm": 1180.7884521484375,
      "learning_rate": 1.9962773068006836e-05,
      "loss": 6.4866,
      "step": 110
    },
    {
      "epoch": 0.005837859789837047,
      "grad_norm": 156.94796752929688,
      "learning_rate": 1.996108093473442e-05,
      "loss": 5.8496,
      "step": 115
    },
    {
      "epoch": 0.006091679780699528,
      "grad_norm": 256.8822326660156,
      "learning_rate": 1.9959388801462007e-05,
      "loss": 5.4725,
      "step": 120
    },
    {
      "epoch": 0.006345499771562008,
      "grad_norm": 242.91868591308594,
      "learning_rate": 1.9957696668189587e-05,
      "loss": 5.0473,
      "step": 125
    },
    {
      "epoch": 0.006599319762424488,
      "grad_norm": 140.21670532226562,
      "learning_rate": 1.9956004534917174e-05,
      "loss": 4.739,
      "step": 130
    },
    {
      "epoch": 0.006853139753286969,
      "grad_norm": 2685.30712890625,
      "learning_rate": 1.9954312401644754e-05,
      "loss": 4.309,
      "step": 135
    },
    {
      "epoch": 0.007106959744149449,
      "grad_norm": 51.844295501708984,
      "learning_rate": 1.9952620268372337e-05,
      "loss": 4.074,
      "step": 140
    },
    {
      "epoch": 0.007360779735011929,
      "grad_norm": 15.273852348327637,
      "learning_rate": 1.9950928135099924e-05,
      "loss": 3.4298,
      "step": 145
    },
    {
      "epoch": 0.00761459972587441,
      "grad_norm": 17.046186447143555,
      "learning_rate": 1.9949236001827505e-05,
      "loss": 3.1516,
      "step": 150
    },
    {
      "epoch": 0.00786841971673689,
      "grad_norm": 13.263907432556152,
      "learning_rate": 1.9947543868555088e-05,
      "loss": 2.6151,
      "step": 155
    },
    {
      "epoch": 0.008122239707599371,
      "grad_norm": 17.467811584472656,
      "learning_rate": 1.994585173528267e-05,
      "loss": 2.4349,
      "step": 160
    },
    {
      "epoch": 0.00837605969846185,
      "grad_norm": 26.78594398498535,
      "learning_rate": 1.9944159602010255e-05,
      "loss": 2.1512,
      "step": 165
    },
    {
      "epoch": 0.008629879689324332,
      "grad_norm": 7.074516296386719,
      "learning_rate": 1.994246746873784e-05,
      "loss": 1.9379,
      "step": 170
    },
    {
      "epoch": 0.008883699680186811,
      "grad_norm": 6.19895601272583,
      "learning_rate": 1.9940775335465422e-05,
      "loss": 1.7137,
      "step": 175
    },
    {
      "epoch": 0.009137519671049291,
      "grad_norm": 7.024235725402832,
      "learning_rate": 1.9939083202193006e-05,
      "loss": 1.6583,
      "step": 180
    },
    {
      "epoch": 0.009391339661911772,
      "grad_norm": 5.285205841064453,
      "learning_rate": 1.993739106892059e-05,
      "loss": 1.5457,
      "step": 185
    },
    {
      "epoch": 0.009645159652774252,
      "grad_norm": 4.7672600746154785,
      "learning_rate": 1.9935698935648173e-05,
      "loss": 1.423,
      "step": 190
    },
    {
      "epoch": 0.009898979643636733,
      "grad_norm": 4.843871116638184,
      "learning_rate": 1.9934006802375756e-05,
      "loss": 1.4004,
      "step": 195
    },
    {
      "epoch": 0.010152799634499212,
      "grad_norm": 3.7040324211120605,
      "learning_rate": 1.993231466910334e-05,
      "loss": 1.3773,
      "step": 200
    },
    {
      "epoch": 0.010406619625361694,
      "grad_norm": 3.55208420753479,
      "learning_rate": 1.9930622535830924e-05,
      "loss": 1.287,
      "step": 205
    },
    {
      "epoch": 0.010660439616224173,
      "grad_norm": 4.512004852294922,
      "learning_rate": 1.9928930402558507e-05,
      "loss": 1.215,
      "step": 210
    },
    {
      "epoch": 0.010914259607086654,
      "grad_norm": 3.2358570098876953,
      "learning_rate": 1.992723826928609e-05,
      "loss": 1.1807,
      "step": 215
    },
    {
      "epoch": 0.011168079597949134,
      "grad_norm": 2.9898245334625244,
      "learning_rate": 1.9925546136013674e-05,
      "loss": 1.1454,
      "step": 220
    },
    {
      "epoch": 0.011421899588811615,
      "grad_norm": 3.377270221710205,
      "learning_rate": 1.9923854002741258e-05,
      "loss": 1.1211,
      "step": 225
    },
    {
      "epoch": 0.011675719579674095,
      "grad_norm": 2.5435738563537598,
      "learning_rate": 1.992216186946884e-05,
      "loss": 1.0875,
      "step": 230
    },
    {
      "epoch": 0.011929539570536576,
      "grad_norm": 2.505054473876953,
      "learning_rate": 1.9920469736196425e-05,
      "loss": 1.0262,
      "step": 235
    },
    {
      "epoch": 0.012183359561399055,
      "grad_norm": 2.3626339435577393,
      "learning_rate": 1.991877760292401e-05,
      "loss": 0.971,
      "step": 240
    },
    {
      "epoch": 0.012437179552261536,
      "grad_norm": 2.3419456481933594,
      "learning_rate": 1.9917085469651592e-05,
      "loss": 0.9605,
      "step": 245
    },
    {
      "epoch": 0.012690999543124016,
      "grad_norm": 2.3525550365448,
      "learning_rate": 1.9915393336379175e-05,
      "loss": 0.9515,
      "step": 250
    },
    {
      "epoch": 0.012944819533986497,
      "grad_norm": 2.0995123386383057,
      "learning_rate": 1.991370120310676e-05,
      "loss": 0.9169,
      "step": 255
    },
    {
      "epoch": 0.013198639524848977,
      "grad_norm": 2.3146281242370605,
      "learning_rate": 1.9912009069834343e-05,
      "loss": 0.9073,
      "step": 260
    },
    {
      "epoch": 0.013452459515711458,
      "grad_norm": 2.5654919147491455,
      "learning_rate": 1.9910316936561923e-05,
      "loss": 0.8649,
      "step": 265
    },
    {
      "epoch": 0.013706279506573937,
      "grad_norm": 2.566847562789917,
      "learning_rate": 1.990862480328951e-05,
      "loss": 0.8795,
      "step": 270
    },
    {
      "epoch": 0.013960099497436419,
      "grad_norm": 2.0294430255889893,
      "learning_rate": 1.9906932670017093e-05,
      "loss": 0.8524,
      "step": 275
    },
    {
      "epoch": 0.014213919488298898,
      "grad_norm": 1.80044424533844,
      "learning_rate": 1.9905240536744677e-05,
      "loss": 0.8287,
      "step": 280
    },
    {
      "epoch": 0.01446773947916138,
      "grad_norm": 3.0150132179260254,
      "learning_rate": 1.990354840347226e-05,
      "loss": 0.7852,
      "step": 285
    },
    {
      "epoch": 0.014721559470023859,
      "grad_norm": 1.9330862760543823,
      "learning_rate": 1.990185627019984e-05,
      "loss": 0.7896,
      "step": 290
    },
    {
      "epoch": 0.01497537946088634,
      "grad_norm": 1.4856704473495483,
      "learning_rate": 1.9900164136927427e-05,
      "loss": 0.7634,
      "step": 295
    },
    {
      "epoch": 0.01522919945174882,
      "grad_norm": 2.0876917839050293,
      "learning_rate": 1.989847200365501e-05,
      "loss": 0.8272,
      "step": 300
    },
    {
      "epoch": 0.0154830194426113,
      "grad_norm": 1.916139006614685,
      "learning_rate": 1.989677987038259e-05,
      "loss": 0.7789,
      "step": 305
    },
    {
      "epoch": 0.01573683943347378,
      "grad_norm": 1.5130165815353394,
      "learning_rate": 1.9895087737110178e-05,
      "loss": 0.7559,
      "step": 310
    },
    {
      "epoch": 0.01599065942433626,
      "grad_norm": 1.9628571271896362,
      "learning_rate": 1.9893395603837758e-05,
      "loss": 0.7496,
      "step": 315
    },
    {
      "epoch": 0.016244479415198743,
      "grad_norm": 2.4967174530029297,
      "learning_rate": 1.9891703470565345e-05,
      "loss": 0.7925,
      "step": 320
    },
    {
      "epoch": 0.01649829940606122,
      "grad_norm": 1.711539626121521,
      "learning_rate": 1.989001133729293e-05,
      "loss": 0.7409,
      "step": 325
    },
    {
      "epoch": 0.0167521193969237,
      "grad_norm": 1.7477493286132812,
      "learning_rate": 1.988831920402051e-05,
      "loss": 0.7136,
      "step": 330
    },
    {
      "epoch": 0.017005939387786183,
      "grad_norm": 4.01808500289917,
      "learning_rate": 1.9886627070748096e-05,
      "loss": 0.7446,
      "step": 335
    },
    {
      "epoch": 0.017259759378648664,
      "grad_norm": 2.3399136066436768,
      "learning_rate": 1.9884934937475676e-05,
      "loss": 0.8573,
      "step": 340
    },
    {
      "epoch": 0.017513579369511142,
      "grad_norm": 1.5367169380187988,
      "learning_rate": 1.988324280420326e-05,
      "loss": 0.6873,
      "step": 345
    },
    {
      "epoch": 0.017767399360373623,
      "grad_norm": 3.4695355892181396,
      "learning_rate": 1.9881550670930846e-05,
      "loss": 0.7075,
      "step": 350
    },
    {
      "epoch": 0.018021219351236104,
      "grad_norm": 1.5102581977844238,
      "learning_rate": 1.9879858537658427e-05,
      "loss": 0.6935,
      "step": 355
    },
    {
      "epoch": 0.018275039342098582,
      "grad_norm": 1.7649434804916382,
      "learning_rate": 1.987816640438601e-05,
      "loss": 0.6854,
      "step": 360
    },
    {
      "epoch": 0.018528859332961063,
      "grad_norm": 1.312583088874817,
      "learning_rate": 1.9876474271113594e-05,
      "loss": 0.796,
      "step": 365
    },
    {
      "epoch": 0.018782679323823544,
      "grad_norm": 1.6248362064361572,
      "learning_rate": 1.9874782137841177e-05,
      "loss": 0.6442,
      "step": 370
    },
    {
      "epoch": 0.019036499314686026,
      "grad_norm": 1.6744344234466553,
      "learning_rate": 1.9873090004568764e-05,
      "loss": 0.6433,
      "step": 375
    },
    {
      "epoch": 0.019290319305548503,
      "grad_norm": 1.5160279273986816,
      "learning_rate": 1.9871397871296344e-05,
      "loss": 0.6437,
      "step": 380
    },
    {
      "epoch": 0.019544139296410985,
      "grad_norm": 1.3933593034744263,
      "learning_rate": 1.9869705738023928e-05,
      "loss": 0.6469,
      "step": 385
    },
    {
      "epoch": 0.019797959287273466,
      "grad_norm": 1.2953556776046753,
      "learning_rate": 1.986801360475151e-05,
      "loss": 0.6602,
      "step": 390
    },
    {
      "epoch": 0.020051779278135947,
      "grad_norm": 1.387939214706421,
      "learning_rate": 1.9866321471479095e-05,
      "loss": 0.6201,
      "step": 395
    },
    {
      "epoch": 0.020305599268998425,
      "grad_norm": 1.3470954895019531,
      "learning_rate": 1.986462933820668e-05,
      "loss": 0.6238,
      "step": 400
    },
    {
      "epoch": 0.020559419259860906,
      "grad_norm": 1.760582447052002,
      "learning_rate": 1.9862937204934262e-05,
      "loss": 0.6314,
      "step": 405
    },
    {
      "epoch": 0.020813239250723387,
      "grad_norm": 1.6875195503234863,
      "learning_rate": 1.9861245071661846e-05,
      "loss": 0.6141,
      "step": 410
    },
    {
      "epoch": 0.02106705924158587,
      "grad_norm": 1.4659909009933472,
      "learning_rate": 1.985955293838943e-05,
      "loss": 0.6416,
      "step": 415
    },
    {
      "epoch": 0.021320879232448346,
      "grad_norm": 1.3127461671829224,
      "learning_rate": 1.9857860805117013e-05,
      "loss": 0.5959,
      "step": 420
    },
    {
      "epoch": 0.021574699223310827,
      "grad_norm": 1.2059125900268555,
      "learning_rate": 1.9856168671844596e-05,
      "loss": 0.6423,
      "step": 425
    },
    {
      "epoch": 0.02182851921417331,
      "grad_norm": 1.4507384300231934,
      "learning_rate": 1.985447653857218e-05,
      "loss": 0.619,
      "step": 430
    },
    {
      "epoch": 0.02208233920503579,
      "grad_norm": 1.2196464538574219,
      "learning_rate": 1.9852784405299763e-05,
      "loss": 0.6224,
      "step": 435
    },
    {
      "epoch": 0.022336159195898268,
      "grad_norm": 1.8387898206710815,
      "learning_rate": 1.9851092272027347e-05,
      "loss": 0.5911,
      "step": 440
    },
    {
      "epoch": 0.02258997918676075,
      "grad_norm": 0.9832326769828796,
      "learning_rate": 1.984940013875493e-05,
      "loss": 0.5794,
      "step": 445
    },
    {
      "epoch": 0.02284379917762323,
      "grad_norm": 1.2416331768035889,
      "learning_rate": 1.9847708005482514e-05,
      "loss": 0.5693,
      "step": 450
    },
    {
      "epoch": 0.02309761916848571,
      "grad_norm": 1.0847058296203613,
      "learning_rate": 1.9846015872210097e-05,
      "loss": 0.5366,
      "step": 455
    },
    {
      "epoch": 0.02335143915934819,
      "grad_norm": 1.1567232608795166,
      "learning_rate": 1.984432373893768e-05,
      "loss": 0.5673,
      "step": 460
    },
    {
      "epoch": 0.02360525915021067,
      "grad_norm": 1.3223860263824463,
      "learning_rate": 1.9842631605665265e-05,
      "loss": 0.5835,
      "step": 465
    },
    {
      "epoch": 0.02385907914107315,
      "grad_norm": 1.5882163047790527,
      "learning_rate": 1.9840939472392848e-05,
      "loss": 0.5227,
      "step": 470
    },
    {
      "epoch": 0.024112899131935633,
      "grad_norm": 2.358612537384033,
      "learning_rate": 1.983924733912043e-05,
      "loss": 0.5583,
      "step": 475
    },
    {
      "epoch": 0.02436671912279811,
      "grad_norm": 1.134081244468689,
      "learning_rate": 1.9837555205848015e-05,
      "loss": 0.5728,
      "step": 480
    },
    {
      "epoch": 0.02462053911366059,
      "grad_norm": 1.3474904298782349,
      "learning_rate": 1.98358630725756e-05,
      "loss": 0.5207,
      "step": 485
    },
    {
      "epoch": 0.024874359104523073,
      "grad_norm": 0.914055585861206,
      "learning_rate": 1.9834170939303182e-05,
      "loss": 0.5529,
      "step": 490
    },
    {
      "epoch": 0.025128179095385554,
      "grad_norm": 1.1716861724853516,
      "learning_rate": 1.9832478806030762e-05,
      "loss": 0.5516,
      "step": 495
    },
    {
      "epoch": 0.025381999086248032,
      "grad_norm": 1.6925829648971558,
      "learning_rate": 1.983078667275835e-05,
      "loss": 0.537,
      "step": 500
    },
    {
      "epoch": 0.025635819077110513,
      "grad_norm": 4.091062545776367,
      "learning_rate": 1.9829094539485933e-05,
      "loss": 0.6366,
      "step": 505
    },
    {
      "epoch": 0.025889639067972994,
      "grad_norm": 1.6175552606582642,
      "learning_rate": 1.9827402406213513e-05,
      "loss": 0.503,
      "step": 510
    },
    {
      "epoch": 0.026143459058835476,
      "grad_norm": 1.2992746829986572,
      "learning_rate": 1.98257102729411e-05,
      "loss": 0.5253,
      "step": 515
    },
    {
      "epoch": 0.026397279049697953,
      "grad_norm": 0.9036175012588501,
      "learning_rate": 1.982401813966868e-05,
      "loss": 0.532,
      "step": 520
    },
    {
      "epoch": 0.026651099040560434,
      "grad_norm": 1.241174340248108,
      "learning_rate": 1.9822326006396267e-05,
      "loss": 0.5049,
      "step": 525
    },
    {
      "epoch": 0.026904919031422916,
      "grad_norm": 1.9662011861801147,
      "learning_rate": 1.982063387312385e-05,
      "loss": 0.4869,
      "step": 530
    },
    {
      "epoch": 0.027158739022285393,
      "grad_norm": 1.3847944736480713,
      "learning_rate": 1.981894173985143e-05,
      "loss": 0.5257,
      "step": 535
    },
    {
      "epoch": 0.027412559013147875,
      "grad_norm": 1.0714912414550781,
      "learning_rate": 1.9817249606579018e-05,
      "loss": 0.4935,
      "step": 540
    },
    {
      "epoch": 0.027666379004010356,
      "grad_norm": 1.3018203973770142,
      "learning_rate": 1.9815557473306598e-05,
      "loss": 0.4876,
      "step": 545
    },
    {
      "epoch": 0.027920198994872837,
      "grad_norm": 1.309104084968567,
      "learning_rate": 1.981386534003418e-05,
      "loss": 0.5586,
      "step": 550
    },
    {
      "epoch": 0.028174018985735315,
      "grad_norm": 1.4573897123336792,
      "learning_rate": 1.9812173206761768e-05,
      "loss": 0.4816,
      "step": 555
    },
    {
      "epoch": 0.028427838976597796,
      "grad_norm": 1.175330638885498,
      "learning_rate": 1.981048107348935e-05,
      "loss": 0.5306,
      "step": 560
    },
    {
      "epoch": 0.028681658967460277,
      "grad_norm": 3.1696276664733887,
      "learning_rate": 1.9808788940216935e-05,
      "loss": 0.6168,
      "step": 565
    },
    {
      "epoch": 0.02893547895832276,
      "grad_norm": 1.5825937986373901,
      "learning_rate": 1.9807096806944516e-05,
      "loss": 0.5071,
      "step": 570
    },
    {
      "epoch": 0.029189298949185236,
      "grad_norm": 2.6860151290893555,
      "learning_rate": 1.98054046736721e-05,
      "loss": 0.5171,
      "step": 575
    },
    {
      "epoch": 0.029443118940047718,
      "grad_norm": 1.351448655128479,
      "learning_rate": 1.9803712540399686e-05,
      "loss": 0.4982,
      "step": 580
    },
    {
      "epoch": 0.0296969389309102,
      "grad_norm": 21.1049747467041,
      "learning_rate": 1.9802020407127266e-05,
      "loss": 0.5326,
      "step": 585
    },
    {
      "epoch": 0.02995075892177268,
      "grad_norm": 93.52015686035156,
      "learning_rate": 1.980032827385485e-05,
      "loss": 0.4622,
      "step": 590
    },
    {
      "epoch": 0.030204578912635158,
      "grad_norm": 1.095617651939392,
      "learning_rate": 1.9798636140582433e-05,
      "loss": 0.457,
      "step": 595
    },
    {
      "epoch": 0.03045839890349764,
      "grad_norm": 1.1381208896636963,
      "learning_rate": 1.9796944007310017e-05,
      "loss": 0.4665,
      "step": 600
    },
    {
      "epoch": 0.03071221889436012,
      "grad_norm": 146.4381866455078,
      "learning_rate": 1.97952518740376e-05,
      "loss": 0.4723,
      "step": 605
    },
    {
      "epoch": 0.0309660388852226,
      "grad_norm": 1.255440592765808,
      "learning_rate": 1.9793559740765184e-05,
      "loss": 0.4205,
      "step": 610
    },
    {
      "epoch": 0.03121985887608508,
      "grad_norm": 1.259681224822998,
      "learning_rate": 1.9791867607492767e-05,
      "loss": 0.4506,
      "step": 615
    },
    {
      "epoch": 0.03147367886694756,
      "grad_norm": 1.818813681602478,
      "learning_rate": 1.979017547422035e-05,
      "loss": 0.4028,
      "step": 620
    },
    {
      "epoch": 0.03172749885781004,
      "grad_norm": 1.2193713188171387,
      "learning_rate": 1.9788483340947935e-05,
      "loss": 0.4232,
      "step": 625
    },
    {
      "epoch": 0.03198131884867252,
      "grad_norm": 1.1357531547546387,
      "learning_rate": 1.9786791207675518e-05,
      "loss": 0.4146,
      "step": 630
    },
    {
      "epoch": 0.032235138839535,
      "grad_norm": 37.55704879760742,
      "learning_rate": 1.97850990744031e-05,
      "loss": 0.4468,
      "step": 635
    },
    {
      "epoch": 0.032488958830397485,
      "grad_norm": 1.28550386428833,
      "learning_rate": 1.9783406941130685e-05,
      "loss": 0.4205,
      "step": 640
    },
    {
      "epoch": 0.03274277882125996,
      "grad_norm": 1.4312454462051392,
      "learning_rate": 1.978171480785827e-05,
      "loss": 0.4807,
      "step": 645
    },
    {
      "epoch": 0.03299659881212244,
      "grad_norm": 397.8225402832031,
      "learning_rate": 1.9780022674585852e-05,
      "loss": 0.423,
      "step": 650
    },
    {
      "epoch": 0.033250418802984925,
      "grad_norm": 1.1653971672058105,
      "learning_rate": 1.9778330541313436e-05,
      "loss": 0.4284,
      "step": 655
    },
    {
      "epoch": 0.0335042387938474,
      "grad_norm": 1.3684107065200806,
      "learning_rate": 1.977663840804102e-05,
      "loss": 0.4189,
      "step": 660
    },
    {
      "epoch": 0.03375805878470988,
      "grad_norm": 1.4786688089370728,
      "learning_rate": 1.9774946274768603e-05,
      "loss": 0.3919,
      "step": 665
    },
    {
      "epoch": 0.034011878775572366,
      "grad_norm": 1.0354373455047607,
      "learning_rate": 1.9773254141496186e-05,
      "loss": 0.4386,
      "step": 670
    },
    {
      "epoch": 0.03426569876643484,
      "grad_norm": 2.865295171737671,
      "learning_rate": 1.977156200822377e-05,
      "loss": 0.4476,
      "step": 675
    },
    {
      "epoch": 0.03451951875729733,
      "grad_norm": 0.9796468615531921,
      "learning_rate": 1.9769869874951354e-05,
      "loss": 0.3822,
      "step": 680
    },
    {
      "epoch": 0.034773338748159806,
      "grad_norm": 1.1535001993179321,
      "learning_rate": 1.9768177741678937e-05,
      "loss": 0.3935,
      "step": 685
    },
    {
      "epoch": 0.035027158739022284,
      "grad_norm": 1.33408522605896,
      "learning_rate": 1.976648560840652e-05,
      "loss": 0.4267,
      "step": 690
    },
    {
      "epoch": 0.03528097872988477,
      "grad_norm": 1.2834749221801758,
      "learning_rate": 1.9764793475134104e-05,
      "loss": 0.3954,
      "step": 695
    },
    {
      "epoch": 0.035534798720747246,
      "grad_norm": 0.8446087837219238,
      "learning_rate": 1.9763101341861684e-05,
      "loss": 0.3865,
      "step": 700
    },
    {
      "epoch": 0.035788618711609724,
      "grad_norm": 1.3027830123901367,
      "learning_rate": 1.976140920858927e-05,
      "loss": 0.4294,
      "step": 705
    },
    {
      "epoch": 0.03604243870247221,
      "grad_norm": 0.8204338550567627,
      "learning_rate": 1.9759717075316855e-05,
      "loss": 0.4074,
      "step": 710
    },
    {
      "epoch": 0.036296258693334686,
      "grad_norm": 1.232735276222229,
      "learning_rate": 1.975802494204444e-05,
      "loss": 0.413,
      "step": 715
    },
    {
      "epoch": 0.036550078684197164,
      "grad_norm": 1.436716079711914,
      "learning_rate": 1.9756332808772022e-05,
      "loss": 0.4133,
      "step": 720
    },
    {
      "epoch": 0.03680389867505965,
      "grad_norm": 0.9789804816246033,
      "learning_rate": 1.9754640675499602e-05,
      "loss": 0.4078,
      "step": 725
    },
    {
      "epoch": 0.037057718665922126,
      "grad_norm": 1.6822584867477417,
      "learning_rate": 1.975294854222719e-05,
      "loss": 0.4372,
      "step": 730
    },
    {
      "epoch": 0.03731153865678461,
      "grad_norm": 1.0740268230438232,
      "learning_rate": 1.9751256408954773e-05,
      "loss": 0.4162,
      "step": 735
    },
    {
      "epoch": 0.03756535864764709,
      "grad_norm": 1.06478750705719,
      "learning_rate": 1.9749564275682353e-05,
      "loss": 0.4091,
      "step": 740
    },
    {
      "epoch": 0.03781917863850957,
      "grad_norm": 0.8764176368713379,
      "learning_rate": 1.974787214240994e-05,
      "loss": 0.4438,
      "step": 745
    },
    {
      "epoch": 0.03807299862937205,
      "grad_norm": 0.9054062366485596,
      "learning_rate": 1.974618000913752e-05,
      "loss": 0.4423,
      "step": 750
    },
    {
      "epoch": 0.03832681862023453,
      "grad_norm": 1.0033007860183716,
      "learning_rate": 1.9744487875865103e-05,
      "loss": 0.4017,
      "step": 755
    },
    {
      "epoch": 0.03858063861109701,
      "grad_norm": 1.1299196481704712,
      "learning_rate": 1.974279574259269e-05,
      "loss": 0.3955,
      "step": 760
    },
    {
      "epoch": 0.03883445860195949,
      "grad_norm": 1.2052655220031738,
      "learning_rate": 1.974110360932027e-05,
      "loss": 0.3745,
      "step": 765
    },
    {
      "epoch": 0.03908827859282197,
      "grad_norm": 3.022794246673584,
      "learning_rate": 1.9739411476047857e-05,
      "loss": 0.3888,
      "step": 770
    },
    {
      "epoch": 0.039342098583684454,
      "grad_norm": 5.622093200683594,
      "learning_rate": 1.9737719342775437e-05,
      "loss": 0.379,
      "step": 775
    },
    {
      "epoch": 0.03959591857454693,
      "grad_norm": 1.1566879749298096,
      "learning_rate": 1.973602720950302e-05,
      "loss": 0.3739,
      "step": 780
    },
    {
      "epoch": 0.03984973856540941,
      "grad_norm": 2.6651649475097656,
      "learning_rate": 1.9734335076230608e-05,
      "loss": 0.3838,
      "step": 785
    },
    {
      "epoch": 0.040103558556271894,
      "grad_norm": 1.4303189516067505,
      "learning_rate": 1.9732642942958188e-05,
      "loss": 0.4171,
      "step": 790
    },
    {
      "epoch": 0.04035737854713437,
      "grad_norm": 0.97078937292099,
      "learning_rate": 1.973095080968577e-05,
      "loss": 0.3513,
      "step": 795
    },
    {
      "epoch": 0.04061119853799685,
      "grad_norm": 0.9570413827896118,
      "learning_rate": 1.9729258676413355e-05,
      "loss": 0.3702,
      "step": 800
    },
    {
      "epoch": 0.040865018528859334,
      "grad_norm": 1.7016774415969849,
      "learning_rate": 1.972756654314094e-05,
      "loss": 0.3608,
      "step": 805
    },
    {
      "epoch": 0.04111883851972181,
      "grad_norm": 1.1286097764968872,
      "learning_rate": 1.9725874409868522e-05,
      "loss": 0.3915,
      "step": 810
    },
    {
      "epoch": 0.0413726585105843,
      "grad_norm": 1.0915040969848633,
      "learning_rate": 1.9724182276596106e-05,
      "loss": 0.3501,
      "step": 815
    },
    {
      "epoch": 0.041626478501446774,
      "grad_norm": 3.019122362136841,
      "learning_rate": 1.972249014332369e-05,
      "loss": 0.3413,
      "step": 820
    },
    {
      "epoch": 0.04188029849230925,
      "grad_norm": 1.6340800523757935,
      "learning_rate": 1.9720798010051273e-05,
      "loss": 0.3902,
      "step": 825
    },
    {
      "epoch": 0.04213411848317174,
      "grad_norm": 4.086325645446777,
      "learning_rate": 1.9719105876778856e-05,
      "loss": 0.4549,
      "step": 830
    },
    {
      "epoch": 0.042387938474034215,
      "grad_norm": 1.0035637617111206,
      "learning_rate": 1.971741374350644e-05,
      "loss": 0.3406,
      "step": 835
    },
    {
      "epoch": 0.04264175846489669,
      "grad_norm": 0.8659387230873108,
      "learning_rate": 1.9715721610234024e-05,
      "loss": 0.3907,
      "step": 840
    },
    {
      "epoch": 0.04289557845575918,
      "grad_norm": 1.208949327468872,
      "learning_rate": 1.9714029476961607e-05,
      "loss": 0.3448,
      "step": 845
    },
    {
      "epoch": 0.043149398446621655,
      "grad_norm": 1.0771589279174805,
      "learning_rate": 1.971233734368919e-05,
      "loss": 0.3617,
      "step": 850
    },
    {
      "epoch": 0.04340321843748414,
      "grad_norm": 1.216821312904358,
      "learning_rate": 1.9710645210416774e-05,
      "loss": 0.3516,
      "step": 855
    },
    {
      "epoch": 0.04365703842834662,
      "grad_norm": 0.8121851086616516,
      "learning_rate": 1.9708953077144358e-05,
      "loss": 0.3514,
      "step": 860
    },
    {
      "epoch": 0.043910858419209095,
      "grad_norm": 0.9845564961433411,
      "learning_rate": 1.970726094387194e-05,
      "loss": 0.3626,
      "step": 865
    },
    {
      "epoch": 0.04416467841007158,
      "grad_norm": 276.0869140625,
      "learning_rate": 1.9705568810599525e-05,
      "loss": 0.3701,
      "step": 870
    },
    {
      "epoch": 0.04441849840093406,
      "grad_norm": 0.9632354974746704,
      "learning_rate": 1.970387667732711e-05,
      "loss": 0.3553,
      "step": 875
    },
    {
      "epoch": 0.044672318391796535,
      "grad_norm": 1.2075852155685425,
      "learning_rate": 1.9702184544054692e-05,
      "loss": 0.3355,
      "step": 880
    },
    {
      "epoch": 0.04492613838265902,
      "grad_norm": 1.1684023141860962,
      "learning_rate": 1.9700492410782275e-05,
      "loss": 0.3593,
      "step": 885
    },
    {
      "epoch": 0.0451799583735215,
      "grad_norm": 1.093794345855713,
      "learning_rate": 1.969880027750986e-05,
      "loss": 0.3217,
      "step": 890
    },
    {
      "epoch": 0.045433778364383975,
      "grad_norm": 0.8607542514801025,
      "learning_rate": 1.9697108144237443e-05,
      "loss": 0.3595,
      "step": 895
    },
    {
      "epoch": 0.04568759835524646,
      "grad_norm": 1.0586198568344116,
      "learning_rate": 1.9695416010965026e-05,
      "loss": 0.3402,
      "step": 900
    },
    {
      "epoch": 0.04594141834610894,
      "grad_norm": 1.3861478567123413,
      "learning_rate": 1.9693723877692606e-05,
      "loss": 0.3389,
      "step": 905
    },
    {
      "epoch": 0.04619523833697142,
      "grad_norm": 1.1892589330673218,
      "learning_rate": 1.9692031744420193e-05,
      "loss": 0.3371,
      "step": 910
    },
    {
      "epoch": 0.0464490583278339,
      "grad_norm": 1.0808658599853516,
      "learning_rate": 1.9690339611147777e-05,
      "loss": 0.3632,
      "step": 915
    },
    {
      "epoch": 0.04670287831869638,
      "grad_norm": 6.600335121154785,
      "learning_rate": 1.968864747787536e-05,
      "loss": 0.3464,
      "step": 920
    },
    {
      "epoch": 0.04695669830955886,
      "grad_norm": 0.9508718252182007,
      "learning_rate": 1.9686955344602944e-05,
      "loss": 0.3804,
      "step": 925
    },
    {
      "epoch": 0.04721051830042134,
      "grad_norm": 0.8976243734359741,
      "learning_rate": 1.9685263211330524e-05,
      "loss": 0.3379,
      "step": 930
    },
    {
      "epoch": 0.04746433829128382,
      "grad_norm": 0.816540539264679,
      "learning_rate": 1.968357107805811e-05,
      "loss": 0.3479,
      "step": 935
    },
    {
      "epoch": 0.0477181582821463,
      "grad_norm": 1.652158498764038,
      "learning_rate": 1.9681878944785694e-05,
      "loss": 0.3458,
      "step": 940
    },
    {
      "epoch": 0.04797197827300878,
      "grad_norm": 0.913906991481781,
      "learning_rate": 1.9680186811513275e-05,
      "loss": 0.3456,
      "step": 945
    },
    {
      "epoch": 0.048225798263871265,
      "grad_norm": 1.0308032035827637,
      "learning_rate": 1.967849467824086e-05,
      "loss": 0.3384,
      "step": 950
    },
    {
      "epoch": 0.04847961825473374,
      "grad_norm": 1.0393309593200684,
      "learning_rate": 1.9676802544968442e-05,
      "loss": 0.3399,
      "step": 955
    },
    {
      "epoch": 0.04873343824559622,
      "grad_norm": 0.9788406491279602,
      "learning_rate": 1.967511041169603e-05,
      "loss": 0.3387,
      "step": 960
    },
    {
      "epoch": 0.048987258236458706,
      "grad_norm": 1.0427477359771729,
      "learning_rate": 1.9673418278423612e-05,
      "loss": 0.3398,
      "step": 965
    },
    {
      "epoch": 0.04924107822732118,
      "grad_norm": 0.898420512676239,
      "learning_rate": 1.9671726145151192e-05,
      "loss": 0.3273,
      "step": 970
    },
    {
      "epoch": 0.04949489821818366,
      "grad_norm": 1.0824823379516602,
      "learning_rate": 1.967003401187878e-05,
      "loss": 0.3541,
      "step": 975
    },
    {
      "epoch": 0.049748718209046146,
      "grad_norm": 0.8971902132034302,
      "learning_rate": 1.966834187860636e-05,
      "loss": 0.304,
      "step": 980
    },
    {
      "epoch": 0.050002538199908624,
      "grad_norm": 2.5218119621276855,
      "learning_rate": 1.9666649745333943e-05,
      "loss": 0.3209,
      "step": 985
    },
    {
      "epoch": 0.05025635819077111,
      "grad_norm": 0.8553778529167175,
      "learning_rate": 1.9664957612061527e-05,
      "loss": 0.3208,
      "step": 990
    },
    {
      "epoch": 0.050510178181633586,
      "grad_norm": 1.212137222290039,
      "learning_rate": 1.966326547878911e-05,
      "loss": 0.3412,
      "step": 995
    },
    {
      "epoch": 0.050763998172496064,
      "grad_norm": 0.7518376708030701,
      "learning_rate": 1.9661573345516694e-05,
      "loss": 0.4053,
      "step": 1000
    },
    {
      "epoch": 0.05101781816335855,
      "grad_norm": 0.9260375499725342,
      "learning_rate": 1.9659881212244277e-05,
      "loss": 0.3254,
      "step": 1005
    },
    {
      "epoch": 0.051271638154221026,
      "grad_norm": 1.103661060333252,
      "learning_rate": 1.965818907897186e-05,
      "loss": 0.3211,
      "step": 1010
    },
    {
      "epoch": 0.051525458145083504,
      "grad_norm": 1.1954058408737183,
      "learning_rate": 1.9656496945699444e-05,
      "loss": 0.3486,
      "step": 1015
    },
    {
      "epoch": 0.05177927813594599,
      "grad_norm": 0.992374837398529,
      "learning_rate": 1.9654804812427028e-05,
      "loss": 0.3211,
      "step": 1020
    },
    {
      "epoch": 0.052033098126808466,
      "grad_norm": 4.1818413734436035,
      "learning_rate": 1.965311267915461e-05,
      "loss": 0.3095,
      "step": 1025
    },
    {
      "epoch": 0.05228691811767095,
      "grad_norm": 1.0134354829788208,
      "learning_rate": 1.9651420545882195e-05,
      "loss": 0.3281,
      "step": 1030
    },
    {
      "epoch": 0.05254073810853343,
      "grad_norm": 0.9592474102973938,
      "learning_rate": 1.964972841260978e-05,
      "loss": 0.3289,
      "step": 1035
    },
    {
      "epoch": 0.05279455809939591,
      "grad_norm": 0.8658552765846252,
      "learning_rate": 1.9648036279337362e-05,
      "loss": 0.3318,
      "step": 1040
    },
    {
      "epoch": 0.05304837809025839,
      "grad_norm": 0.9752548933029175,
      "learning_rate": 1.9646344146064946e-05,
      "loss": 0.3342,
      "step": 1045
    },
    {
      "epoch": 0.05330219808112087,
      "grad_norm": 1.1210874319076538,
      "learning_rate": 1.964465201279253e-05,
      "loss": 0.2928,
      "step": 1050
    },
    {
      "epoch": 0.05355601807198335,
      "grad_norm": 0.7496079206466675,
      "learning_rate": 1.9642959879520113e-05,
      "loss": 0.3433,
      "step": 1055
    },
    {
      "epoch": 0.05380983806284583,
      "grad_norm": 1.119456171989441,
      "learning_rate": 1.9641267746247696e-05,
      "loss": 0.3038,
      "step": 1060
    },
    {
      "epoch": 0.05406365805370831,
      "grad_norm": 0.9271594882011414,
      "learning_rate": 1.963957561297528e-05,
      "loss": 0.3138,
      "step": 1065
    },
    {
      "epoch": 0.05431747804457079,
      "grad_norm": 1.0885213613510132,
      "learning_rate": 1.9637883479702863e-05,
      "loss": 0.3176,
      "step": 1070
    },
    {
      "epoch": 0.05457129803543327,
      "grad_norm": 1.075253963470459,
      "learning_rate": 1.9636191346430447e-05,
      "loss": 0.2996,
      "step": 1075
    },
    {
      "epoch": 0.05482511802629575,
      "grad_norm": 0.9437046647071838,
      "learning_rate": 1.963449921315803e-05,
      "loss": 0.3203,
      "step": 1080
    },
    {
      "epoch": 0.055078938017158234,
      "grad_norm": 0.8787136077880859,
      "learning_rate": 1.9632807079885614e-05,
      "loss": 0.3017,
      "step": 1085
    },
    {
      "epoch": 0.05533275800802071,
      "grad_norm": 0.9308958649635315,
      "learning_rate": 1.9631114946613197e-05,
      "loss": 0.3023,
      "step": 1090
    },
    {
      "epoch": 0.05558657799888319,
      "grad_norm": 0.8047986626625061,
      "learning_rate": 1.962942281334078e-05,
      "loss": 0.3042,
      "step": 1095
    },
    {
      "epoch": 0.055840397989745674,
      "grad_norm": 0.6756587028503418,
      "learning_rate": 1.9627730680068364e-05,
      "loss": 0.2961,
      "step": 1100
    },
    {
      "epoch": 0.05609421798060815,
      "grad_norm": 1.1307016611099243,
      "learning_rate": 1.9626038546795948e-05,
      "loss": 0.3447,
      "step": 1105
    },
    {
      "epoch": 0.05634803797147063,
      "grad_norm": 0.9675906896591187,
      "learning_rate": 1.962434641352353e-05,
      "loss": 0.3325,
      "step": 1110
    },
    {
      "epoch": 0.056601857962333114,
      "grad_norm": 0.9537900686264038,
      "learning_rate": 1.9622654280251115e-05,
      "loss": 0.3241,
      "step": 1115
    },
    {
      "epoch": 0.05685567795319559,
      "grad_norm": 0.9516997337341309,
      "learning_rate": 1.96209621469787e-05,
      "loss": 0.3301,
      "step": 1120
    },
    {
      "epoch": 0.05710949794405808,
      "grad_norm": 1.157469391822815,
      "learning_rate": 1.9619270013706282e-05,
      "loss": 0.306,
      "step": 1125
    },
    {
      "epoch": 0.057363317934920555,
      "grad_norm": 0.7865618467330933,
      "learning_rate": 1.9617577880433866e-05,
      "loss": 0.3097,
      "step": 1130
    },
    {
      "epoch": 0.05761713792578303,
      "grad_norm": 0.8256521821022034,
      "learning_rate": 1.9615885747161446e-05,
      "loss": 0.3012,
      "step": 1135
    },
    {
      "epoch": 0.05787095791664552,
      "grad_norm": 0.7554883360862732,
      "learning_rate": 1.9614193613889033e-05,
      "loss": 0.3226,
      "step": 1140
    },
    {
      "epoch": 0.058124777907507995,
      "grad_norm": 0.8436192870140076,
      "learning_rate": 1.9612501480616616e-05,
      "loss": 0.3068,
      "step": 1145
    },
    {
      "epoch": 0.05837859789837047,
      "grad_norm": 1.5259509086608887,
      "learning_rate": 1.9610809347344197e-05,
      "loss": 0.3158,
      "step": 1150
    },
    {
      "epoch": 0.05863241788923296,
      "grad_norm": 1.0376449823379517,
      "learning_rate": 1.9609117214071783e-05,
      "loss": 0.2944,
      "step": 1155
    },
    {
      "epoch": 0.058886237880095435,
      "grad_norm": 0.9757311344146729,
      "learning_rate": 1.9607425080799364e-05,
      "loss": 0.3029,
      "step": 1160
    },
    {
      "epoch": 0.05914005787095792,
      "grad_norm": 1.0501054525375366,
      "learning_rate": 1.960573294752695e-05,
      "loss": 0.2927,
      "step": 1165
    },
    {
      "epoch": 0.0593938778618204,
      "grad_norm": 0.7940589189529419,
      "learning_rate": 1.9604040814254534e-05,
      "loss": 0.3092,
      "step": 1170
    },
    {
      "epoch": 0.059647697852682875,
      "grad_norm": 1.0338493585586548,
      "learning_rate": 1.9602348680982114e-05,
      "loss": 0.3086,
      "step": 1175
    },
    {
      "epoch": 0.05990151784354536,
      "grad_norm": 0.8323164582252502,
      "learning_rate": 1.96006565477097e-05,
      "loss": 0.2797,
      "step": 1180
    },
    {
      "epoch": 0.06015533783440784,
      "grad_norm": 1.0480012893676758,
      "learning_rate": 1.959896441443728e-05,
      "loss": 0.2908,
      "step": 1185
    },
    {
      "epoch": 0.060409157825270315,
      "grad_norm": 0.8771175742149353,
      "learning_rate": 1.9597272281164865e-05,
      "loss": 0.2945,
      "step": 1190
    },
    {
      "epoch": 0.0606629778161328,
      "grad_norm": 0.968286395072937,
      "learning_rate": 1.959558014789245e-05,
      "loss": 0.2989,
      "step": 1195
    },
    {
      "epoch": 0.06091679780699528,
      "grad_norm": 1.787429690361023,
      "learning_rate": 1.9593888014620032e-05,
      "loss": 0.3255,
      "step": 1200
    },
    {
      "epoch": 0.06117061779785776,
      "grad_norm": 1.2332099676132202,
      "learning_rate": 1.959219588134762e-05,
      "loss": 0.2937,
      "step": 1205
    },
    {
      "epoch": 0.06142443778872024,
      "grad_norm": 0.9430225491523743,
      "learning_rate": 1.95905037480752e-05,
      "loss": 0.309,
      "step": 1210
    },
    {
      "epoch": 0.06167825777958272,
      "grad_norm": 0.9462845921516418,
      "learning_rate": 1.9588811614802783e-05,
      "loss": 0.3156,
      "step": 1215
    },
    {
      "epoch": 0.0619320777704452,
      "grad_norm": 0.9460931420326233,
      "learning_rate": 1.9587119481530366e-05,
      "loss": 0.2851,
      "step": 1220
    },
    {
      "epoch": 0.06218589776130768,
      "grad_norm": 0.8023846745491028,
      "learning_rate": 1.958542734825795e-05,
      "loss": 0.304,
      "step": 1225
    },
    {
      "epoch": 0.06243971775217016,
      "grad_norm": 1.055830955505371,
      "learning_rate": 1.9583735214985533e-05,
      "loss": 0.3123,
      "step": 1230
    },
    {
      "epoch": 0.06269353774303264,
      "grad_norm": 0.7749910354614258,
      "learning_rate": 1.9582043081713117e-05,
      "loss": 0.3202,
      "step": 1235
    },
    {
      "epoch": 0.06294735773389512,
      "grad_norm": 1.0144844055175781,
      "learning_rate": 1.95803509484407e-05,
      "loss": 0.2781,
      "step": 1240
    },
    {
      "epoch": 0.0632011777247576,
      "grad_norm": 0.9046682715415955,
      "learning_rate": 1.9578658815168284e-05,
      "loss": 0.3233,
      "step": 1245
    },
    {
      "epoch": 0.06345499771562008,
      "grad_norm": 71.55082702636719,
      "learning_rate": 1.9576966681895867e-05,
      "loss": 0.2724,
      "step": 1250
    },
    {
      "epoch": 0.06370881770648257,
      "grad_norm": 0.8242619037628174,
      "learning_rate": 1.957527454862345e-05,
      "loss": 0.2722,
      "step": 1255
    },
    {
      "epoch": 0.06396263769734505,
      "grad_norm": 1.0677359104156494,
      "learning_rate": 1.9573582415351035e-05,
      "loss": 0.3172,
      "step": 1260
    },
    {
      "epoch": 0.06421645768820752,
      "grad_norm": 0.8488295078277588,
      "learning_rate": 1.9571890282078618e-05,
      "loss": 0.2966,
      "step": 1265
    },
    {
      "epoch": 0.06447027767907,
      "grad_norm": 0.703192949295044,
      "learning_rate": 1.95701981488062e-05,
      "loss": 0.3108,
      "step": 1270
    },
    {
      "epoch": 0.06472409766993248,
      "grad_norm": 0.8458622694015503,
      "learning_rate": 1.9568506015533785e-05,
      "loss": 0.2884,
      "step": 1275
    },
    {
      "epoch": 0.06497791766079497,
      "grad_norm": 0.9391268491744995,
      "learning_rate": 1.956681388226137e-05,
      "loss": 0.293,
      "step": 1280
    },
    {
      "epoch": 0.06523173765165745,
      "grad_norm": 1.0053985118865967,
      "learning_rate": 1.9565121748988952e-05,
      "loss": 0.3244,
      "step": 1285
    },
    {
      "epoch": 0.06548555764251993,
      "grad_norm": 0.8404228687286377,
      "learning_rate": 1.9563429615716536e-05,
      "loss": 0.2932,
      "step": 1290
    },
    {
      "epoch": 0.0657393776333824,
      "grad_norm": 1.3393089771270752,
      "learning_rate": 1.956173748244412e-05,
      "loss": 0.3064,
      "step": 1295
    },
    {
      "epoch": 0.06599319762424488,
      "grad_norm": 0.6775000691413879,
      "learning_rate": 1.9560045349171703e-05,
      "loss": 0.2931,
      "step": 1300
    },
    {
      "epoch": 0.06624701761510736,
      "grad_norm": 0.9884696006774902,
      "learning_rate": 1.9558353215899286e-05,
      "loss": 0.285,
      "step": 1305
    },
    {
      "epoch": 0.06650083760596985,
      "grad_norm": 1.0637091398239136,
      "learning_rate": 1.955666108262687e-05,
      "loss": 0.2468,
      "step": 1310
    },
    {
      "epoch": 0.06675465759683233,
      "grad_norm": 0.8244103193283081,
      "learning_rate": 1.9554968949354454e-05,
      "loss": 0.2582,
      "step": 1315
    },
    {
      "epoch": 0.0670084775876948,
      "grad_norm": 0.6568355560302734,
      "learning_rate": 1.9553276816082037e-05,
      "loss": 0.2774,
      "step": 1320
    },
    {
      "epoch": 0.06726229757855728,
      "grad_norm": 1.0442551374435425,
      "learning_rate": 1.955158468280962e-05,
      "loss": 0.277,
      "step": 1325
    },
    {
      "epoch": 0.06751611756941976,
      "grad_norm": 0.797654390335083,
      "learning_rate": 1.9549892549537204e-05,
      "loss": 0.2658,
      "step": 1330
    },
    {
      "epoch": 0.06776993756028225,
      "grad_norm": 0.8469611406326294,
      "learning_rate": 1.9548200416264788e-05,
      "loss": 0.319,
      "step": 1335
    },
    {
      "epoch": 0.06802375755114473,
      "grad_norm": 38.28089141845703,
      "learning_rate": 1.9546508282992368e-05,
      "loss": 0.268,
      "step": 1340
    },
    {
      "epoch": 0.06827757754200721,
      "grad_norm": 0.961740255355835,
      "learning_rate": 1.9544816149719955e-05,
      "loss": 0.3031,
      "step": 1345
    },
    {
      "epoch": 0.06853139753286969,
      "grad_norm": 0.9473285675048828,
      "learning_rate": 1.954312401644754e-05,
      "loss": 0.2707,
      "step": 1350
    },
    {
      "epoch": 0.06878521752373216,
      "grad_norm": 1.0564128160476685,
      "learning_rate": 1.9541431883175122e-05,
      "loss": 0.2747,
      "step": 1355
    },
    {
      "epoch": 0.06903903751459466,
      "grad_norm": 1.3247959613800049,
      "learning_rate": 1.9539739749902705e-05,
      "loss": 0.2801,
      "step": 1360
    },
    {
      "epoch": 0.06929285750545713,
      "grad_norm": 1.2612837553024292,
      "learning_rate": 1.9538047616630286e-05,
      "loss": 0.314,
      "step": 1365
    },
    {
      "epoch": 0.06954667749631961,
      "grad_norm": 0.8873483538627625,
      "learning_rate": 1.9536355483357873e-05,
      "loss": 0.2818,
      "step": 1370
    },
    {
      "epoch": 0.06980049748718209,
      "grad_norm": 1.1142306327819824,
      "learning_rate": 1.9534663350085453e-05,
      "loss": 0.2964,
      "step": 1375
    },
    {
      "epoch": 0.07005431747804457,
      "grad_norm": 0.7834559679031372,
      "learning_rate": 1.9532971216813036e-05,
      "loss": 0.3129,
      "step": 1380
    },
    {
      "epoch": 0.07030813746890704,
      "grad_norm": 0.6153595447540283,
      "learning_rate": 1.9531279083540623e-05,
      "loss": 0.2775,
      "step": 1385
    },
    {
      "epoch": 0.07056195745976954,
      "grad_norm": 0.9495353102684021,
      "learning_rate": 1.9529586950268203e-05,
      "loss": 0.2902,
      "step": 1390
    },
    {
      "epoch": 0.07081577745063201,
      "grad_norm": 1.0067836046218872,
      "learning_rate": 1.9527894816995787e-05,
      "loss": 0.2918,
      "step": 1395
    },
    {
      "epoch": 0.07106959744149449,
      "grad_norm": 1.068027377128601,
      "learning_rate": 1.952620268372337e-05,
      "loss": 0.2845,
      "step": 1400
    },
    {
      "epoch": 0.07132341743235697,
      "grad_norm": 1.1914137601852417,
      "learning_rate": 1.9524510550450954e-05,
      "loss": 0.2867,
      "step": 1405
    },
    {
      "epoch": 0.07157723742321945,
      "grad_norm": 0.9969173669815063,
      "learning_rate": 1.952281841717854e-05,
      "loss": 0.2888,
      "step": 1410
    },
    {
      "epoch": 0.07183105741408194,
      "grad_norm": 0.8332482576370239,
      "learning_rate": 1.952112628390612e-05,
      "loss": 0.2833,
      "step": 1415
    },
    {
      "epoch": 0.07208487740494442,
      "grad_norm": 1.3746370077133179,
      "learning_rate": 1.9519434150633705e-05,
      "loss": 0.2867,
      "step": 1420
    },
    {
      "epoch": 0.0723386973958069,
      "grad_norm": 0.6186951398849487,
      "learning_rate": 1.9517742017361288e-05,
      "loss": 0.2754,
      "step": 1425
    },
    {
      "epoch": 0.07259251738666937,
      "grad_norm": 0.668218195438385,
      "learning_rate": 1.951604988408887e-05,
      "loss": 0.2538,
      "step": 1430
    },
    {
      "epoch": 0.07284633737753185,
      "grad_norm": 0.900421679019928,
      "learning_rate": 1.9514357750816455e-05,
      "loss": 0.2657,
      "step": 1435
    },
    {
      "epoch": 0.07310015736839433,
      "grad_norm": 0.7691309452056885,
      "learning_rate": 1.951266561754404e-05,
      "loss": 0.3226,
      "step": 1440
    },
    {
      "epoch": 0.07335397735925682,
      "grad_norm": 0.9059398770332336,
      "learning_rate": 1.9510973484271622e-05,
      "loss": 0.3005,
      "step": 1445
    },
    {
      "epoch": 0.0736077973501193,
      "grad_norm": 1.2089287042617798,
      "learning_rate": 1.9509281350999206e-05,
      "loss": 0.2816,
      "step": 1450
    },
    {
      "epoch": 0.07386161734098178,
      "grad_norm": 0.8258754014968872,
      "learning_rate": 1.950758921772679e-05,
      "loss": 0.2874,
      "step": 1455
    },
    {
      "epoch": 0.07411543733184425,
      "grad_norm": 0.7340827584266663,
      "learning_rate": 1.9505897084454373e-05,
      "loss": 0.2884,
      "step": 1460
    },
    {
      "epoch": 0.07436925732270673,
      "grad_norm": 0.741913914680481,
      "learning_rate": 1.9504204951181956e-05,
      "loss": 0.2929,
      "step": 1465
    },
    {
      "epoch": 0.07462307731356922,
      "grad_norm": 0.7973793745040894,
      "learning_rate": 1.950251281790954e-05,
      "loss": 0.2963,
      "step": 1470
    },
    {
      "epoch": 0.0748768973044317,
      "grad_norm": 0.9016576409339905,
      "learning_rate": 1.9500820684637124e-05,
      "loss": 0.265,
      "step": 1475
    },
    {
      "epoch": 0.07513071729529418,
      "grad_norm": 1.0395702123641968,
      "learning_rate": 1.9499128551364707e-05,
      "loss": 0.3539,
      "step": 1480
    },
    {
      "epoch": 0.07538453728615666,
      "grad_norm": 0.7906758189201355,
      "learning_rate": 1.949743641809229e-05,
      "loss": 0.2769,
      "step": 1485
    },
    {
      "epoch": 0.07563835727701913,
      "grad_norm": 0.7581782341003418,
      "learning_rate": 1.9495744284819874e-05,
      "loss": 0.2911,
      "step": 1490
    },
    {
      "epoch": 0.07589217726788162,
      "grad_norm": 0.9939795732498169,
      "learning_rate": 1.9494052151547458e-05,
      "loss": 0.2759,
      "step": 1495
    },
    {
      "epoch": 0.0761459972587441,
      "grad_norm": 0.7973515391349792,
      "learning_rate": 1.949236001827504e-05,
      "loss": 0.2673,
      "step": 1500
    },
    {
      "epoch": 0.07639981724960658,
      "grad_norm": 0.793901264667511,
      "learning_rate": 1.9490667885002625e-05,
      "loss": 0.2643,
      "step": 1505
    },
    {
      "epoch": 0.07665363724046906,
      "grad_norm": 0.9730905294418335,
      "learning_rate": 1.948897575173021e-05,
      "loss": 0.2505,
      "step": 1510
    },
    {
      "epoch": 0.07690745723133154,
      "grad_norm": 0.7291544079780579,
      "learning_rate": 1.9487283618457792e-05,
      "loss": 0.2337,
      "step": 1515
    },
    {
      "epoch": 0.07716127722219401,
      "grad_norm": 0.836552083492279,
      "learning_rate": 1.9485591485185375e-05,
      "loss": 0.274,
      "step": 1520
    },
    {
      "epoch": 0.0774150972130565,
      "grad_norm": 0.7705031037330627,
      "learning_rate": 1.948389935191296e-05,
      "loss": 0.2706,
      "step": 1525
    },
    {
      "epoch": 0.07766891720391898,
      "grad_norm": 0.6896925568580627,
      "learning_rate": 1.9482207218640543e-05,
      "loss": 0.2807,
      "step": 1530
    },
    {
      "epoch": 0.07792273719478146,
      "grad_norm": 1.161415696144104,
      "learning_rate": 1.9480515085368126e-05,
      "loss": 0.2574,
      "step": 1535
    },
    {
      "epoch": 0.07817655718564394,
      "grad_norm": 1.0795843601226807,
      "learning_rate": 1.947882295209571e-05,
      "loss": 0.2574,
      "step": 1540
    },
    {
      "epoch": 0.07843037717650642,
      "grad_norm": 0.6460023522377014,
      "learning_rate": 1.947713081882329e-05,
      "loss": 0.2794,
      "step": 1545
    },
    {
      "epoch": 0.07868419716736891,
      "grad_norm": 0.9282433390617371,
      "learning_rate": 1.9475438685550877e-05,
      "loss": 0.2564,
      "step": 1550
    },
    {
      "epoch": 0.07893801715823139,
      "grad_norm": 0.9009420871734619,
      "learning_rate": 1.9473746552278457e-05,
      "loss": 0.2607,
      "step": 1555
    },
    {
      "epoch": 0.07919183714909386,
      "grad_norm": 0.8371971845626831,
      "learning_rate": 1.9472054419006044e-05,
      "loss": 0.2732,
      "step": 1560
    },
    {
      "epoch": 0.07944565713995634,
      "grad_norm": 0.9188066124916077,
      "learning_rate": 1.9470362285733627e-05,
      "loss": 0.2457,
      "step": 1565
    },
    {
      "epoch": 0.07969947713081882,
      "grad_norm": 0.7665749788284302,
      "learning_rate": 1.9468670152461208e-05,
      "loss": 0.2373,
      "step": 1570
    },
    {
      "epoch": 0.07995329712168131,
      "grad_norm": 1.9648408889770508,
      "learning_rate": 1.9466978019188794e-05,
      "loss": 0.2852,
      "step": 1575
    },
    {
      "epoch": 0.08020711711254379,
      "grad_norm": 0.8872603178024292,
      "learning_rate": 1.9465285885916375e-05,
      "loss": 0.3144,
      "step": 1580
    },
    {
      "epoch": 0.08046093710340627,
      "grad_norm": 1.1950451135635376,
      "learning_rate": 1.9463593752643958e-05,
      "loss": 0.2723,
      "step": 1585
    },
    {
      "epoch": 0.08071475709426874,
      "grad_norm": 0.8541845083236694,
      "learning_rate": 1.9461901619371545e-05,
      "loss": 0.2556,
      "step": 1590
    },
    {
      "epoch": 0.08096857708513122,
      "grad_norm": 0.8095240592956543,
      "learning_rate": 1.9460209486099125e-05,
      "loss": 0.2707,
      "step": 1595
    },
    {
      "epoch": 0.0812223970759937,
      "grad_norm": 0.8487701416015625,
      "learning_rate": 1.9458517352826712e-05,
      "loss": 0.2712,
      "step": 1600
    },
    {
      "epoch": 0.08147621706685619,
      "grad_norm": 1.3312042951583862,
      "learning_rate": 1.9456825219554292e-05,
      "loss": 0.2592,
      "step": 1605
    },
    {
      "epoch": 0.08173003705771867,
      "grad_norm": 0.7110800743103027,
      "learning_rate": 1.9455133086281876e-05,
      "loss": 0.2652,
      "step": 1610
    },
    {
      "epoch": 0.08198385704858115,
      "grad_norm": 0.933272659778595,
      "learning_rate": 1.9453440953009463e-05,
      "loss": 0.3137,
      "step": 1615
    },
    {
      "epoch": 0.08223767703944362,
      "grad_norm": 1.1123526096343994,
      "learning_rate": 1.9451748819737043e-05,
      "loss": 0.2681,
      "step": 1620
    },
    {
      "epoch": 0.0824914970303061,
      "grad_norm": 1.181645393371582,
      "learning_rate": 1.9450056686464627e-05,
      "loss": 0.2715,
      "step": 1625
    },
    {
      "epoch": 0.0827453170211686,
      "grad_norm": 1.169384241104126,
      "learning_rate": 1.944836455319221e-05,
      "loss": 0.2534,
      "step": 1630
    },
    {
      "epoch": 0.08299913701203107,
      "grad_norm": 0.8438488245010376,
      "learning_rate": 1.9446672419919794e-05,
      "loss": 0.2797,
      "step": 1635
    },
    {
      "epoch": 0.08325295700289355,
      "grad_norm": 0.6939857602119446,
      "learning_rate": 1.9444980286647377e-05,
      "loss": 0.2649,
      "step": 1640
    },
    {
      "epoch": 0.08350677699375603,
      "grad_norm": 0.803497314453125,
      "learning_rate": 1.944328815337496e-05,
      "loss": 0.2699,
      "step": 1645
    },
    {
      "epoch": 0.0837605969846185,
      "grad_norm": 0.6957250237464905,
      "learning_rate": 1.9441596020102544e-05,
      "loss": 0.2411,
      "step": 1650
    },
    {
      "epoch": 0.08401441697548098,
      "grad_norm": 0.8274784684181213,
      "learning_rate": 1.9439903886830128e-05,
      "loss": 0.2345,
      "step": 1655
    },
    {
      "epoch": 0.08426823696634347,
      "grad_norm": 0.9105510711669922,
      "learning_rate": 1.943821175355771e-05,
      "loss": 0.2779,
      "step": 1660
    },
    {
      "epoch": 0.08452205695720595,
      "grad_norm": 0.8826274275779724,
      "learning_rate": 1.9436519620285295e-05,
      "loss": 0.2501,
      "step": 1665
    },
    {
      "epoch": 0.08477587694806843,
      "grad_norm": 1.227468729019165,
      "learning_rate": 1.943482748701288e-05,
      "loss": 0.2628,
      "step": 1670
    },
    {
      "epoch": 0.08502969693893091,
      "grad_norm": 0.8807783722877502,
      "learning_rate": 1.9433135353740462e-05,
      "loss": 0.319,
      "step": 1675
    },
    {
      "epoch": 0.08528351692979338,
      "grad_norm": 0.829677164554596,
      "learning_rate": 1.9431443220468045e-05,
      "loss": 0.2426,
      "step": 1680
    },
    {
      "epoch": 0.08553733692065588,
      "grad_norm": 0.7359742522239685,
      "learning_rate": 1.942975108719563e-05,
      "loss": 0.2444,
      "step": 1685
    },
    {
      "epoch": 0.08579115691151835,
      "grad_norm": 0.8583529591560364,
      "learning_rate": 1.9428058953923213e-05,
      "loss": 0.2667,
      "step": 1690
    },
    {
      "epoch": 0.08604497690238083,
      "grad_norm": 0.8555963039398193,
      "learning_rate": 1.9426366820650796e-05,
      "loss": 0.2694,
      "step": 1695
    },
    {
      "epoch": 0.08629879689324331,
      "grad_norm": 0.7486214637756348,
      "learning_rate": 1.942467468737838e-05,
      "loss": 0.2464,
      "step": 1700
    },
    {
      "epoch": 0.08655261688410579,
      "grad_norm": 0.9317395687103271,
      "learning_rate": 1.9422982554105963e-05,
      "loss": 0.2421,
      "step": 1705
    },
    {
      "epoch": 0.08680643687496828,
      "grad_norm": 1.1423553228378296,
      "learning_rate": 1.9421290420833547e-05,
      "loss": 0.2773,
      "step": 1710
    },
    {
      "epoch": 0.08706025686583076,
      "grad_norm": 0.7504492402076721,
      "learning_rate": 1.941959828756113e-05,
      "loss": 0.2531,
      "step": 1715
    },
    {
      "epoch": 0.08731407685669323,
      "grad_norm": 0.7574618458747864,
      "learning_rate": 1.9417906154288714e-05,
      "loss": 0.2537,
      "step": 1720
    },
    {
      "epoch": 0.08756789684755571,
      "grad_norm": 0.7793534994125366,
      "learning_rate": 1.9416214021016297e-05,
      "loss": 0.2856,
      "step": 1725
    },
    {
      "epoch": 0.08782171683841819,
      "grad_norm": 0.7327775955200195,
      "learning_rate": 1.941452188774388e-05,
      "loss": 0.2499,
      "step": 1730
    },
    {
      "epoch": 0.08807553682928067,
      "grad_norm": 0.7954163551330566,
      "learning_rate": 1.941282975447146e-05,
      "loss": 0.2599,
      "step": 1735
    },
    {
      "epoch": 0.08832935682014316,
      "grad_norm": 0.7551532983779907,
      "learning_rate": 1.9411137621199048e-05,
      "loss": 0.2548,
      "step": 1740
    },
    {
      "epoch": 0.08858317681100564,
      "grad_norm": 0.8406553864479065,
      "learning_rate": 1.940944548792663e-05,
      "loss": 0.2291,
      "step": 1745
    },
    {
      "epoch": 0.08883699680186811,
      "grad_norm": 0.6450394988059998,
      "learning_rate": 1.9407753354654215e-05,
      "loss": 0.2654,
      "step": 1750
    },
    {
      "epoch": 0.08909081679273059,
      "grad_norm": 0.6348497271537781,
      "learning_rate": 1.94060612213818e-05,
      "loss": 0.2302,
      "step": 1755
    },
    {
      "epoch": 0.08934463678359307,
      "grad_norm": 0.589336097240448,
      "learning_rate": 1.940436908810938e-05,
      "loss": 0.242,
      "step": 1760
    },
    {
      "epoch": 0.08959845677445556,
      "grad_norm": 1.3192259073257446,
      "learning_rate": 1.9402676954836966e-05,
      "loss": 0.2606,
      "step": 1765
    },
    {
      "epoch": 0.08985227676531804,
      "grad_norm": 1.3367799520492554,
      "learning_rate": 1.940098482156455e-05,
      "loss": 0.2755,
      "step": 1770
    },
    {
      "epoch": 0.09010609675618052,
      "grad_norm": 0.8470580577850342,
      "learning_rate": 1.939929268829213e-05,
      "loss": 0.2596,
      "step": 1775
    },
    {
      "epoch": 0.090359916747043,
      "grad_norm": 0.7992636561393738,
      "learning_rate": 1.9397600555019716e-05,
      "loss": 0.259,
      "step": 1780
    },
    {
      "epoch": 0.09061373673790547,
      "grad_norm": 0.6407599449157715,
      "learning_rate": 1.9395908421747297e-05,
      "loss": 0.2247,
      "step": 1785
    },
    {
      "epoch": 0.09086755672876795,
      "grad_norm": 0.9677866101264954,
      "learning_rate": 1.939421628847488e-05,
      "loss": 0.2329,
      "step": 1790
    },
    {
      "epoch": 0.09112137671963044,
      "grad_norm": 0.7822664976119995,
      "learning_rate": 1.9392524155202467e-05,
      "loss": 0.2402,
      "step": 1795
    },
    {
      "epoch": 0.09137519671049292,
      "grad_norm": 0.7724019289016724,
      "learning_rate": 1.9390832021930047e-05,
      "loss": 0.2504,
      "step": 1800
    },
    {
      "epoch": 0.0916290167013554,
      "grad_norm": 0.7643800973892212,
      "learning_rate": 1.9389139888657634e-05,
      "loss": 0.2409,
      "step": 1805
    },
    {
      "epoch": 0.09188283669221788,
      "grad_norm": 0.84905606508255,
      "learning_rate": 1.9387447755385214e-05,
      "loss": 0.2391,
      "step": 1810
    },
    {
      "epoch": 0.09213665668308035,
      "grad_norm": 3.136202812194824,
      "learning_rate": 1.9385755622112798e-05,
      "loss": 0.3056,
      "step": 1815
    },
    {
      "epoch": 0.09239047667394285,
      "grad_norm": 1.0175585746765137,
      "learning_rate": 1.9384063488840385e-05,
      "loss": 0.2591,
      "step": 1820
    },
    {
      "epoch": 0.09264429666480532,
      "grad_norm": 1.1569483280181885,
      "learning_rate": 1.9382371355567965e-05,
      "loss": 0.2738,
      "step": 1825
    },
    {
      "epoch": 0.0928981166556678,
      "grad_norm": 0.648151159286499,
      "learning_rate": 1.938067922229555e-05,
      "loss": 0.2288,
      "step": 1830
    },
    {
      "epoch": 0.09315193664653028,
      "grad_norm": 0.5930745005607605,
      "learning_rate": 1.9378987089023132e-05,
      "loss": 0.2324,
      "step": 1835
    },
    {
      "epoch": 0.09340575663739276,
      "grad_norm": 0.7623791098594666,
      "learning_rate": 1.9377294955750716e-05,
      "loss": 0.2287,
      "step": 1840
    },
    {
      "epoch": 0.09365957662825525,
      "grad_norm": 1.552193284034729,
      "learning_rate": 1.9375602822478302e-05,
      "loss": 0.2388,
      "step": 1845
    },
    {
      "epoch": 0.09391339661911773,
      "grad_norm": 0.6569783687591553,
      "learning_rate": 1.9373910689205883e-05,
      "loss": 0.2415,
      "step": 1850
    },
    {
      "epoch": 0.0941672166099802,
      "grad_norm": 0.8682032823562622,
      "learning_rate": 1.9372218555933466e-05,
      "loss": 0.2376,
      "step": 1855
    },
    {
      "epoch": 0.09442103660084268,
      "grad_norm": 0.7503842115402222,
      "learning_rate": 1.937052642266105e-05,
      "loss": 0.2479,
      "step": 1860
    },
    {
      "epoch": 0.09467485659170516,
      "grad_norm": 1.0479599237442017,
      "learning_rate": 1.9368834289388633e-05,
      "loss": 0.2598,
      "step": 1865
    },
    {
      "epoch": 0.09492867658256764,
      "grad_norm": 3.8259048461914062,
      "learning_rate": 1.9367142156116217e-05,
      "loss": 0.2883,
      "step": 1870
    },
    {
      "epoch": 0.09518249657343013,
      "grad_norm": 0.7968519330024719,
      "learning_rate": 1.93654500228438e-05,
      "loss": 0.2404,
      "step": 1875
    },
    {
      "epoch": 0.0954363165642926,
      "grad_norm": 0.9822812080383301,
      "learning_rate": 1.9363757889571384e-05,
      "loss": 0.2585,
      "step": 1880
    },
    {
      "epoch": 0.09569013655515508,
      "grad_norm": 1.0278716087341309,
      "learning_rate": 1.9362065756298967e-05,
      "loss": 0.2263,
      "step": 1885
    },
    {
      "epoch": 0.09594395654601756,
      "grad_norm": 0.7152266502380371,
      "learning_rate": 1.936037362302655e-05,
      "loss": 0.2329,
      "step": 1890
    },
    {
      "epoch": 0.09619777653688004,
      "grad_norm": 0.9978631138801575,
      "learning_rate": 1.9358681489754135e-05,
      "loss": 0.2461,
      "step": 1895
    },
    {
      "epoch": 0.09645159652774253,
      "grad_norm": 0.8091859221458435,
      "learning_rate": 1.9356989356481718e-05,
      "loss": 0.248,
      "step": 1900
    },
    {
      "epoch": 0.09670541651860501,
      "grad_norm": 0.714453935623169,
      "learning_rate": 1.93552972232093e-05,
      "loss": 0.2521,
      "step": 1905
    },
    {
      "epoch": 0.09695923650946749,
      "grad_norm": 0.9081348180770874,
      "learning_rate": 1.9353605089936885e-05,
      "loss": 0.2538,
      "step": 1910
    },
    {
      "epoch": 0.09721305650032996,
      "grad_norm": 1.0938876867294312,
      "learning_rate": 1.935191295666447e-05,
      "loss": 0.2338,
      "step": 1915
    },
    {
      "epoch": 0.09746687649119244,
      "grad_norm": 0.6565501093864441,
      "learning_rate": 1.9350220823392052e-05,
      "loss": 0.2505,
      "step": 1920
    },
    {
      "epoch": 0.09772069648205493,
      "grad_norm": 0.615381121635437,
      "learning_rate": 1.9348528690119636e-05,
      "loss": 0.228,
      "step": 1925
    },
    {
      "epoch": 0.09797451647291741,
      "grad_norm": 1.128936529159546,
      "learning_rate": 1.934683655684722e-05,
      "loss": 0.2206,
      "step": 1930
    },
    {
      "epoch": 0.09822833646377989,
      "grad_norm": 0.9524216651916504,
      "learning_rate": 1.9345144423574803e-05,
      "loss": 0.2608,
      "step": 1935
    },
    {
      "epoch": 0.09848215645464237,
      "grad_norm": 0.7258911728858948,
      "learning_rate": 1.9343452290302383e-05,
      "loss": 0.247,
      "step": 1940
    },
    {
      "epoch": 0.09873597644550484,
      "grad_norm": 0.7854008674621582,
      "learning_rate": 1.934176015702997e-05,
      "loss": 0.2611,
      "step": 1945
    },
    {
      "epoch": 0.09898979643636732,
      "grad_norm": 0.9118169546127319,
      "learning_rate": 1.9340068023757554e-05,
      "loss": 0.2428,
      "step": 1950
    },
    {
      "epoch": 0.09924361642722981,
      "grad_norm": 0.9793441891670227,
      "learning_rate": 1.9338375890485137e-05,
      "loss": 0.2296,
      "step": 1955
    },
    {
      "epoch": 0.09949743641809229,
      "grad_norm": 0.601921558380127,
      "learning_rate": 1.933668375721272e-05,
      "loss": 0.2492,
      "step": 1960
    },
    {
      "epoch": 0.09975125640895477,
      "grad_norm": 0.6406486630439758,
      "learning_rate": 1.93349916239403e-05,
      "loss": 0.2456,
      "step": 1965
    },
    {
      "epoch": 0.10000507639981725,
      "grad_norm": 0.935122013092041,
      "learning_rate": 1.9333299490667888e-05,
      "loss": 0.2373,
      "step": 1970
    },
    {
      "epoch": 0.10025889639067972,
      "grad_norm": 0.6741816997528076,
      "learning_rate": 1.933160735739547e-05,
      "loss": 0.2215,
      "step": 1975
    },
    {
      "epoch": 0.10051271638154222,
      "grad_norm": 1.050551414489746,
      "learning_rate": 1.932991522412305e-05,
      "loss": 0.2436,
      "step": 1980
    },
    {
      "epoch": 0.1007665363724047,
      "grad_norm": 0.9753983020782471,
      "learning_rate": 1.932822309085064e-05,
      "loss": 0.2506,
      "step": 1985
    },
    {
      "epoch": 0.10102035636326717,
      "grad_norm": 0.5514481067657471,
      "learning_rate": 1.932653095757822e-05,
      "loss": 0.2287,
      "step": 1990
    },
    {
      "epoch": 0.10127417635412965,
      "grad_norm": 0.6138185262680054,
      "learning_rate": 1.9324838824305805e-05,
      "loss": 0.2546,
      "step": 1995
    },
    {
      "epoch": 0.10152799634499213,
      "grad_norm": 0.7834172248840332,
      "learning_rate": 1.932314669103339e-05,
      "loss": 0.2398,
      "step": 2000
    },
    {
      "epoch": 0.1017818163358546,
      "grad_norm": 0.7345184683799744,
      "learning_rate": 1.932145455776097e-05,
      "loss": 0.2264,
      "step": 2005
    },
    {
      "epoch": 0.1020356363267171,
      "grad_norm": 0.7603473663330078,
      "learning_rate": 1.9319762424488556e-05,
      "loss": 0.2579,
      "step": 2010
    },
    {
      "epoch": 0.10228945631757957,
      "grad_norm": 0.5276412963867188,
      "learning_rate": 1.9318070291216136e-05,
      "loss": 0.2575,
      "step": 2015
    },
    {
      "epoch": 0.10254327630844205,
      "grad_norm": 0.5985664129257202,
      "learning_rate": 1.931637815794372e-05,
      "loss": 0.2357,
      "step": 2020
    },
    {
      "epoch": 0.10279709629930453,
      "grad_norm": 1.0004132986068726,
      "learning_rate": 1.9314686024671307e-05,
      "loss": 0.2415,
      "step": 2025
    },
    {
      "epoch": 0.10305091629016701,
      "grad_norm": 0.9058099389076233,
      "learning_rate": 1.9312993891398887e-05,
      "loss": 0.2489,
      "step": 2030
    },
    {
      "epoch": 0.1033047362810295,
      "grad_norm": 0.6580535769462585,
      "learning_rate": 1.931130175812647e-05,
      "loss": 0.2282,
      "step": 2035
    },
    {
      "epoch": 0.10355855627189198,
      "grad_norm": 0.8212989568710327,
      "learning_rate": 1.9309609624854054e-05,
      "loss": 0.2486,
      "step": 2040
    },
    {
      "epoch": 0.10381237626275445,
      "grad_norm": 0.6100918650627136,
      "learning_rate": 1.9307917491581637e-05,
      "loss": 0.2817,
      "step": 2045
    },
    {
      "epoch": 0.10406619625361693,
      "grad_norm": 0.9934706687927246,
      "learning_rate": 1.9306225358309224e-05,
      "loss": 0.2319,
      "step": 2050
    },
    {
      "epoch": 0.10432001624447941,
      "grad_norm": 0.7338758111000061,
      "learning_rate": 1.9304533225036805e-05,
      "loss": 0.2284,
      "step": 2055
    },
    {
      "epoch": 0.1045738362353419,
      "grad_norm": 0.7655614614486694,
      "learning_rate": 1.9302841091764388e-05,
      "loss": 0.2106,
      "step": 2060
    },
    {
      "epoch": 0.10482765622620438,
      "grad_norm": 0.8056983351707458,
      "learning_rate": 1.930114895849197e-05,
      "loss": 0.2301,
      "step": 2065
    },
    {
      "epoch": 0.10508147621706686,
      "grad_norm": 0.6935755610466003,
      "learning_rate": 1.9299456825219555e-05,
      "loss": 0.1977,
      "step": 2070
    },
    {
      "epoch": 0.10533529620792934,
      "grad_norm": 0.7659327387809753,
      "learning_rate": 1.929776469194714e-05,
      "loss": 0.2319,
      "step": 2075
    },
    {
      "epoch": 0.10558911619879181,
      "grad_norm": 0.8755415081977844,
      "learning_rate": 1.9296072558674722e-05,
      "loss": 0.2531,
      "step": 2080
    },
    {
      "epoch": 0.10584293618965429,
      "grad_norm": 0.7641196846961975,
      "learning_rate": 1.9294380425402306e-05,
      "loss": 0.2262,
      "step": 2085
    },
    {
      "epoch": 0.10609675618051678,
      "grad_norm": 0.5032942295074463,
      "learning_rate": 1.929268829212989e-05,
      "loss": 0.2371,
      "step": 2090
    },
    {
      "epoch": 0.10635057617137926,
      "grad_norm": 0.8890644311904907,
      "learning_rate": 1.9290996158857473e-05,
      "loss": 0.2262,
      "step": 2095
    },
    {
      "epoch": 0.10660439616224174,
      "grad_norm": 0.6854649186134338,
      "learning_rate": 1.9289304025585056e-05,
      "loss": 0.2407,
      "step": 2100
    },
    {
      "epoch": 0.10685821615310422,
      "grad_norm": 0.7222452163696289,
      "learning_rate": 1.928761189231264e-05,
      "loss": 0.2246,
      "step": 2105
    },
    {
      "epoch": 0.1071120361439667,
      "grad_norm": 0.5661618709564209,
      "learning_rate": 1.9285919759040224e-05,
      "loss": 0.236,
      "step": 2110
    },
    {
      "epoch": 0.10736585613482919,
      "grad_norm": 0.6868450045585632,
      "learning_rate": 1.9284227625767807e-05,
      "loss": 0.2206,
      "step": 2115
    },
    {
      "epoch": 0.10761967612569166,
      "grad_norm": 0.6375486254692078,
      "learning_rate": 1.928253549249539e-05,
      "loss": 0.2075,
      "step": 2120
    },
    {
      "epoch": 0.10787349611655414,
      "grad_norm": 0.7467653751373291,
      "learning_rate": 1.9280843359222974e-05,
      "loss": 0.2168,
      "step": 2125
    },
    {
      "epoch": 0.10812731610741662,
      "grad_norm": 1.1524518728256226,
      "learning_rate": 1.9279151225950558e-05,
      "loss": 0.2622,
      "step": 2130
    },
    {
      "epoch": 0.1083811360982791,
      "grad_norm": 0.8847719430923462,
      "learning_rate": 1.927745909267814e-05,
      "loss": 0.2297,
      "step": 2135
    },
    {
      "epoch": 0.10863495608914157,
      "grad_norm": 1.002732515335083,
      "learning_rate": 1.9275766959405725e-05,
      "loss": 0.2322,
      "step": 2140
    },
    {
      "epoch": 0.10888877608000407,
      "grad_norm": 1.2165062427520752,
      "learning_rate": 1.927407482613331e-05,
      "loss": 0.2213,
      "step": 2145
    },
    {
      "epoch": 0.10914259607086654,
      "grad_norm": 0.7660351395606995,
      "learning_rate": 1.9272382692860892e-05,
      "loss": 0.2369,
      "step": 2150
    },
    {
      "epoch": 0.10939641606172902,
      "grad_norm": 0.6399320960044861,
      "learning_rate": 1.9270690559588475e-05,
      "loss": 0.2226,
      "step": 2155
    },
    {
      "epoch": 0.1096502360525915,
      "grad_norm": 0.929460883140564,
      "learning_rate": 1.926899842631606e-05,
      "loss": 0.2345,
      "step": 2160
    },
    {
      "epoch": 0.10990405604345398,
      "grad_norm": 0.7910303473472595,
      "learning_rate": 1.9267306293043643e-05,
      "loss": 0.1851,
      "step": 2165
    },
    {
      "epoch": 0.11015787603431647,
      "grad_norm": 0.6536149978637695,
      "learning_rate": 1.9265614159771223e-05,
      "loss": 0.2259,
      "step": 2170
    },
    {
      "epoch": 0.11041169602517895,
      "grad_norm": 0.6428697109222412,
      "learning_rate": 1.926392202649881e-05,
      "loss": 0.2169,
      "step": 2175
    },
    {
      "epoch": 0.11066551601604142,
      "grad_norm": 0.6982537508010864,
      "learning_rate": 1.9262229893226393e-05,
      "loss": 0.2084,
      "step": 2180
    },
    {
      "epoch": 0.1109193360069039,
      "grad_norm": 0.7507615089416504,
      "learning_rate": 1.9260537759953973e-05,
      "loss": 0.2433,
      "step": 2185
    },
    {
      "epoch": 0.11117315599776638,
      "grad_norm": 0.8022400140762329,
      "learning_rate": 1.925884562668156e-05,
      "loss": 0.2215,
      "step": 2190
    },
    {
      "epoch": 0.11142697598862887,
      "grad_norm": 0.5533608794212341,
      "learning_rate": 1.925715349340914e-05,
      "loss": 0.2013,
      "step": 2195
    },
    {
      "epoch": 0.11168079597949135,
      "grad_norm": 0.6407077312469482,
      "learning_rate": 1.9255461360136727e-05,
      "loss": 0.2131,
      "step": 2200
    },
    {
      "epoch": 0.11193461597035383,
      "grad_norm": 1.1785430908203125,
      "learning_rate": 1.925376922686431e-05,
      "loss": 0.2623,
      "step": 2205
    },
    {
      "epoch": 0.1121884359612163,
      "grad_norm": 0.6303373575210571,
      "learning_rate": 1.925207709359189e-05,
      "loss": 0.2143,
      "step": 2210
    },
    {
      "epoch": 0.11244225595207878,
      "grad_norm": 0.8575406074523926,
      "learning_rate": 1.9250384960319478e-05,
      "loss": 0.2237,
      "step": 2215
    },
    {
      "epoch": 0.11269607594294126,
      "grad_norm": 0.9760498404502869,
      "learning_rate": 1.9248692827047058e-05,
      "loss": 0.214,
      "step": 2220
    },
    {
      "epoch": 0.11294989593380375,
      "grad_norm": 1.455056071281433,
      "learning_rate": 1.924700069377464e-05,
      "loss": 0.2518,
      "step": 2225
    },
    {
      "epoch": 0.11320371592466623,
      "grad_norm": 0.8172942996025085,
      "learning_rate": 1.924530856050223e-05,
      "loss": 0.2165,
      "step": 2230
    },
    {
      "epoch": 0.1134575359155287,
      "grad_norm": 0.9945189952850342,
      "learning_rate": 1.924361642722981e-05,
      "loss": 0.2459,
      "step": 2235
    },
    {
      "epoch": 0.11371135590639118,
      "grad_norm": 1.0252959728240967,
      "learning_rate": 1.9241924293957396e-05,
      "loss": 0.2323,
      "step": 2240
    },
    {
      "epoch": 0.11396517589725366,
      "grad_norm": 0.5688400268554688,
      "learning_rate": 1.9240232160684976e-05,
      "loss": 0.1913,
      "step": 2245
    },
    {
      "epoch": 0.11421899588811615,
      "grad_norm": 0.6286960244178772,
      "learning_rate": 1.923854002741256e-05,
      "loss": 0.2405,
      "step": 2250
    },
    {
      "epoch": 0.11447281587897863,
      "grad_norm": 1.2339733839035034,
      "learning_rate": 1.9236847894140146e-05,
      "loss": 0.2043,
      "step": 2255
    },
    {
      "epoch": 0.11472663586984111,
      "grad_norm": 0.5691444873809814,
      "learning_rate": 1.9235155760867726e-05,
      "loss": 0.221,
      "step": 2260
    },
    {
      "epoch": 0.11498045586070359,
      "grad_norm": 1.1626230478286743,
      "learning_rate": 1.923346362759531e-05,
      "loss": 0.225,
      "step": 2265
    },
    {
      "epoch": 0.11523427585156606,
      "grad_norm": 0.7354618906974792,
      "learning_rate": 1.9231771494322894e-05,
      "loss": 0.2281,
      "step": 2270
    },
    {
      "epoch": 0.11548809584242856,
      "grad_norm": 0.5635653734207153,
      "learning_rate": 1.9230079361050477e-05,
      "loss": 0.2105,
      "step": 2275
    },
    {
      "epoch": 0.11574191583329103,
      "grad_norm": 1.019686222076416,
      "learning_rate": 1.922838722777806e-05,
      "loss": 0.2232,
      "step": 2280
    },
    {
      "epoch": 0.11599573582415351,
      "grad_norm": 0.5987036228179932,
      "learning_rate": 1.9226695094505644e-05,
      "loss": 0.2232,
      "step": 2285
    },
    {
      "epoch": 0.11624955581501599,
      "grad_norm": 0.8209441304206848,
      "learning_rate": 1.9225002961233228e-05,
      "loss": 0.2016,
      "step": 2290
    },
    {
      "epoch": 0.11650337580587847,
      "grad_norm": 0.576745331287384,
      "learning_rate": 1.922331082796081e-05,
      "loss": 0.239,
      "step": 2295
    },
    {
      "epoch": 0.11675719579674095,
      "grad_norm": 1.0648269653320312,
      "learning_rate": 1.9221618694688395e-05,
      "loss": 0.1987,
      "step": 2300
    },
    {
      "epoch": 0.11701101578760344,
      "grad_norm": 0.5774346590042114,
      "learning_rate": 1.921992656141598e-05,
      "loss": 0.225,
      "step": 2305
    },
    {
      "epoch": 0.11726483577846591,
      "grad_norm": 0.6183851957321167,
      "learning_rate": 1.9218234428143562e-05,
      "loss": 0.2028,
      "step": 2310
    },
    {
      "epoch": 0.11751865576932839,
      "grad_norm": 0.8308761715888977,
      "learning_rate": 1.9216542294871145e-05,
      "loss": 0.2405,
      "step": 2315
    },
    {
      "epoch": 0.11777247576019087,
      "grad_norm": 0.8884351253509521,
      "learning_rate": 1.921485016159873e-05,
      "loss": 0.2338,
      "step": 2320
    },
    {
      "epoch": 0.11802629575105335,
      "grad_norm": 0.8659582138061523,
      "learning_rate": 1.9213158028326313e-05,
      "loss": 0.2082,
      "step": 2325
    },
    {
      "epoch": 0.11828011574191584,
      "grad_norm": 0.7193389534950256,
      "learning_rate": 1.9211465895053896e-05,
      "loss": 0.211,
      "step": 2330
    },
    {
      "epoch": 0.11853393573277832,
      "grad_norm": 0.6831737756729126,
      "learning_rate": 1.920977376178148e-05,
      "loss": 0.2214,
      "step": 2335
    },
    {
      "epoch": 0.1187877557236408,
      "grad_norm": 0.7798734903335571,
      "learning_rate": 1.9208081628509063e-05,
      "loss": 0.2062,
      "step": 2340
    },
    {
      "epoch": 0.11904157571450327,
      "grad_norm": 0.7005655169487,
      "learning_rate": 1.9206389495236647e-05,
      "loss": 0.2081,
      "step": 2345
    },
    {
      "epoch": 0.11929539570536575,
      "grad_norm": 0.694050669670105,
      "learning_rate": 1.920469736196423e-05,
      "loss": 0.2348,
      "step": 2350
    },
    {
      "epoch": 0.11954921569622823,
      "grad_norm": 0.5242777466773987,
      "learning_rate": 1.9203005228691814e-05,
      "loss": 0.2103,
      "step": 2355
    },
    {
      "epoch": 0.11980303568709072,
      "grad_norm": 0.9790678024291992,
      "learning_rate": 1.9201313095419397e-05,
      "loss": 0.2186,
      "step": 2360
    },
    {
      "epoch": 0.1200568556779532,
      "grad_norm": 0.6226420998573303,
      "learning_rate": 1.919962096214698e-05,
      "loss": 0.2132,
      "step": 2365
    },
    {
      "epoch": 0.12031067566881568,
      "grad_norm": 0.6048428416252136,
      "learning_rate": 1.9197928828874564e-05,
      "loss": 0.2111,
      "step": 2370
    },
    {
      "epoch": 0.12056449565967815,
      "grad_norm": 0.4867452383041382,
      "learning_rate": 1.9196236695602145e-05,
      "loss": 0.1954,
      "step": 2375
    },
    {
      "epoch": 0.12081831565054063,
      "grad_norm": 0.8069561123847961,
      "learning_rate": 1.919454456232973e-05,
      "loss": 0.2194,
      "step": 2380
    },
    {
      "epoch": 0.12107213564140312,
      "grad_norm": 0.9302831888198853,
      "learning_rate": 1.9192852429057315e-05,
      "loss": 0.2132,
      "step": 2385
    },
    {
      "epoch": 0.1213259556322656,
      "grad_norm": 0.6558440327644348,
      "learning_rate": 1.91911602957849e-05,
      "loss": 0.2138,
      "step": 2390
    },
    {
      "epoch": 0.12157977562312808,
      "grad_norm": 0.9373264908790588,
      "learning_rate": 1.9189468162512482e-05,
      "loss": 0.2006,
      "step": 2395
    },
    {
      "epoch": 0.12183359561399056,
      "grad_norm": 1.0025991201400757,
      "learning_rate": 1.9187776029240062e-05,
      "loss": 0.2473,
      "step": 2400
    },
    {
      "epoch": 0.12208741560485303,
      "grad_norm": 0.6057084202766418,
      "learning_rate": 1.918608389596765e-05,
      "loss": 0.2271,
      "step": 2405
    },
    {
      "epoch": 0.12234123559571553,
      "grad_norm": 0.7220245599746704,
      "learning_rate": 1.9184391762695233e-05,
      "loss": 0.2219,
      "step": 2410
    },
    {
      "epoch": 0.122595055586578,
      "grad_norm": 0.7604368329048157,
      "learning_rate": 1.9182699629422813e-05,
      "loss": 0.2019,
      "step": 2415
    },
    {
      "epoch": 0.12284887557744048,
      "grad_norm": 0.5124613046646118,
      "learning_rate": 1.91810074961504e-05,
      "loss": 0.2054,
      "step": 2420
    },
    {
      "epoch": 0.12310269556830296,
      "grad_norm": 0.6220692992210388,
      "learning_rate": 1.917931536287798e-05,
      "loss": 0.2201,
      "step": 2425
    },
    {
      "epoch": 0.12335651555916544,
      "grad_norm": 1.0920076370239258,
      "learning_rate": 1.9177623229605564e-05,
      "loss": 0.2245,
      "step": 2430
    },
    {
      "epoch": 0.12361033555002791,
      "grad_norm": 0.9251731038093567,
      "learning_rate": 1.917593109633315e-05,
      "loss": 0.2204,
      "step": 2435
    },
    {
      "epoch": 0.1238641555408904,
      "grad_norm": 0.9124245047569275,
      "learning_rate": 1.917423896306073e-05,
      "loss": 0.215,
      "step": 2440
    },
    {
      "epoch": 0.12411797553175288,
      "grad_norm": 2.2037713527679443,
      "learning_rate": 1.9172546829788318e-05,
      "loss": 0.2279,
      "step": 2445
    },
    {
      "epoch": 0.12437179552261536,
      "grad_norm": 0.696209728717804,
      "learning_rate": 1.9170854696515898e-05,
      "loss": 0.2353,
      "step": 2450
    },
    {
      "epoch": 0.12462561551347784,
      "grad_norm": 0.8244996070861816,
      "learning_rate": 1.916916256324348e-05,
      "loss": 0.2185,
      "step": 2455
    },
    {
      "epoch": 0.12487943550434032,
      "grad_norm": 1.1219316720962524,
      "learning_rate": 1.9167470429971068e-05,
      "loss": 0.2361,
      "step": 2460
    },
    {
      "epoch": 0.1251332554952028,
      "grad_norm": 0.6167690753936768,
      "learning_rate": 1.916577829669865e-05,
      "loss": 0.2075,
      "step": 2465
    },
    {
      "epoch": 0.12538707548606529,
      "grad_norm": 0.663831889629364,
      "learning_rate": 1.9164086163426232e-05,
      "loss": 0.2179,
      "step": 2470
    },
    {
      "epoch": 0.12564089547692775,
      "grad_norm": 0.7804519534111023,
      "learning_rate": 1.9162394030153816e-05,
      "loss": 0.2319,
      "step": 2475
    },
    {
      "epoch": 0.12589471546779024,
      "grad_norm": 0.5112632513046265,
      "learning_rate": 1.91607018968814e-05,
      "loss": 0.2121,
      "step": 2480
    },
    {
      "epoch": 0.12614853545865273,
      "grad_norm": 1.0492627620697021,
      "learning_rate": 1.9159009763608986e-05,
      "loss": 0.1928,
      "step": 2485
    },
    {
      "epoch": 0.1264023554495152,
      "grad_norm": 0.8314067721366882,
      "learning_rate": 1.9157317630336566e-05,
      "loss": 0.2102,
      "step": 2490
    },
    {
      "epoch": 0.1266561754403777,
      "grad_norm": 0.6615179181098938,
      "learning_rate": 1.915562549706415e-05,
      "loss": 0.2098,
      "step": 2495
    },
    {
      "epoch": 0.12690999543124015,
      "grad_norm": 1.4945785999298096,
      "learning_rate": 1.9153933363791733e-05,
      "loss": 0.2256,
      "step": 2500
    },
    {
      "epoch": 0.12716381542210264,
      "grad_norm": 0.6474151015281677,
      "learning_rate": 1.9152241230519317e-05,
      "loss": 0.2119,
      "step": 2505
    },
    {
      "epoch": 0.12741763541296514,
      "grad_norm": 0.7189993858337402,
      "learning_rate": 1.91505490972469e-05,
      "loss": 0.1996,
      "step": 2510
    },
    {
      "epoch": 0.1276714554038276,
      "grad_norm": 0.6964658498764038,
      "learning_rate": 1.9148856963974484e-05,
      "loss": 0.247,
      "step": 2515
    },
    {
      "epoch": 0.1279252753946901,
      "grad_norm": 0.5863429307937622,
      "learning_rate": 1.9147164830702067e-05,
      "loss": 0.2197,
      "step": 2520
    },
    {
      "epoch": 0.12817909538555255,
      "grad_norm": 0.7309147715568542,
      "learning_rate": 1.914547269742965e-05,
      "loss": 0.2016,
      "step": 2525
    },
    {
      "epoch": 0.12843291537641505,
      "grad_norm": 0.6691656112670898,
      "learning_rate": 1.9143780564157235e-05,
      "loss": 0.2326,
      "step": 2530
    },
    {
      "epoch": 0.12868673536727754,
      "grad_norm": 1.2683653831481934,
      "learning_rate": 1.9142088430884818e-05,
      "loss": 0.214,
      "step": 2535
    },
    {
      "epoch": 0.12894055535814,
      "grad_norm": 0.9203475713729858,
      "learning_rate": 1.91403962976124e-05,
      "loss": 0.2125,
      "step": 2540
    },
    {
      "epoch": 0.1291943753490025,
      "grad_norm": 0.636577308177948,
      "learning_rate": 1.9138704164339985e-05,
      "loss": 0.2013,
      "step": 2545
    },
    {
      "epoch": 0.12944819533986496,
      "grad_norm": 0.5804703235626221,
      "learning_rate": 1.913701203106757e-05,
      "loss": 0.2053,
      "step": 2550
    },
    {
      "epoch": 0.12970201533072745,
      "grad_norm": 0.5855862498283386,
      "learning_rate": 1.9135319897795152e-05,
      "loss": 0.1958,
      "step": 2555
    },
    {
      "epoch": 0.12995583532158994,
      "grad_norm": 0.7482487559318542,
      "learning_rate": 1.9133627764522736e-05,
      "loss": 0.2257,
      "step": 2560
    },
    {
      "epoch": 0.1302096553124524,
      "grad_norm": 0.7025090456008911,
      "learning_rate": 1.913193563125032e-05,
      "loss": 0.226,
      "step": 2565
    },
    {
      "epoch": 0.1304634753033149,
      "grad_norm": 0.8828746676445007,
      "learning_rate": 1.9130243497977903e-05,
      "loss": 0.2187,
      "step": 2570
    },
    {
      "epoch": 0.13071729529417736,
      "grad_norm": 1.0203245878219604,
      "learning_rate": 1.9128551364705486e-05,
      "loss": 0.2288,
      "step": 2575
    },
    {
      "epoch": 0.13097111528503985,
      "grad_norm": 0.6295384168624878,
      "learning_rate": 1.912685923143307e-05,
      "loss": 0.2034,
      "step": 2580
    },
    {
      "epoch": 0.13122493527590234,
      "grad_norm": 1.2643804550170898,
      "learning_rate": 1.9125167098160654e-05,
      "loss": 0.2278,
      "step": 2585
    },
    {
      "epoch": 0.1314787552667648,
      "grad_norm": 0.6384063959121704,
      "learning_rate": 1.9123474964888237e-05,
      "loss": 0.1976,
      "step": 2590
    },
    {
      "epoch": 0.1317325752576273,
      "grad_norm": 0.6057654619216919,
      "learning_rate": 1.912178283161582e-05,
      "loss": 0.2108,
      "step": 2595
    },
    {
      "epoch": 0.13198639524848976,
      "grad_norm": 0.6316978335380554,
      "learning_rate": 1.9120090698343404e-05,
      "loss": 0.2051,
      "step": 2600
    },
    {
      "epoch": 0.13224021523935225,
      "grad_norm": 0.6967063546180725,
      "learning_rate": 1.9118398565070984e-05,
      "loss": 0.2016,
      "step": 2605
    },
    {
      "epoch": 0.13249403523021472,
      "grad_norm": 0.8182034492492676,
      "learning_rate": 1.911670643179857e-05,
      "loss": 0.1894,
      "step": 2610
    },
    {
      "epoch": 0.1327478552210772,
      "grad_norm": 0.5409026741981506,
      "learning_rate": 1.9115014298526155e-05,
      "loss": 0.2166,
      "step": 2615
    },
    {
      "epoch": 0.1330016752119397,
      "grad_norm": 0.7620669007301331,
      "learning_rate": 1.9113322165253735e-05,
      "loss": 0.2018,
      "step": 2620
    },
    {
      "epoch": 0.13325549520280217,
      "grad_norm": 0.8058112859725952,
      "learning_rate": 1.9111630031981322e-05,
      "loss": 0.1956,
      "step": 2625
    },
    {
      "epoch": 0.13350931519366466,
      "grad_norm": 0.9924262166023254,
      "learning_rate": 1.9109937898708902e-05,
      "loss": 0.2005,
      "step": 2630
    },
    {
      "epoch": 0.13376313518452712,
      "grad_norm": 0.7677115797996521,
      "learning_rate": 1.910824576543649e-05,
      "loss": 0.2136,
      "step": 2635
    },
    {
      "epoch": 0.1340169551753896,
      "grad_norm": 0.8996549844741821,
      "learning_rate": 1.9106553632164072e-05,
      "loss": 0.2144,
      "step": 2640
    },
    {
      "epoch": 0.1342707751662521,
      "grad_norm": 1.055097222328186,
      "learning_rate": 1.9104861498891653e-05,
      "loss": 0.199,
      "step": 2645
    },
    {
      "epoch": 0.13452459515711457,
      "grad_norm": 0.6172511577606201,
      "learning_rate": 1.910316936561924e-05,
      "loss": 0.2001,
      "step": 2650
    },
    {
      "epoch": 0.13477841514797706,
      "grad_norm": 0.7040294408798218,
      "learning_rate": 1.910147723234682e-05,
      "loss": 0.2091,
      "step": 2655
    },
    {
      "epoch": 0.13503223513883952,
      "grad_norm": 0.6608620285987854,
      "learning_rate": 1.9099785099074403e-05,
      "loss": 0.1907,
      "step": 2660
    },
    {
      "epoch": 0.13528605512970202,
      "grad_norm": 0.7029122114181519,
      "learning_rate": 1.909809296580199e-05,
      "loss": 0.207,
      "step": 2665
    },
    {
      "epoch": 0.1355398751205645,
      "grad_norm": 0.7303573489189148,
      "learning_rate": 1.909640083252957e-05,
      "loss": 0.2103,
      "step": 2670
    },
    {
      "epoch": 0.13579369511142697,
      "grad_norm": 0.7197701334953308,
      "learning_rate": 1.9094708699257154e-05,
      "loss": 0.198,
      "step": 2675
    },
    {
      "epoch": 0.13604751510228946,
      "grad_norm": 2.0488266944885254,
      "learning_rate": 1.9093016565984737e-05,
      "loss": 0.2288,
      "step": 2680
    },
    {
      "epoch": 0.13630133509315193,
      "grad_norm": 0.7889509201049805,
      "learning_rate": 1.909132443271232e-05,
      "loss": 0.2184,
      "step": 2685
    },
    {
      "epoch": 0.13655515508401442,
      "grad_norm": 0.8902899622917175,
      "learning_rate": 1.9089632299439908e-05,
      "loss": 0.2161,
      "step": 2690
    },
    {
      "epoch": 0.1368089750748769,
      "grad_norm": 0.6209053993225098,
      "learning_rate": 1.9087940166167488e-05,
      "loss": 0.1956,
      "step": 2695
    },
    {
      "epoch": 0.13706279506573937,
      "grad_norm": 0.5760391354560852,
      "learning_rate": 1.908624803289507e-05,
      "loss": 0.2005,
      "step": 2700
    },
    {
      "epoch": 0.13731661505660187,
      "grad_norm": 0.9128169417381287,
      "learning_rate": 1.9084555899622655e-05,
      "loss": 0.2036,
      "step": 2705
    },
    {
      "epoch": 0.13757043504746433,
      "grad_norm": 0.8320951461791992,
      "learning_rate": 1.908286376635024e-05,
      "loss": 0.2108,
      "step": 2710
    },
    {
      "epoch": 0.13782425503832682,
      "grad_norm": 0.7470078468322754,
      "learning_rate": 1.9081171633077822e-05,
      "loss": 0.2058,
      "step": 2715
    },
    {
      "epoch": 0.1380780750291893,
      "grad_norm": 0.7988982796669006,
      "learning_rate": 1.9079479499805406e-05,
      "loss": 0.242,
      "step": 2720
    },
    {
      "epoch": 0.13833189502005178,
      "grad_norm": 0.8993115425109863,
      "learning_rate": 1.907778736653299e-05,
      "loss": 0.212,
      "step": 2725
    },
    {
      "epoch": 0.13858571501091427,
      "grad_norm": 0.6929076313972473,
      "learning_rate": 1.9076095233260573e-05,
      "loss": 0.2509,
      "step": 2730
    },
    {
      "epoch": 0.13883953500177673,
      "grad_norm": 0.6944275498390198,
      "learning_rate": 1.9074403099988156e-05,
      "loss": 0.2116,
      "step": 2735
    },
    {
      "epoch": 0.13909335499263922,
      "grad_norm": 0.6491143107414246,
      "learning_rate": 1.907271096671574e-05,
      "loss": 0.1943,
      "step": 2740
    },
    {
      "epoch": 0.1393471749835017,
      "grad_norm": 0.6222745776176453,
      "learning_rate": 1.9071018833443324e-05,
      "loss": 0.2144,
      "step": 2745
    },
    {
      "epoch": 0.13960099497436418,
      "grad_norm": 0.863029420375824,
      "learning_rate": 1.9069326700170907e-05,
      "loss": 0.1972,
      "step": 2750
    },
    {
      "epoch": 0.13985481496522667,
      "grad_norm": 0.5419744253158569,
      "learning_rate": 1.906763456689849e-05,
      "loss": 0.1913,
      "step": 2755
    },
    {
      "epoch": 0.14010863495608913,
      "grad_norm": 0.5654199719429016,
      "learning_rate": 1.9065942433626074e-05,
      "loss": 0.1994,
      "step": 2760
    },
    {
      "epoch": 0.14036245494695163,
      "grad_norm": 0.7003618478775024,
      "learning_rate": 1.9064250300353658e-05,
      "loss": 0.1955,
      "step": 2765
    },
    {
      "epoch": 0.1406162749378141,
      "grad_norm": 1.1119288206100464,
      "learning_rate": 1.906255816708124e-05,
      "loss": 0.1984,
      "step": 2770
    },
    {
      "epoch": 0.14087009492867658,
      "grad_norm": 0.496934711933136,
      "learning_rate": 1.9060866033808825e-05,
      "loss": 0.1855,
      "step": 2775
    },
    {
      "epoch": 0.14112391491953907,
      "grad_norm": 0.9058437943458557,
      "learning_rate": 1.905917390053641e-05,
      "loss": 0.188,
      "step": 2780
    },
    {
      "epoch": 0.14137773491040154,
      "grad_norm": 0.6056883335113525,
      "learning_rate": 1.9057481767263992e-05,
      "loss": 0.2088,
      "step": 2785
    },
    {
      "epoch": 0.14163155490126403,
      "grad_norm": 0.6516966223716736,
      "learning_rate": 1.9055789633991575e-05,
      "loss": 0.1905,
      "step": 2790
    },
    {
      "epoch": 0.1418853748921265,
      "grad_norm": 0.6040582060813904,
      "learning_rate": 1.905409750071916e-05,
      "loss": 0.2125,
      "step": 2795
    },
    {
      "epoch": 0.14213919488298898,
      "grad_norm": 0.5323441624641418,
      "learning_rate": 1.9052405367446743e-05,
      "loss": 0.1771,
      "step": 2800
    },
    {
      "epoch": 0.14239301487385148,
      "grad_norm": 0.4898989796638489,
      "learning_rate": 1.9050713234174326e-05,
      "loss": 0.1863,
      "step": 2805
    },
    {
      "epoch": 0.14264683486471394,
      "grad_norm": 0.5168548226356506,
      "learning_rate": 1.9049021100901906e-05,
      "loss": 0.1941,
      "step": 2810
    },
    {
      "epoch": 0.14290065485557643,
      "grad_norm": 0.5550641417503357,
      "learning_rate": 1.9047328967629493e-05,
      "loss": 0.2024,
      "step": 2815
    },
    {
      "epoch": 0.1431544748464389,
      "grad_norm": 0.8888419270515442,
      "learning_rate": 1.9045636834357077e-05,
      "loss": 0.2091,
      "step": 2820
    },
    {
      "epoch": 0.1434082948373014,
      "grad_norm": 0.6956480741500854,
      "learning_rate": 1.904394470108466e-05,
      "loss": 0.2006,
      "step": 2825
    },
    {
      "epoch": 0.14366211482816388,
      "grad_norm": 0.6455702781677246,
      "learning_rate": 1.9042252567812244e-05,
      "loss": 0.1969,
      "step": 2830
    },
    {
      "epoch": 0.14391593481902634,
      "grad_norm": 0.7298540472984314,
      "learning_rate": 1.9040560434539824e-05,
      "loss": 0.2041,
      "step": 2835
    },
    {
      "epoch": 0.14416975480988883,
      "grad_norm": 0.9407358169555664,
      "learning_rate": 1.903886830126741e-05,
      "loss": 0.2226,
      "step": 2840
    },
    {
      "epoch": 0.1444235748007513,
      "grad_norm": 0.6761470437049866,
      "learning_rate": 1.9037176167994994e-05,
      "loss": 0.2073,
      "step": 2845
    },
    {
      "epoch": 0.1446773947916138,
      "grad_norm": 0.9858958125114441,
      "learning_rate": 1.9035484034722575e-05,
      "loss": 0.2191,
      "step": 2850
    },
    {
      "epoch": 0.14493121478247628,
      "grad_norm": 0.7643491625785828,
      "learning_rate": 1.903379190145016e-05,
      "loss": 0.1855,
      "step": 2855
    },
    {
      "epoch": 0.14518503477333874,
      "grad_norm": 0.5545926690101624,
      "learning_rate": 1.903209976817774e-05,
      "loss": 0.1854,
      "step": 2860
    },
    {
      "epoch": 0.14543885476420124,
      "grad_norm": 0.7087584137916565,
      "learning_rate": 1.9030407634905325e-05,
      "loss": 0.1925,
      "step": 2865
    },
    {
      "epoch": 0.1456926747550637,
      "grad_norm": 0.9376761317253113,
      "learning_rate": 1.9028715501632912e-05,
      "loss": 0.2171,
      "step": 2870
    },
    {
      "epoch": 0.1459464947459262,
      "grad_norm": 0.5529223680496216,
      "learning_rate": 1.9027023368360492e-05,
      "loss": 0.1849,
      "step": 2875
    },
    {
      "epoch": 0.14620031473678866,
      "grad_norm": 0.6258545517921448,
      "learning_rate": 1.902533123508808e-05,
      "loss": 0.195,
      "step": 2880
    },
    {
      "epoch": 0.14645413472765115,
      "grad_norm": 0.5800721645355225,
      "learning_rate": 1.902363910181566e-05,
      "loss": 0.1879,
      "step": 2885
    },
    {
      "epoch": 0.14670795471851364,
      "grad_norm": 0.9016756415367126,
      "learning_rate": 1.9021946968543243e-05,
      "loss": 0.1978,
      "step": 2890
    },
    {
      "epoch": 0.1469617747093761,
      "grad_norm": 0.6598945260047913,
      "learning_rate": 1.902025483527083e-05,
      "loss": 0.2201,
      "step": 2895
    },
    {
      "epoch": 0.1472155947002386,
      "grad_norm": 0.5882731080055237,
      "learning_rate": 1.901856270199841e-05,
      "loss": 0.1942,
      "step": 2900
    },
    {
      "epoch": 0.14746941469110106,
      "grad_norm": 0.898997962474823,
      "learning_rate": 1.9016870568725994e-05,
      "loss": 0.2089,
      "step": 2905
    },
    {
      "epoch": 0.14772323468196355,
      "grad_norm": 0.547217071056366,
      "learning_rate": 1.9015178435453577e-05,
      "loss": 0.1957,
      "step": 2910
    },
    {
      "epoch": 0.14797705467282604,
      "grad_norm": 0.5666927099227905,
      "learning_rate": 1.901348630218116e-05,
      "loss": 0.2036,
      "step": 2915
    },
    {
      "epoch": 0.1482308746636885,
      "grad_norm": 0.7027495503425598,
      "learning_rate": 1.9011794168908744e-05,
      "loss": 0.2027,
      "step": 2920
    },
    {
      "epoch": 0.148484694654551,
      "grad_norm": 0.9105992317199707,
      "learning_rate": 1.9010102035636328e-05,
      "loss": 0.2185,
      "step": 2925
    },
    {
      "epoch": 0.14873851464541346,
      "grad_norm": 0.9565317630767822,
      "learning_rate": 1.900840990236391e-05,
      "loss": 0.1891,
      "step": 2930
    },
    {
      "epoch": 0.14899233463627595,
      "grad_norm": 0.6969782710075378,
      "learning_rate": 1.9006717769091495e-05,
      "loss": 0.215,
      "step": 2935
    },
    {
      "epoch": 0.14924615462713844,
      "grad_norm": 0.5370941758155823,
      "learning_rate": 1.900502563581908e-05,
      "loss": 0.1872,
      "step": 2940
    },
    {
      "epoch": 0.1494999746180009,
      "grad_norm": 0.932841420173645,
      "learning_rate": 1.9003333502546662e-05,
      "loss": 0.2144,
      "step": 2945
    },
    {
      "epoch": 0.1497537946088634,
      "grad_norm": 0.6696744561195374,
      "learning_rate": 1.9001641369274245e-05,
      "loss": 0.2132,
      "step": 2950
    },
    {
      "epoch": 0.15000761459972586,
      "grad_norm": 0.8368297219276428,
      "learning_rate": 1.899994923600183e-05,
      "loss": 0.1922,
      "step": 2955
    },
    {
      "epoch": 0.15026143459058836,
      "grad_norm": 0.7820245623588562,
      "learning_rate": 1.8998257102729413e-05,
      "loss": 0.1922,
      "step": 2960
    },
    {
      "epoch": 0.15051525458145085,
      "grad_norm": 0.6239885091781616,
      "learning_rate": 1.8996564969456996e-05,
      "loss": 0.2171,
      "step": 2965
    },
    {
      "epoch": 0.1507690745723133,
      "grad_norm": 0.6773269176483154,
      "learning_rate": 1.899487283618458e-05,
      "loss": 0.2039,
      "step": 2970
    },
    {
      "epoch": 0.1510228945631758,
      "grad_norm": 0.6102594137191772,
      "learning_rate": 1.8993180702912163e-05,
      "loss": 0.1843,
      "step": 2975
    },
    {
      "epoch": 0.15127671455403827,
      "grad_norm": 0.7514229416847229,
      "learning_rate": 1.8991488569639747e-05,
      "loss": 0.1819,
      "step": 2980
    },
    {
      "epoch": 0.15153053454490076,
      "grad_norm": 3.802446126937866,
      "learning_rate": 1.898979643636733e-05,
      "loss": 0.1858,
      "step": 2985
    },
    {
      "epoch": 0.15178435453576325,
      "grad_norm": 0.984550416469574,
      "learning_rate": 1.8988104303094914e-05,
      "loss": 0.1861,
      "step": 2990
    },
    {
      "epoch": 0.1520381745266257,
      "grad_norm": 0.672860324382782,
      "learning_rate": 1.8986412169822497e-05,
      "loss": 0.1966,
      "step": 2995
    },
    {
      "epoch": 0.1522919945174882,
      "grad_norm": 0.5952281355857849,
      "learning_rate": 1.898472003655008e-05,
      "loss": 0.1876,
      "step": 3000
    },
    {
      "epoch": 0.15254581450835067,
      "grad_norm": 0.8867749571800232,
      "learning_rate": 1.8983027903277664e-05,
      "loss": 0.2116,
      "step": 3005
    },
    {
      "epoch": 0.15279963449921316,
      "grad_norm": 0.7406168580055237,
      "learning_rate": 1.8981335770005248e-05,
      "loss": 0.1867,
      "step": 3010
    },
    {
      "epoch": 0.15305345449007565,
      "grad_norm": 0.8236103653907776,
      "learning_rate": 1.8979643636732828e-05,
      "loss": 0.2004,
      "step": 3015
    },
    {
      "epoch": 0.15330727448093812,
      "grad_norm": 0.5611258149147034,
      "learning_rate": 1.8977951503460415e-05,
      "loss": 0.1773,
      "step": 3020
    },
    {
      "epoch": 0.1535610944718006,
      "grad_norm": 0.8456403017044067,
      "learning_rate": 1.8976259370188e-05,
      "loss": 0.2209,
      "step": 3025
    },
    {
      "epoch": 0.15381491446266307,
      "grad_norm": 1.378568172454834,
      "learning_rate": 1.8974567236915582e-05,
      "loss": 0.2122,
      "step": 3030
    },
    {
      "epoch": 0.15406873445352556,
      "grad_norm": 0.599615216255188,
      "learning_rate": 1.8972875103643166e-05,
      "loss": 0.1886,
      "step": 3035
    },
    {
      "epoch": 0.15432255444438803,
      "grad_norm": 0.5790411233901978,
      "learning_rate": 1.8971182970370746e-05,
      "loss": 0.1923,
      "step": 3040
    },
    {
      "epoch": 0.15457637443525052,
      "grad_norm": 5.2787299156188965,
      "learning_rate": 1.8969490837098333e-05,
      "loss": 0.1875,
      "step": 3045
    },
    {
      "epoch": 0.154830194426113,
      "grad_norm": 0.5498223304748535,
      "learning_rate": 1.8967798703825916e-05,
      "loss": 0.1898,
      "step": 3050
    },
    {
      "epoch": 0.15508401441697547,
      "grad_norm": 0.5836355686187744,
      "learning_rate": 1.8966106570553497e-05,
      "loss": 0.1873,
      "step": 3055
    },
    {
      "epoch": 0.15533783440783797,
      "grad_norm": 0.6591739654541016,
      "learning_rate": 1.8964414437281083e-05,
      "loss": 0.2141,
      "step": 3060
    },
    {
      "epoch": 0.15559165439870043,
      "grad_norm": 0.5879199504852295,
      "learning_rate": 1.8962722304008664e-05,
      "loss": 0.199,
      "step": 3065
    },
    {
      "epoch": 0.15584547438956292,
      "grad_norm": 0.6252302527427673,
      "learning_rate": 1.8961030170736247e-05,
      "loss": 0.183,
      "step": 3070
    },
    {
      "epoch": 0.1560992943804254,
      "grad_norm": 0.6322395205497742,
      "learning_rate": 1.8959338037463834e-05,
      "loss": 0.2204,
      "step": 3075
    },
    {
      "epoch": 0.15635311437128788,
      "grad_norm": 0.8022140860557556,
      "learning_rate": 1.8957645904191414e-05,
      "loss": 0.1974,
      "step": 3080
    },
    {
      "epoch": 0.15660693436215037,
      "grad_norm": 0.7623772025108337,
      "learning_rate": 1.8955953770919e-05,
      "loss": 0.1824,
      "step": 3085
    },
    {
      "epoch": 0.15686075435301283,
      "grad_norm": 0.6760655641555786,
      "learning_rate": 1.895426163764658e-05,
      "loss": 0.2041,
      "step": 3090
    },
    {
      "epoch": 0.15711457434387532,
      "grad_norm": 0.6074882745742798,
      "learning_rate": 1.8952569504374165e-05,
      "loss": 0.19,
      "step": 3095
    },
    {
      "epoch": 0.15736839433473782,
      "grad_norm": 0.5363246202468872,
      "learning_rate": 1.8950877371101752e-05,
      "loss": 0.1732,
      "step": 3100
    },
    {
      "epoch": 0.15762221432560028,
      "grad_norm": 0.9593762159347534,
      "learning_rate": 1.8949185237829332e-05,
      "loss": 0.1973,
      "step": 3105
    },
    {
      "epoch": 0.15787603431646277,
      "grad_norm": 3.48103666305542,
      "learning_rate": 1.8947493104556916e-05,
      "loss": 0.2306,
      "step": 3110
    },
    {
      "epoch": 0.15812985430732523,
      "grad_norm": 0.593743085861206,
      "learning_rate": 1.89458009712845e-05,
      "loss": 0.2023,
      "step": 3115
    },
    {
      "epoch": 0.15838367429818773,
      "grad_norm": 0.8416429758071899,
      "learning_rate": 1.8944108838012083e-05,
      "loss": 0.1661,
      "step": 3120
    },
    {
      "epoch": 0.15863749428905022,
      "grad_norm": 0.5875362157821655,
      "learning_rate": 1.894241670473967e-05,
      "loss": 0.1693,
      "step": 3125
    },
    {
      "epoch": 0.15889131427991268,
      "grad_norm": 0.5337428450584412,
      "learning_rate": 1.894072457146725e-05,
      "loss": 0.2014,
      "step": 3130
    },
    {
      "epoch": 0.15914513427077517,
      "grad_norm": 0.705847978591919,
      "learning_rate": 1.8939032438194833e-05,
      "loss": 0.1769,
      "step": 3135
    },
    {
      "epoch": 0.15939895426163764,
      "grad_norm": 1.11298406124115,
      "learning_rate": 1.8937340304922417e-05,
      "loss": 0.1957,
      "step": 3140
    },
    {
      "epoch": 0.15965277425250013,
      "grad_norm": 0.7969094514846802,
      "learning_rate": 1.893564817165e-05,
      "loss": 0.2062,
      "step": 3145
    },
    {
      "epoch": 0.15990659424336262,
      "grad_norm": 0.6891659498214722,
      "learning_rate": 1.8933956038377584e-05,
      "loss": 0.1837,
      "step": 3150
    },
    {
      "epoch": 0.16016041423422508,
      "grad_norm": 0.7349233627319336,
      "learning_rate": 1.8932263905105167e-05,
      "loss": 0.1962,
      "step": 3155
    },
    {
      "epoch": 0.16041423422508758,
      "grad_norm": 0.6409569382667542,
      "learning_rate": 1.893057177183275e-05,
      "loss": 0.1762,
      "step": 3160
    },
    {
      "epoch": 0.16066805421595004,
      "grad_norm": 0.6636890769004822,
      "learning_rate": 1.8928879638560335e-05,
      "loss": 0.1863,
      "step": 3165
    },
    {
      "epoch": 0.16092187420681253,
      "grad_norm": 0.6937609314918518,
      "learning_rate": 1.8927187505287918e-05,
      "loss": 0.1784,
      "step": 3170
    },
    {
      "epoch": 0.161175694197675,
      "grad_norm": 0.5253648161888123,
      "learning_rate": 1.89254953720155e-05,
      "loss": 0.1917,
      "step": 3175
    },
    {
      "epoch": 0.1614295141885375,
      "grad_norm": 0.6246563196182251,
      "learning_rate": 1.8923803238743085e-05,
      "loss": 0.1687,
      "step": 3180
    },
    {
      "epoch": 0.16168333417939998,
      "grad_norm": 0.664107084274292,
      "learning_rate": 1.892211110547067e-05,
      "loss": 0.1761,
      "step": 3185
    },
    {
      "epoch": 0.16193715417026244,
      "grad_norm": 0.7611233592033386,
      "learning_rate": 1.8920418972198252e-05,
      "loss": 0.1977,
      "step": 3190
    },
    {
      "epoch": 0.16219097416112493,
      "grad_norm": 0.6064574718475342,
      "learning_rate": 1.8918726838925836e-05,
      "loss": 0.1824,
      "step": 3195
    },
    {
      "epoch": 0.1624447941519874,
      "grad_norm": 0.8105899691581726,
      "learning_rate": 1.891703470565342e-05,
      "loss": 0.1892,
      "step": 3200
    },
    {
      "epoch": 0.1626986141428499,
      "grad_norm": 0.6670052409172058,
      "learning_rate": 1.8915342572381003e-05,
      "loss": 0.1657,
      "step": 3205
    },
    {
      "epoch": 0.16295243413371238,
      "grad_norm": 0.8309715986251831,
      "learning_rate": 1.8913650439108586e-05,
      "loss": 0.1826,
      "step": 3210
    },
    {
      "epoch": 0.16320625412457485,
      "grad_norm": 0.6957319378852844,
      "learning_rate": 1.891195830583617e-05,
      "loss": 0.1981,
      "step": 3215
    },
    {
      "epoch": 0.16346007411543734,
      "grad_norm": 0.5244278311729431,
      "learning_rate": 1.8910266172563753e-05,
      "loss": 0.1835,
      "step": 3220
    },
    {
      "epoch": 0.1637138941062998,
      "grad_norm": 0.6687745451927185,
      "learning_rate": 1.8908574039291337e-05,
      "loss": 0.1818,
      "step": 3225
    },
    {
      "epoch": 0.1639677140971623,
      "grad_norm": 0.5225896835327148,
      "learning_rate": 1.890688190601892e-05,
      "loss": 0.1972,
      "step": 3230
    },
    {
      "epoch": 0.16422153408802478,
      "grad_norm": 0.71306973695755,
      "learning_rate": 1.8905189772746504e-05,
      "loss": 0.1793,
      "step": 3235
    },
    {
      "epoch": 0.16447535407888725,
      "grad_norm": 0.6916504502296448,
      "learning_rate": 1.8903497639474088e-05,
      "loss": 0.1796,
      "step": 3240
    },
    {
      "epoch": 0.16472917406974974,
      "grad_norm": 0.7372540831565857,
      "learning_rate": 1.8901805506201668e-05,
      "loss": 0.1891,
      "step": 3245
    },
    {
      "epoch": 0.1649829940606122,
      "grad_norm": 0.5736752152442932,
      "learning_rate": 1.8900113372929255e-05,
      "loss": 0.2053,
      "step": 3250
    },
    {
      "epoch": 0.1652368140514747,
      "grad_norm": 0.5286284685134888,
      "learning_rate": 1.889842123965684e-05,
      "loss": 0.1989,
      "step": 3255
    },
    {
      "epoch": 0.1654906340423372,
      "grad_norm": 0.9421578049659729,
      "learning_rate": 1.889672910638442e-05,
      "loss": 0.2002,
      "step": 3260
    },
    {
      "epoch": 0.16574445403319965,
      "grad_norm": 0.721328616142273,
      "learning_rate": 1.8895036973112005e-05,
      "loss": 0.1938,
      "step": 3265
    },
    {
      "epoch": 0.16599827402406214,
      "grad_norm": 0.6766708493232727,
      "learning_rate": 1.8893344839839586e-05,
      "loss": 0.1981,
      "step": 3270
    },
    {
      "epoch": 0.1662520940149246,
      "grad_norm": 0.7163949608802795,
      "learning_rate": 1.8891652706567172e-05,
      "loss": 0.1856,
      "step": 3275
    },
    {
      "epoch": 0.1665059140057871,
      "grad_norm": 0.673416793346405,
      "learning_rate": 1.8889960573294756e-05,
      "loss": 0.1928,
      "step": 3280
    },
    {
      "epoch": 0.1667597339966496,
      "grad_norm": 0.6823815703392029,
      "learning_rate": 1.8888268440022336e-05,
      "loss": 0.1953,
      "step": 3285
    },
    {
      "epoch": 0.16701355398751205,
      "grad_norm": 0.6587371826171875,
      "learning_rate": 1.8886576306749923e-05,
      "loss": 0.1985,
      "step": 3290
    },
    {
      "epoch": 0.16726737397837455,
      "grad_norm": 0.7395918965339661,
      "learning_rate": 1.8884884173477503e-05,
      "loss": 0.1859,
      "step": 3295
    },
    {
      "epoch": 0.167521193969237,
      "grad_norm": 0.6410045623779297,
      "learning_rate": 1.8883192040205087e-05,
      "loss": 0.1916,
      "step": 3300
    },
    {
      "epoch": 0.1677750139600995,
      "grad_norm": 0.5092994570732117,
      "learning_rate": 1.8881499906932674e-05,
      "loss": 0.1782,
      "step": 3305
    },
    {
      "epoch": 0.16802883395096196,
      "grad_norm": 0.6259738206863403,
      "learning_rate": 1.8879807773660254e-05,
      "loss": 0.2074,
      "step": 3310
    },
    {
      "epoch": 0.16828265394182446,
      "grad_norm": 0.8457812070846558,
      "learning_rate": 1.8878115640387837e-05,
      "loss": 0.1983,
      "step": 3315
    },
    {
      "epoch": 0.16853647393268695,
      "grad_norm": 0.4995182752609253,
      "learning_rate": 1.887642350711542e-05,
      "loss": 0.1781,
      "step": 3320
    },
    {
      "epoch": 0.1687902939235494,
      "grad_norm": 0.6008071899414062,
      "learning_rate": 1.8874731373843005e-05,
      "loss": 0.1904,
      "step": 3325
    },
    {
      "epoch": 0.1690441139144119,
      "grad_norm": 0.7149403691291809,
      "learning_rate": 1.8873039240570588e-05,
      "loss": 0.1846,
      "step": 3330
    },
    {
      "epoch": 0.16929793390527437,
      "grad_norm": 1.0425550937652588,
      "learning_rate": 1.887134710729817e-05,
      "loss": 0.1979,
      "step": 3335
    },
    {
      "epoch": 0.16955175389613686,
      "grad_norm": 0.6860085725784302,
      "learning_rate": 1.8869654974025755e-05,
      "loss": 0.1808,
      "step": 3340
    },
    {
      "epoch": 0.16980557388699935,
      "grad_norm": 0.8324023485183716,
      "learning_rate": 1.886796284075334e-05,
      "loss": 0.2056,
      "step": 3345
    },
    {
      "epoch": 0.17005939387786181,
      "grad_norm": 0.6411967873573303,
      "learning_rate": 1.8866270707480922e-05,
      "loss": 0.1771,
      "step": 3350
    },
    {
      "epoch": 0.1703132138687243,
      "grad_norm": 0.5328414440155029,
      "learning_rate": 1.8864578574208506e-05,
      "loss": 0.1957,
      "step": 3355
    },
    {
      "epoch": 0.17056703385958677,
      "grad_norm": 0.7603042125701904,
      "learning_rate": 1.886288644093609e-05,
      "loss": 0.2017,
      "step": 3360
    },
    {
      "epoch": 0.17082085385044926,
      "grad_norm": 0.8825428485870361,
      "learning_rate": 1.8861194307663673e-05,
      "loss": 0.1989,
      "step": 3365
    },
    {
      "epoch": 0.17107467384131175,
      "grad_norm": 0.657129168510437,
      "learning_rate": 1.8859502174391256e-05,
      "loss": 0.1801,
      "step": 3370
    },
    {
      "epoch": 0.17132849383217422,
      "grad_norm": 0.7885820269584656,
      "learning_rate": 1.885781004111884e-05,
      "loss": 0.2056,
      "step": 3375
    },
    {
      "epoch": 0.1715823138230367,
      "grad_norm": 0.5244271159172058,
      "learning_rate": 1.8856117907846424e-05,
      "loss": 0.1646,
      "step": 3380
    },
    {
      "epoch": 0.17183613381389917,
      "grad_norm": 0.6291254758834839,
      "learning_rate": 1.8854425774574007e-05,
      "loss": 0.1832,
      "step": 3385
    },
    {
      "epoch": 0.17208995380476166,
      "grad_norm": 0.6272834539413452,
      "learning_rate": 1.885273364130159e-05,
      "loss": 0.1888,
      "step": 3390
    },
    {
      "epoch": 0.17234377379562416,
      "grad_norm": 0.727254331111908,
      "learning_rate": 1.8851041508029174e-05,
      "loss": 0.1812,
      "step": 3395
    },
    {
      "epoch": 0.17259759378648662,
      "grad_norm": 0.8115093111991882,
      "learning_rate": 1.8849349374756758e-05,
      "loss": 0.1872,
      "step": 3400
    },
    {
      "epoch": 0.1728514137773491,
      "grad_norm": 0.6561703085899353,
      "learning_rate": 1.884765724148434e-05,
      "loss": 0.1813,
      "step": 3405
    },
    {
      "epoch": 0.17310523376821157,
      "grad_norm": 0.5098863840103149,
      "learning_rate": 1.8845965108211925e-05,
      "loss": 0.1873,
      "step": 3410
    },
    {
      "epoch": 0.17335905375907407,
      "grad_norm": 0.6273576021194458,
      "learning_rate": 1.884427297493951e-05,
      "loss": 0.1893,
      "step": 3415
    },
    {
      "epoch": 0.17361287374993656,
      "grad_norm": 0.622138261795044,
      "learning_rate": 1.8842580841667092e-05,
      "loss": 0.1596,
      "step": 3420
    },
    {
      "epoch": 0.17386669374079902,
      "grad_norm": 0.8276758790016174,
      "learning_rate": 1.8840888708394675e-05,
      "loss": 0.1774,
      "step": 3425
    },
    {
      "epoch": 0.1741205137316615,
      "grad_norm": 0.6109856963157654,
      "learning_rate": 1.883919657512226e-05,
      "loss": 0.1927,
      "step": 3430
    },
    {
      "epoch": 0.17437433372252398,
      "grad_norm": 0.6221029758453369,
      "learning_rate": 1.8837504441849843e-05,
      "loss": 0.1951,
      "step": 3435
    },
    {
      "epoch": 0.17462815371338647,
      "grad_norm": 0.567482590675354,
      "learning_rate": 1.8835812308577426e-05,
      "loss": 0.1848,
      "step": 3440
    },
    {
      "epoch": 0.17488197370424893,
      "grad_norm": 0.7713445425033569,
      "learning_rate": 1.883412017530501e-05,
      "loss": 0.1676,
      "step": 3445
    },
    {
      "epoch": 0.17513579369511142,
      "grad_norm": 0.6191247701644897,
      "learning_rate": 1.883242804203259e-05,
      "loss": 0.1867,
      "step": 3450
    },
    {
      "epoch": 0.17538961368597392,
      "grad_norm": 0.40963295102119446,
      "learning_rate": 1.8830735908760177e-05,
      "loss": 0.1683,
      "step": 3455
    },
    {
      "epoch": 0.17564343367683638,
      "grad_norm": 0.5506500601768494,
      "learning_rate": 1.882904377548776e-05,
      "loss": 0.1755,
      "step": 3460
    },
    {
      "epoch": 0.17589725366769887,
      "grad_norm": 0.5656499266624451,
      "learning_rate": 1.8827351642215344e-05,
      "loss": 0.1856,
      "step": 3465
    },
    {
      "epoch": 0.17615107365856134,
      "grad_norm": 0.9259144067764282,
      "learning_rate": 1.8825659508942927e-05,
      "loss": 0.1683,
      "step": 3470
    },
    {
      "epoch": 0.17640489364942383,
      "grad_norm": 0.6214718818664551,
      "learning_rate": 1.8823967375670507e-05,
      "loss": 0.1877,
      "step": 3475
    },
    {
      "epoch": 0.17665871364028632,
      "grad_norm": 0.6122065186500549,
      "learning_rate": 1.8822275242398094e-05,
      "loss": 0.1823,
      "step": 3480
    },
    {
      "epoch": 0.17691253363114878,
      "grad_norm": 0.5101046562194824,
      "learning_rate": 1.8820583109125678e-05,
      "loss": 0.1558,
      "step": 3485
    },
    {
      "epoch": 0.17716635362201127,
      "grad_norm": 0.6742383241653442,
      "learning_rate": 1.8818890975853258e-05,
      "loss": 0.1767,
      "step": 3490
    },
    {
      "epoch": 0.17742017361287374,
      "grad_norm": 0.7784730792045593,
      "learning_rate": 1.8817198842580845e-05,
      "loss": 0.1944,
      "step": 3495
    },
    {
      "epoch": 0.17767399360373623,
      "grad_norm": 0.5944966673851013,
      "learning_rate": 1.8815506709308425e-05,
      "loss": 0.1625,
      "step": 3500
    },
    {
      "epoch": 0.17792781359459872,
      "grad_norm": 0.5932590961456299,
      "learning_rate": 1.881381457603601e-05,
      "loss": 0.1994,
      "step": 3505
    },
    {
      "epoch": 0.17818163358546119,
      "grad_norm": 0.7129168510437012,
      "learning_rate": 1.8812122442763596e-05,
      "loss": 0.2029,
      "step": 3510
    },
    {
      "epoch": 0.17843545357632368,
      "grad_norm": 1.4822226762771606,
      "learning_rate": 1.8810430309491176e-05,
      "loss": 0.1733,
      "step": 3515
    },
    {
      "epoch": 0.17868927356718614,
      "grad_norm": 0.5432773232460022,
      "learning_rate": 1.8808738176218763e-05,
      "loss": 0.1995,
      "step": 3520
    },
    {
      "epoch": 0.17894309355804863,
      "grad_norm": 0.6018402576446533,
      "learning_rate": 1.8807046042946343e-05,
      "loss": 0.1706,
      "step": 3525
    },
    {
      "epoch": 0.17919691354891112,
      "grad_norm": 0.7911956906318665,
      "learning_rate": 1.8805353909673926e-05,
      "loss": 0.2016,
      "step": 3530
    },
    {
      "epoch": 0.1794507335397736,
      "grad_norm": 0.9234727621078491,
      "learning_rate": 1.880366177640151e-05,
      "loss": 0.1936,
      "step": 3535
    },
    {
      "epoch": 0.17970455353063608,
      "grad_norm": 0.5961397886276245,
      "learning_rate": 1.8801969643129094e-05,
      "loss": 0.1628,
      "step": 3540
    },
    {
      "epoch": 0.17995837352149854,
      "grad_norm": 0.689795196056366,
      "learning_rate": 1.8800277509856677e-05,
      "loss": 0.1835,
      "step": 3545
    },
    {
      "epoch": 0.18021219351236104,
      "grad_norm": 0.538791835308075,
      "learning_rate": 1.879858537658426e-05,
      "loss": 0.185,
      "step": 3550
    },
    {
      "epoch": 0.18046601350322353,
      "grad_norm": 0.6588658094406128,
      "learning_rate": 1.8796893243311844e-05,
      "loss": 0.1787,
      "step": 3555
    },
    {
      "epoch": 0.180719833494086,
      "grad_norm": 0.5752840042114258,
      "learning_rate": 1.8795201110039428e-05,
      "loss": 0.1643,
      "step": 3560
    },
    {
      "epoch": 0.18097365348494848,
      "grad_norm": 1.0449694395065308,
      "learning_rate": 1.879350897676701e-05,
      "loss": 0.167,
      "step": 3565
    },
    {
      "epoch": 0.18122747347581095,
      "grad_norm": 0.5879854559898376,
      "learning_rate": 1.8791816843494595e-05,
      "loss": 0.1879,
      "step": 3570
    },
    {
      "epoch": 0.18148129346667344,
      "grad_norm": 0.782319962978363,
      "learning_rate": 1.879012471022218e-05,
      "loss": 0.1702,
      "step": 3575
    },
    {
      "epoch": 0.1817351134575359,
      "grad_norm": 0.6967921853065491,
      "learning_rate": 1.8788432576949762e-05,
      "loss": 0.1811,
      "step": 3580
    },
    {
      "epoch": 0.1819889334483984,
      "grad_norm": 0.6189330220222473,
      "learning_rate": 1.8786740443677345e-05,
      "loss": 0.1682,
      "step": 3585
    },
    {
      "epoch": 0.18224275343926088,
      "grad_norm": 1.0096155405044556,
      "learning_rate": 1.878504831040493e-05,
      "loss": 0.1538,
      "step": 3590
    },
    {
      "epoch": 0.18249657343012335,
      "grad_norm": 0.9320985078811646,
      "learning_rate": 1.8783356177132513e-05,
      "loss": 0.1963,
      "step": 3595
    },
    {
      "epoch": 0.18275039342098584,
      "grad_norm": 0.6774333119392395,
      "learning_rate": 1.8781664043860096e-05,
      "loss": 0.1907,
      "step": 3600
    },
    {
      "epoch": 0.1830042134118483,
      "grad_norm": 0.7954360246658325,
      "learning_rate": 1.877997191058768e-05,
      "loss": 0.2024,
      "step": 3605
    },
    {
      "epoch": 0.1832580334027108,
      "grad_norm": 0.5827684998512268,
      "learning_rate": 1.8778279777315263e-05,
      "loss": 0.1882,
      "step": 3610
    },
    {
      "epoch": 0.1835118533935733,
      "grad_norm": 0.9476773738861084,
      "learning_rate": 1.8776587644042847e-05,
      "loss": 0.2031,
      "step": 3615
    },
    {
      "epoch": 0.18376567338443575,
      "grad_norm": 0.5946481227874756,
      "learning_rate": 1.877489551077043e-05,
      "loss": 0.1796,
      "step": 3620
    },
    {
      "epoch": 0.18401949337529824,
      "grad_norm": 0.9748368859291077,
      "learning_rate": 1.8773203377498014e-05,
      "loss": 0.1859,
      "step": 3625
    },
    {
      "epoch": 0.1842733133661607,
      "grad_norm": 0.680027425289154,
      "learning_rate": 1.8771511244225597e-05,
      "loss": 0.1949,
      "step": 3630
    },
    {
      "epoch": 0.1845271333570232,
      "grad_norm": 0.7523426413536072,
      "learning_rate": 1.876981911095318e-05,
      "loss": 0.1911,
      "step": 3635
    },
    {
      "epoch": 0.1847809533478857,
      "grad_norm": 0.6500402092933655,
      "learning_rate": 1.8768126977680764e-05,
      "loss": 0.163,
      "step": 3640
    },
    {
      "epoch": 0.18503477333874815,
      "grad_norm": 0.8817210793495178,
      "learning_rate": 1.8766434844408348e-05,
      "loss": 0.1681,
      "step": 3645
    },
    {
      "epoch": 0.18528859332961065,
      "grad_norm": 0.6788656115531921,
      "learning_rate": 1.876474271113593e-05,
      "loss": 0.1521,
      "step": 3650
    },
    {
      "epoch": 0.1855424133204731,
      "grad_norm": 0.5651068687438965,
      "learning_rate": 1.8763050577863512e-05,
      "loss": 0.167,
      "step": 3655
    },
    {
      "epoch": 0.1857962333113356,
      "grad_norm": 0.6204758286476135,
      "learning_rate": 1.87613584445911e-05,
      "loss": 0.1718,
      "step": 3660
    },
    {
      "epoch": 0.1860500533021981,
      "grad_norm": 0.6937487721443176,
      "learning_rate": 1.8759666311318682e-05,
      "loss": 0.2,
      "step": 3665
    },
    {
      "epoch": 0.18630387329306056,
      "grad_norm": 0.791344165802002,
      "learning_rate": 1.8757974178046266e-05,
      "loss": 0.1746,
      "step": 3670
    },
    {
      "epoch": 0.18655769328392305,
      "grad_norm": 0.7099347114562988,
      "learning_rate": 1.875628204477385e-05,
      "loss": 0.1779,
      "step": 3675
    },
    {
      "epoch": 0.1868115132747855,
      "grad_norm": 0.6113842129707336,
      "learning_rate": 1.875458991150143e-05,
      "loss": 0.1821,
      "step": 3680
    },
    {
      "epoch": 0.187065333265648,
      "grad_norm": 0.6101746559143066,
      "learning_rate": 1.8752897778229016e-05,
      "loss": 0.1616,
      "step": 3685
    },
    {
      "epoch": 0.1873191532565105,
      "grad_norm": 0.7758021354675293,
      "learning_rate": 1.87512056449566e-05,
      "loss": 0.1774,
      "step": 3690
    },
    {
      "epoch": 0.18757297324737296,
      "grad_norm": 0.632546603679657,
      "learning_rate": 1.874951351168418e-05,
      "loss": 0.1653,
      "step": 3695
    },
    {
      "epoch": 0.18782679323823545,
      "grad_norm": 1.211270809173584,
      "learning_rate": 1.8747821378411767e-05,
      "loss": 0.1907,
      "step": 3700
    },
    {
      "epoch": 0.18808061322909791,
      "grad_norm": 0.8178271055221558,
      "learning_rate": 1.8746129245139347e-05,
      "loss": 0.1833,
      "step": 3705
    },
    {
      "epoch": 0.1883344332199604,
      "grad_norm": 0.720481276512146,
      "learning_rate": 1.8744437111866934e-05,
      "loss": 0.1865,
      "step": 3710
    },
    {
      "epoch": 0.1885882532108229,
      "grad_norm": 0.6861073970794678,
      "learning_rate": 1.8742744978594514e-05,
      "loss": 0.159,
      "step": 3715
    },
    {
      "epoch": 0.18884207320168536,
      "grad_norm": 0.8942195773124695,
      "learning_rate": 1.8741052845322098e-05,
      "loss": 0.17,
      "step": 3720
    },
    {
      "epoch": 0.18909589319254785,
      "grad_norm": 0.6733516454696655,
      "learning_rate": 1.8739360712049685e-05,
      "loss": 0.2026,
      "step": 3725
    },
    {
      "epoch": 0.18934971318341032,
      "grad_norm": 0.7968901991844177,
      "learning_rate": 1.8737668578777265e-05,
      "loss": 0.1768,
      "step": 3730
    },
    {
      "epoch": 0.1896035331742728,
      "grad_norm": 0.7362722158432007,
      "learning_rate": 1.873597644550485e-05,
      "loss": 0.1793,
      "step": 3735
    },
    {
      "epoch": 0.18985735316513527,
      "grad_norm": 0.5821983814239502,
      "learning_rate": 1.8734284312232432e-05,
      "loss": 0.1935,
      "step": 3740
    },
    {
      "epoch": 0.19011117315599776,
      "grad_norm": 0.6550444960594177,
      "learning_rate": 1.8732592178960016e-05,
      "loss": 0.1815,
      "step": 3745
    },
    {
      "epoch": 0.19036499314686026,
      "grad_norm": 1.057003140449524,
      "learning_rate": 1.87309000456876e-05,
      "loss": 0.1735,
      "step": 3750
    },
    {
      "epoch": 0.19061881313772272,
      "grad_norm": 0.6318512558937073,
      "learning_rate": 1.8729207912415183e-05,
      "loss": 0.1894,
      "step": 3755
    },
    {
      "epoch": 0.1908726331285852,
      "grad_norm": 0.514367401599884,
      "learning_rate": 1.8727515779142766e-05,
      "loss": 0.162,
      "step": 3760
    },
    {
      "epoch": 0.19112645311944768,
      "grad_norm": 0.5353686809539795,
      "learning_rate": 1.872582364587035e-05,
      "loss": 0.1942,
      "step": 3765
    },
    {
      "epoch": 0.19138027311031017,
      "grad_norm": 0.7195748090744019,
      "learning_rate": 1.8724131512597933e-05,
      "loss": 0.1638,
      "step": 3770
    },
    {
      "epoch": 0.19163409310117266,
      "grad_norm": 0.5870863199234009,
      "learning_rate": 1.8722439379325517e-05,
      "loss": 0.1616,
      "step": 3775
    },
    {
      "epoch": 0.19188791309203512,
      "grad_norm": 0.5601217150688171,
      "learning_rate": 1.87207472460531e-05,
      "loss": 0.175,
      "step": 3780
    },
    {
      "epoch": 0.19214173308289761,
      "grad_norm": 0.6515225768089294,
      "learning_rate": 1.8719055112780684e-05,
      "loss": 0.196,
      "step": 3785
    },
    {
      "epoch": 0.19239555307376008,
      "grad_norm": 0.5685335397720337,
      "learning_rate": 1.8717362979508267e-05,
      "loss": 0.173,
      "step": 3790
    },
    {
      "epoch": 0.19264937306462257,
      "grad_norm": 0.5485351085662842,
      "learning_rate": 1.871567084623585e-05,
      "loss": 0.1613,
      "step": 3795
    },
    {
      "epoch": 0.19290319305548506,
      "grad_norm": 0.5211474299430847,
      "learning_rate": 1.8713978712963435e-05,
      "loss": 0.1811,
      "step": 3800
    },
    {
      "epoch": 0.19315701304634753,
      "grad_norm": 0.5695779323577881,
      "learning_rate": 1.8712286579691018e-05,
      "loss": 0.1737,
      "step": 3805
    },
    {
      "epoch": 0.19341083303721002,
      "grad_norm": 0.781248152256012,
      "learning_rate": 1.87105944464186e-05,
      "loss": 0.1474,
      "step": 3810
    },
    {
      "epoch": 0.19366465302807248,
      "grad_norm": 0.6343421936035156,
      "learning_rate": 1.8708902313146185e-05,
      "loss": 0.1597,
      "step": 3815
    },
    {
      "epoch": 0.19391847301893497,
      "grad_norm": 0.46050354838371277,
      "learning_rate": 1.870721017987377e-05,
      "loss": 0.1726,
      "step": 3820
    },
    {
      "epoch": 0.19417229300979746,
      "grad_norm": 0.7671462297439575,
      "learning_rate": 1.8705518046601352e-05,
      "loss": 0.16,
      "step": 3825
    },
    {
      "epoch": 0.19442611300065993,
      "grad_norm": 0.6018497347831726,
      "learning_rate": 1.8703825913328936e-05,
      "loss": 0.17,
      "step": 3830
    },
    {
      "epoch": 0.19467993299152242,
      "grad_norm": 0.5179815292358398,
      "learning_rate": 1.870213378005652e-05,
      "loss": 0.1667,
      "step": 3835
    },
    {
      "epoch": 0.19493375298238488,
      "grad_norm": 0.6752147674560547,
      "learning_rate": 1.8700441646784103e-05,
      "loss": 0.1778,
      "step": 3840
    },
    {
      "epoch": 0.19518757297324738,
      "grad_norm": 0.5998455286026001,
      "learning_rate": 1.8698749513511686e-05,
      "loss": 0.1804,
      "step": 3845
    },
    {
      "epoch": 0.19544139296410987,
      "grad_norm": 0.7329761981964111,
      "learning_rate": 1.869705738023927e-05,
      "loss": 0.1705,
      "step": 3850
    },
    {
      "epoch": 0.19569521295497233,
      "grad_norm": 0.5798784494400024,
      "learning_rate": 1.8695365246966853e-05,
      "loss": 0.1722,
      "step": 3855
    },
    {
      "epoch": 0.19594903294583482,
      "grad_norm": 0.55069899559021,
      "learning_rate": 1.8693673113694437e-05,
      "loss": 0.1677,
      "step": 3860
    },
    {
      "epoch": 0.1962028529366973,
      "grad_norm": 0.7288419008255005,
      "learning_rate": 1.869198098042202e-05,
      "loss": 0.1609,
      "step": 3865
    },
    {
      "epoch": 0.19645667292755978,
      "grad_norm": 0.5638713836669922,
      "learning_rate": 1.8690288847149604e-05,
      "loss": 0.1744,
      "step": 3870
    },
    {
      "epoch": 0.19671049291842224,
      "grad_norm": 1.066811203956604,
      "learning_rate": 1.8688596713877188e-05,
      "loss": 0.1809,
      "step": 3875
    },
    {
      "epoch": 0.19696431290928473,
      "grad_norm": 0.5710257291793823,
      "learning_rate": 1.868690458060477e-05,
      "loss": 0.177,
      "step": 3880
    },
    {
      "epoch": 0.19721813290014722,
      "grad_norm": 1.2099708318710327,
      "learning_rate": 1.868521244733235e-05,
      "loss": 0.1727,
      "step": 3885
    },
    {
      "epoch": 0.1974719528910097,
      "grad_norm": 0.5482766032218933,
      "learning_rate": 1.8683520314059938e-05,
      "loss": 0.1774,
      "step": 3890
    },
    {
      "epoch": 0.19772577288187218,
      "grad_norm": 0.5520328283309937,
      "learning_rate": 1.868182818078752e-05,
      "loss": 0.2038,
      "step": 3895
    },
    {
      "epoch": 0.19797959287273464,
      "grad_norm": 0.7456613183021545,
      "learning_rate": 1.8680136047515102e-05,
      "loss": 0.2103,
      "step": 3900
    },
    {
      "epoch": 0.19823341286359714,
      "grad_norm": 0.6516144275665283,
      "learning_rate": 1.867844391424269e-05,
      "loss": 0.1716,
      "step": 3905
    },
    {
      "epoch": 0.19848723285445963,
      "grad_norm": 0.6367723345756531,
      "learning_rate": 1.867675178097027e-05,
      "loss": 0.1618,
      "step": 3910
    },
    {
      "epoch": 0.1987410528453221,
      "grad_norm": 0.4886556565761566,
      "learning_rate": 1.8675059647697856e-05,
      "loss": 0.1596,
      "step": 3915
    },
    {
      "epoch": 0.19899487283618458,
      "grad_norm": 0.443977952003479,
      "learning_rate": 1.8673367514425436e-05,
      "loss": 0.1847,
      "step": 3920
    },
    {
      "epoch": 0.19924869282704705,
      "grad_norm": 0.5834622383117676,
      "learning_rate": 1.867167538115302e-05,
      "loss": 0.1697,
      "step": 3925
    },
    {
      "epoch": 0.19950251281790954,
      "grad_norm": 0.6995598673820496,
      "learning_rate": 1.8669983247880607e-05,
      "loss": 0.1785,
      "step": 3930
    },
    {
      "epoch": 0.19975633280877203,
      "grad_norm": 0.5793687105178833,
      "learning_rate": 1.8668291114608187e-05,
      "loss": 0.1835,
      "step": 3935
    },
    {
      "epoch": 0.2000101527996345,
      "grad_norm": 0.6711710691452026,
      "learning_rate": 1.866659898133577e-05,
      "loss": 0.1808,
      "step": 3940
    },
    {
      "epoch": 0.20026397279049699,
      "grad_norm": 0.7024595737457275,
      "learning_rate": 1.8664906848063354e-05,
      "loss": 0.1723,
      "step": 3945
    },
    {
      "epoch": 0.20051779278135945,
      "grad_norm": 0.5017092823982239,
      "learning_rate": 1.8663214714790937e-05,
      "loss": 0.1691,
      "step": 3950
    },
    {
      "epoch": 0.20077161277222194,
      "grad_norm": 0.769305944442749,
      "learning_rate": 1.8661522581518524e-05,
      "loss": 0.1612,
      "step": 3955
    },
    {
      "epoch": 0.20102543276308443,
      "grad_norm": 0.6473913192749023,
      "learning_rate": 1.8659830448246105e-05,
      "loss": 0.1788,
      "step": 3960
    },
    {
      "epoch": 0.2012792527539469,
      "grad_norm": 0.5249276757240295,
      "learning_rate": 1.8658138314973688e-05,
      "loss": 0.1654,
      "step": 3965
    },
    {
      "epoch": 0.2015330727448094,
      "grad_norm": 0.9578689932823181,
      "learning_rate": 1.865644618170127e-05,
      "loss": 0.1652,
      "step": 3970
    },
    {
      "epoch": 0.20178689273567185,
      "grad_norm": 0.580237627029419,
      "learning_rate": 1.8654754048428855e-05,
      "loss": 0.1705,
      "step": 3975
    },
    {
      "epoch": 0.20204071272653434,
      "grad_norm": 0.7080004811286926,
      "learning_rate": 1.865306191515644e-05,
      "loss": 0.1527,
      "step": 3980
    },
    {
      "epoch": 0.20229453271739684,
      "grad_norm": 0.4707263112068176,
      "learning_rate": 1.8651369781884022e-05,
      "loss": 0.1525,
      "step": 3985
    },
    {
      "epoch": 0.2025483527082593,
      "grad_norm": 1.2706327438354492,
      "learning_rate": 1.8649677648611606e-05,
      "loss": 0.1963,
      "step": 3990
    },
    {
      "epoch": 0.2028021726991218,
      "grad_norm": 0.7353994250297546,
      "learning_rate": 1.864798551533919e-05,
      "loss": 0.19,
      "step": 3995
    },
    {
      "epoch": 0.20305599268998425,
      "grad_norm": 0.7245195508003235,
      "learning_rate": 1.8646293382066773e-05,
      "loss": 0.1833,
      "step": 4000
    },
    {
      "epoch": 0.20330981268084675,
      "grad_norm": 0.7345285415649414,
      "learning_rate": 1.8644601248794356e-05,
      "loss": 0.175,
      "step": 4005
    },
    {
      "epoch": 0.2035636326717092,
      "grad_norm": 0.8779308795928955,
      "learning_rate": 1.864290911552194e-05,
      "loss": 0.1683,
      "step": 4010
    },
    {
      "epoch": 0.2038174526625717,
      "grad_norm": 0.49476489424705505,
      "learning_rate": 1.8641216982249524e-05,
      "loss": 0.1581,
      "step": 4015
    },
    {
      "epoch": 0.2040712726534342,
      "grad_norm": 0.5535828471183777,
      "learning_rate": 1.8639524848977107e-05,
      "loss": 0.1901,
      "step": 4020
    },
    {
      "epoch": 0.20432509264429666,
      "grad_norm": 0.5144571661949158,
      "learning_rate": 1.863783271570469e-05,
      "loss": 0.1598,
      "step": 4025
    },
    {
      "epoch": 0.20457891263515915,
      "grad_norm": 0.666415274143219,
      "learning_rate": 1.8636140582432274e-05,
      "loss": 0.1713,
      "step": 4030
    },
    {
      "epoch": 0.2048327326260216,
      "grad_norm": 0.5483223795890808,
      "learning_rate": 1.8634448449159858e-05,
      "loss": 0.1685,
      "step": 4035
    },
    {
      "epoch": 0.2050865526168841,
      "grad_norm": 1.4303494691848755,
      "learning_rate": 1.863275631588744e-05,
      "loss": 0.1635,
      "step": 4040
    },
    {
      "epoch": 0.2053403726077466,
      "grad_norm": 0.8545394539833069,
      "learning_rate": 1.8631064182615025e-05,
      "loss": 0.1647,
      "step": 4045
    },
    {
      "epoch": 0.20559419259860906,
      "grad_norm": 1.2137486934661865,
      "learning_rate": 1.862937204934261e-05,
      "loss": 0.1817,
      "step": 4050
    },
    {
      "epoch": 0.20584801258947155,
      "grad_norm": 0.6643612384796143,
      "learning_rate": 1.8627679916070192e-05,
      "loss": 0.1677,
      "step": 4055
    },
    {
      "epoch": 0.20610183258033402,
      "grad_norm": 0.6444915533065796,
      "learning_rate": 1.8625987782797775e-05,
      "loss": 0.1735,
      "step": 4060
    },
    {
      "epoch": 0.2063556525711965,
      "grad_norm": 0.591526985168457,
      "learning_rate": 1.862429564952536e-05,
      "loss": 0.1719,
      "step": 4065
    },
    {
      "epoch": 0.206609472562059,
      "grad_norm": 0.5368006825447083,
      "learning_rate": 1.8622603516252943e-05,
      "loss": 0.1468,
      "step": 4070
    },
    {
      "epoch": 0.20686329255292146,
      "grad_norm": 0.589856743812561,
      "learning_rate": 1.8620911382980523e-05,
      "loss": 0.1637,
      "step": 4075
    },
    {
      "epoch": 0.20711711254378395,
      "grad_norm": 0.707324206829071,
      "learning_rate": 1.861921924970811e-05,
      "loss": 0.1785,
      "step": 4080
    },
    {
      "epoch": 0.20737093253464642,
      "grad_norm": 0.6533371210098267,
      "learning_rate": 1.8617527116435693e-05,
      "loss": 0.1815,
      "step": 4085
    },
    {
      "epoch": 0.2076247525255089,
      "grad_norm": 1.014189600944519,
      "learning_rate": 1.8615834983163273e-05,
      "loss": 0.1684,
      "step": 4090
    },
    {
      "epoch": 0.2078785725163714,
      "grad_norm": 0.4793491065502167,
      "learning_rate": 1.861414284989086e-05,
      "loss": 0.1621,
      "step": 4095
    },
    {
      "epoch": 0.20813239250723387,
      "grad_norm": 0.7456510663032532,
      "learning_rate": 1.861245071661844e-05,
      "loss": 0.1836,
      "step": 4100
    },
    {
      "epoch": 0.20838621249809636,
      "grad_norm": 0.706316351890564,
      "learning_rate": 1.8610758583346027e-05,
      "loss": 0.176,
      "step": 4105
    },
    {
      "epoch": 0.20864003248895882,
      "grad_norm": 1.0256385803222656,
      "learning_rate": 1.860906645007361e-05,
      "loss": 0.1866,
      "step": 4110
    },
    {
      "epoch": 0.2088938524798213,
      "grad_norm": 0.5895639061927795,
      "learning_rate": 1.860737431680119e-05,
      "loss": 0.1542,
      "step": 4115
    },
    {
      "epoch": 0.2091476724706838,
      "grad_norm": 0.6485773324966431,
      "learning_rate": 1.8605682183528778e-05,
      "loss": 0.1585,
      "step": 4120
    },
    {
      "epoch": 0.20940149246154627,
      "grad_norm": 0.5658993124961853,
      "learning_rate": 1.8603990050256358e-05,
      "loss": 0.189,
      "step": 4125
    },
    {
      "epoch": 0.20965531245240876,
      "grad_norm": 0.5852290987968445,
      "learning_rate": 1.860229791698394e-05,
      "loss": 0.164,
      "step": 4130
    },
    {
      "epoch": 0.20990913244327122,
      "grad_norm": 0.5155125856399536,
      "learning_rate": 1.860060578371153e-05,
      "loss": 0.1706,
      "step": 4135
    },
    {
      "epoch": 0.21016295243413372,
      "grad_norm": 0.5006566047668457,
      "learning_rate": 1.859891365043911e-05,
      "loss": 0.158,
      "step": 4140
    },
    {
      "epoch": 0.21041677242499618,
      "grad_norm": 0.8291502594947815,
      "learning_rate": 1.8597221517166692e-05,
      "loss": 0.1732,
      "step": 4145
    },
    {
      "epoch": 0.21067059241585867,
      "grad_norm": 0.6317897439002991,
      "learning_rate": 1.8595529383894276e-05,
      "loss": 0.1629,
      "step": 4150
    },
    {
      "epoch": 0.21092441240672116,
      "grad_norm": 0.8886175155639648,
      "learning_rate": 1.859383725062186e-05,
      "loss": 0.162,
      "step": 4155
    },
    {
      "epoch": 0.21117823239758363,
      "grad_norm": 1.5014044046401978,
      "learning_rate": 1.8592145117349446e-05,
      "loss": 0.1696,
      "step": 4160
    },
    {
      "epoch": 0.21143205238844612,
      "grad_norm": 0.6006774306297302,
      "learning_rate": 1.8590452984077026e-05,
      "loss": 0.1706,
      "step": 4165
    },
    {
      "epoch": 0.21168587237930858,
      "grad_norm": 0.5629169940948486,
      "learning_rate": 1.858876085080461e-05,
      "loss": 0.1868,
      "step": 4170
    },
    {
      "epoch": 0.21193969237017107,
      "grad_norm": 0.5821768641471863,
      "learning_rate": 1.8587068717532194e-05,
      "loss": 0.1612,
      "step": 4175
    },
    {
      "epoch": 0.21219351236103356,
      "grad_norm": 0.4641706645488739,
      "learning_rate": 1.8585376584259777e-05,
      "loss": 0.1573,
      "step": 4180
    },
    {
      "epoch": 0.21244733235189603,
      "grad_norm": 0.6541835069656372,
      "learning_rate": 1.858368445098736e-05,
      "loss": 0.1757,
      "step": 4185
    },
    {
      "epoch": 0.21270115234275852,
      "grad_norm": 0.6074985861778259,
      "learning_rate": 1.8581992317714944e-05,
      "loss": 0.1877,
      "step": 4190
    },
    {
      "epoch": 0.21295497233362098,
      "grad_norm": 0.63148033618927,
      "learning_rate": 1.8580300184442528e-05,
      "loss": 0.1815,
      "step": 4195
    },
    {
      "epoch": 0.21320879232448348,
      "grad_norm": 0.7583580017089844,
      "learning_rate": 1.857860805117011e-05,
      "loss": 0.1703,
      "step": 4200
    },
    {
      "epoch": 0.21346261231534597,
      "grad_norm": 0.737652599811554,
      "learning_rate": 1.8576915917897695e-05,
      "loss": 0.1595,
      "step": 4205
    },
    {
      "epoch": 0.21371643230620843,
      "grad_norm": 0.6646580696105957,
      "learning_rate": 1.857522378462528e-05,
      "loss": 0.1653,
      "step": 4210
    },
    {
      "epoch": 0.21397025229707092,
      "grad_norm": 1.0683635473251343,
      "learning_rate": 1.8573531651352862e-05,
      "loss": 0.1656,
      "step": 4215
    },
    {
      "epoch": 0.2142240722879334,
      "grad_norm": 0.6982542276382446,
      "learning_rate": 1.8571839518080445e-05,
      "loss": 0.1861,
      "step": 4220
    },
    {
      "epoch": 0.21447789227879588,
      "grad_norm": 0.6811363697052002,
      "learning_rate": 1.857014738480803e-05,
      "loss": 0.1652,
      "step": 4225
    },
    {
      "epoch": 0.21473171226965837,
      "grad_norm": 0.43406397104263306,
      "learning_rate": 1.8568455251535613e-05,
      "loss": 0.1683,
      "step": 4230
    },
    {
      "epoch": 0.21498553226052083,
      "grad_norm": 0.4632825255393982,
      "learning_rate": 1.8566763118263196e-05,
      "loss": 0.1428,
      "step": 4235
    },
    {
      "epoch": 0.21523935225138333,
      "grad_norm": 0.6871059536933899,
      "learning_rate": 1.856507098499078e-05,
      "loss": 0.1557,
      "step": 4240
    },
    {
      "epoch": 0.2154931722422458,
      "grad_norm": 0.4901556968688965,
      "learning_rate": 1.8563378851718363e-05,
      "loss": 0.1651,
      "step": 4245
    },
    {
      "epoch": 0.21574699223310828,
      "grad_norm": 0.5677388310432434,
      "learning_rate": 1.8561686718445947e-05,
      "loss": 0.1552,
      "step": 4250
    },
    {
      "epoch": 0.21600081222397077,
      "grad_norm": 0.7926852107048035,
      "learning_rate": 1.855999458517353e-05,
      "loss": 0.1605,
      "step": 4255
    },
    {
      "epoch": 0.21625463221483324,
      "grad_norm": 0.69388347864151,
      "learning_rate": 1.8558302451901114e-05,
      "loss": 0.1773,
      "step": 4260
    },
    {
      "epoch": 0.21650845220569573,
      "grad_norm": 0.6975191235542297,
      "learning_rate": 1.8556610318628697e-05,
      "loss": 0.1646,
      "step": 4265
    },
    {
      "epoch": 0.2167622721965582,
      "grad_norm": 0.5928018689155579,
      "learning_rate": 1.855491818535628e-05,
      "loss": 0.1664,
      "step": 4270
    },
    {
      "epoch": 0.21701609218742068,
      "grad_norm": 0.7675602436065674,
      "learning_rate": 1.8553226052083864e-05,
      "loss": 0.1727,
      "step": 4275
    },
    {
      "epoch": 0.21726991217828315,
      "grad_norm": 1.1031657457351685,
      "learning_rate": 1.8551533918811445e-05,
      "loss": 0.1664,
      "step": 4280
    },
    {
      "epoch": 0.21752373216914564,
      "grad_norm": 0.46347787976264954,
      "learning_rate": 1.854984178553903e-05,
      "loss": 0.1781,
      "step": 4285
    },
    {
      "epoch": 0.21777755216000813,
      "grad_norm": 0.8614928126335144,
      "learning_rate": 1.8548149652266615e-05,
      "loss": 0.1721,
      "step": 4290
    },
    {
      "epoch": 0.2180313721508706,
      "grad_norm": 0.49578431248664856,
      "learning_rate": 1.8546457518994195e-05,
      "loss": 0.1724,
      "step": 4295
    },
    {
      "epoch": 0.2182851921417331,
      "grad_norm": 0.6700772643089294,
      "learning_rate": 1.8544765385721782e-05,
      "loss": 0.1681,
      "step": 4300
    },
    {
      "epoch": 0.21853901213259555,
      "grad_norm": 0.6420726180076599,
      "learning_rate": 1.8543073252449362e-05,
      "loss": 0.185,
      "step": 4305
    },
    {
      "epoch": 0.21879283212345804,
      "grad_norm": 0.6904594302177429,
      "learning_rate": 1.854138111917695e-05,
      "loss": 0.1689,
      "step": 4310
    },
    {
      "epoch": 0.21904665211432053,
      "grad_norm": 0.5921376347541809,
      "learning_rate": 1.8539688985904533e-05,
      "loss": 0.1651,
      "step": 4315
    },
    {
      "epoch": 0.219300472105183,
      "grad_norm": 0.7793440818786621,
      "learning_rate": 1.8537996852632113e-05,
      "loss": 0.1718,
      "step": 4320
    },
    {
      "epoch": 0.2195542920960455,
      "grad_norm": 0.42908427119255066,
      "learning_rate": 1.85363047193597e-05,
      "loss": 0.1639,
      "step": 4325
    },
    {
      "epoch": 0.21980811208690795,
      "grad_norm": 0.6529392600059509,
      "learning_rate": 1.853461258608728e-05,
      "loss": 0.1845,
      "step": 4330
    },
    {
      "epoch": 0.22006193207777044,
      "grad_norm": 0.4588109254837036,
      "learning_rate": 1.8532920452814864e-05,
      "loss": 0.1653,
      "step": 4335
    },
    {
      "epoch": 0.22031575206863294,
      "grad_norm": 0.5849223136901855,
      "learning_rate": 1.853122831954245e-05,
      "loss": 0.164,
      "step": 4340
    },
    {
      "epoch": 0.2205695720594954,
      "grad_norm": 0.7719104290008545,
      "learning_rate": 1.852953618627003e-05,
      "loss": 0.1663,
      "step": 4345
    },
    {
      "epoch": 0.2208233920503579,
      "grad_norm": 0.5275380611419678,
      "learning_rate": 1.8527844052997618e-05,
      "loss": 0.1622,
      "step": 4350
    },
    {
      "epoch": 0.22107721204122036,
      "grad_norm": 1.1291331052780151,
      "learning_rate": 1.8526151919725198e-05,
      "loss": 0.1736,
      "step": 4355
    },
    {
      "epoch": 0.22133103203208285,
      "grad_norm": 0.7423467636108398,
      "learning_rate": 1.852445978645278e-05,
      "loss": 0.1594,
      "step": 4360
    },
    {
      "epoch": 0.22158485202294534,
      "grad_norm": 0.6104901432991028,
      "learning_rate": 1.8522767653180368e-05,
      "loss": 0.1613,
      "step": 4365
    },
    {
      "epoch": 0.2218386720138078,
      "grad_norm": 0.6495379209518433,
      "learning_rate": 1.852107551990795e-05,
      "loss": 0.1595,
      "step": 4370
    },
    {
      "epoch": 0.2220924920046703,
      "grad_norm": 0.5626767873764038,
      "learning_rate": 1.8519383386635532e-05,
      "loss": 0.1716,
      "step": 4375
    },
    {
      "epoch": 0.22234631199553276,
      "grad_norm": 0.5380212664604187,
      "learning_rate": 1.8517691253363116e-05,
      "loss": 0.1745,
      "step": 4380
    },
    {
      "epoch": 0.22260013198639525,
      "grad_norm": 0.5656577348709106,
      "learning_rate": 1.85159991200907e-05,
      "loss": 0.1812,
      "step": 4385
    },
    {
      "epoch": 0.22285395197725774,
      "grad_norm": 0.5490018129348755,
      "learning_rate": 1.8514306986818283e-05,
      "loss": 0.1724,
      "step": 4390
    },
    {
      "epoch": 0.2231077719681202,
      "grad_norm": 0.563014805316925,
      "learning_rate": 1.8512614853545866e-05,
      "loss": 0.1547,
      "step": 4395
    },
    {
      "epoch": 0.2233615919589827,
      "grad_norm": 0.6768389344215393,
      "learning_rate": 1.851092272027345e-05,
      "loss": 0.1643,
      "step": 4400
    },
    {
      "epoch": 0.22361541194984516,
      "grad_norm": 0.9782013893127441,
      "learning_rate": 1.8509230587001033e-05,
      "loss": 0.1556,
      "step": 4405
    },
    {
      "epoch": 0.22386923194070765,
      "grad_norm": 0.6855179071426392,
      "learning_rate": 1.8507538453728617e-05,
      "loss": 0.167,
      "step": 4410
    },
    {
      "epoch": 0.22412305193157014,
      "grad_norm": 0.9513364434242249,
      "learning_rate": 1.85058463204562e-05,
      "loss": 0.1615,
      "step": 4415
    },
    {
      "epoch": 0.2243768719224326,
      "grad_norm": 0.5813558101654053,
      "learning_rate": 1.8504154187183784e-05,
      "loss": 0.1509,
      "step": 4420
    },
    {
      "epoch": 0.2246306919132951,
      "grad_norm": 0.8219256401062012,
      "learning_rate": 1.8502462053911367e-05,
      "loss": 0.1616,
      "step": 4425
    },
    {
      "epoch": 0.22488451190415756,
      "grad_norm": 0.4321226477622986,
      "learning_rate": 1.850076992063895e-05,
      "loss": 0.155,
      "step": 4430
    },
    {
      "epoch": 0.22513833189502006,
      "grad_norm": 0.6410679817199707,
      "learning_rate": 1.8499077787366534e-05,
      "loss": 0.1658,
      "step": 4435
    },
    {
      "epoch": 0.22539215188588252,
      "grad_norm": 0.6968072056770325,
      "learning_rate": 1.8497385654094118e-05,
      "loss": 0.1574,
      "step": 4440
    },
    {
      "epoch": 0.225645971876745,
      "grad_norm": 0.8063611388206482,
      "learning_rate": 1.84956935208217e-05,
      "loss": 0.149,
      "step": 4445
    },
    {
      "epoch": 0.2258997918676075,
      "grad_norm": 0.5573265552520752,
      "learning_rate": 1.8494001387549285e-05,
      "loss": 0.1624,
      "step": 4450
    },
    {
      "epoch": 0.22615361185846997,
      "grad_norm": 0.6173406839370728,
      "learning_rate": 1.849230925427687e-05,
      "loss": 0.1539,
      "step": 4455
    },
    {
      "epoch": 0.22640743184933246,
      "grad_norm": 0.46205034852027893,
      "learning_rate": 1.8490617121004452e-05,
      "loss": 0.1546,
      "step": 4460
    },
    {
      "epoch": 0.22666125184019492,
      "grad_norm": 0.5302807688713074,
      "learning_rate": 1.8488924987732036e-05,
      "loss": 0.1712,
      "step": 4465
    },
    {
      "epoch": 0.2269150718310574,
      "grad_norm": 0.6777194738388062,
      "learning_rate": 1.848723285445962e-05,
      "loss": 0.1613,
      "step": 4470
    },
    {
      "epoch": 0.2271688918219199,
      "grad_norm": 0.4836040735244751,
      "learning_rate": 1.8485540721187203e-05,
      "loss": 0.166,
      "step": 4475
    },
    {
      "epoch": 0.22742271181278237,
      "grad_norm": 0.7450656890869141,
      "learning_rate": 1.8483848587914786e-05,
      "loss": 0.1776,
      "step": 4480
    },
    {
      "epoch": 0.22767653180364486,
      "grad_norm": 1.0024539232254028,
      "learning_rate": 1.8482156454642367e-05,
      "loss": 0.1649,
      "step": 4485
    },
    {
      "epoch": 0.22793035179450732,
      "grad_norm": 0.8160313963890076,
      "learning_rate": 1.8480464321369953e-05,
      "loss": 0.1389,
      "step": 4490
    },
    {
      "epoch": 0.22818417178536982,
      "grad_norm": 0.495733380317688,
      "learning_rate": 1.8478772188097537e-05,
      "loss": 0.1434,
      "step": 4495
    },
    {
      "epoch": 0.2284379917762323,
      "grad_norm": 0.6479185819625854,
      "learning_rate": 1.847708005482512e-05,
      "loss": 0.1492,
      "step": 4500
    },
    {
      "epoch": 0.22869181176709477,
      "grad_norm": 0.6237260103225708,
      "learning_rate": 1.8475387921552704e-05,
      "loss": 0.1692,
      "step": 4505
    },
    {
      "epoch": 0.22894563175795726,
      "grad_norm": 0.5665149688720703,
      "learning_rate": 1.8473695788280284e-05,
      "loss": 0.1783,
      "step": 4510
    },
    {
      "epoch": 0.22919945174881973,
      "grad_norm": 0.7720448970794678,
      "learning_rate": 1.847200365500787e-05,
      "loss": 0.1766,
      "step": 4515
    },
    {
      "epoch": 0.22945327173968222,
      "grad_norm": 0.6933048963546753,
      "learning_rate": 1.8470311521735455e-05,
      "loss": 0.1655,
      "step": 4520
    },
    {
      "epoch": 0.2297070917305447,
      "grad_norm": 0.7732922434806824,
      "learning_rate": 1.8468619388463035e-05,
      "loss": 0.1609,
      "step": 4525
    },
    {
      "epoch": 0.22996091172140717,
      "grad_norm": 0.9285184741020203,
      "learning_rate": 1.8466927255190622e-05,
      "loss": 0.1476,
      "step": 4530
    },
    {
      "epoch": 0.23021473171226967,
      "grad_norm": 0.5524913668632507,
      "learning_rate": 1.8465235121918202e-05,
      "loss": 0.1541,
      "step": 4535
    },
    {
      "epoch": 0.23046855170313213,
      "grad_norm": 0.5339060425758362,
      "learning_rate": 1.8463542988645786e-05,
      "loss": 0.1802,
      "step": 4540
    },
    {
      "epoch": 0.23072237169399462,
      "grad_norm": 0.5357369184494019,
      "learning_rate": 1.8461850855373372e-05,
      "loss": 0.162,
      "step": 4545
    },
    {
      "epoch": 0.2309761916848571,
      "grad_norm": 0.6212167739868164,
      "learning_rate": 1.8460158722100953e-05,
      "loss": 0.1612,
      "step": 4550
    },
    {
      "epoch": 0.23123001167571958,
      "grad_norm": 0.5287737250328064,
      "learning_rate": 1.845846658882854e-05,
      "loss": 0.1445,
      "step": 4555
    },
    {
      "epoch": 0.23148383166658207,
      "grad_norm": 0.5808826088905334,
      "learning_rate": 1.845677445555612e-05,
      "loss": 0.1443,
      "step": 4560
    },
    {
      "epoch": 0.23173765165744453,
      "grad_norm": 0.47897592186927795,
      "learning_rate": 1.8455082322283703e-05,
      "loss": 0.1503,
      "step": 4565
    },
    {
      "epoch": 0.23199147164830702,
      "grad_norm": 0.7037847638130188,
      "learning_rate": 1.845339018901129e-05,
      "loss": 0.1676,
      "step": 4570
    },
    {
      "epoch": 0.2322452916391695,
      "grad_norm": 0.4233403503894806,
      "learning_rate": 1.845169805573887e-05,
      "loss": 0.1459,
      "step": 4575
    },
    {
      "epoch": 0.23249911163003198,
      "grad_norm": 1.0100395679473877,
      "learning_rate": 1.8450005922466454e-05,
      "loss": 0.1742,
      "step": 4580
    },
    {
      "epoch": 0.23275293162089447,
      "grad_norm": 0.6140120029449463,
      "learning_rate": 1.8448313789194037e-05,
      "loss": 0.1439,
      "step": 4585
    },
    {
      "epoch": 0.23300675161175693,
      "grad_norm": 0.5431662201881409,
      "learning_rate": 1.844662165592162e-05,
      "loss": 0.1444,
      "step": 4590
    },
    {
      "epoch": 0.23326057160261943,
      "grad_norm": 0.7064282894134521,
      "learning_rate": 1.8444929522649208e-05,
      "loss": 0.1598,
      "step": 4595
    },
    {
      "epoch": 0.2335143915934819,
      "grad_norm": 0.5530514717102051,
      "learning_rate": 1.8443237389376788e-05,
      "loss": 0.1514,
      "step": 4600
    },
    {
      "epoch": 0.23376821158434438,
      "grad_norm": 0.5173631906509399,
      "learning_rate": 1.844154525610437e-05,
      "loss": 0.1581,
      "step": 4605
    },
    {
      "epoch": 0.23402203157520687,
      "grad_norm": 0.5507360100746155,
      "learning_rate": 1.8439853122831955e-05,
      "loss": 0.1555,
      "step": 4610
    },
    {
      "epoch": 0.23427585156606934,
      "grad_norm": 0.6322073340415955,
      "learning_rate": 1.843816098955954e-05,
      "loss": 0.1734,
      "step": 4615
    },
    {
      "epoch": 0.23452967155693183,
      "grad_norm": 0.6972060799598694,
      "learning_rate": 1.8436468856287122e-05,
      "loss": 0.1554,
      "step": 4620
    },
    {
      "epoch": 0.2347834915477943,
      "grad_norm": 0.4970358610153198,
      "learning_rate": 1.8434776723014706e-05,
      "loss": 0.1679,
      "step": 4625
    },
    {
      "epoch": 0.23503731153865678,
      "grad_norm": 0.5872799158096313,
      "learning_rate": 1.843308458974229e-05,
      "loss": 0.1489,
      "step": 4630
    },
    {
      "epoch": 0.23529113152951928,
      "grad_norm": 1.1778312921524048,
      "learning_rate": 1.8431392456469873e-05,
      "loss": 0.1502,
      "step": 4635
    },
    {
      "epoch": 0.23554495152038174,
      "grad_norm": 0.5631628036499023,
      "learning_rate": 1.8429700323197456e-05,
      "loss": 0.1709,
      "step": 4640
    },
    {
      "epoch": 0.23579877151124423,
      "grad_norm": 0.7208503484725952,
      "learning_rate": 1.842800818992504e-05,
      "loss": 0.1591,
      "step": 4645
    },
    {
      "epoch": 0.2360525915021067,
      "grad_norm": 0.5835586786270142,
      "learning_rate": 1.8426316056652624e-05,
      "loss": 0.16,
      "step": 4650
    },
    {
      "epoch": 0.2363064114929692,
      "grad_norm": 0.5638494491577148,
      "learning_rate": 1.8424623923380207e-05,
      "loss": 0.1534,
      "step": 4655
    },
    {
      "epoch": 0.23656023148383168,
      "grad_norm": 0.5742624998092651,
      "learning_rate": 1.842293179010779e-05,
      "loss": 0.1587,
      "step": 4660
    },
    {
      "epoch": 0.23681405147469414,
      "grad_norm": 1.1678133010864258,
      "learning_rate": 1.8421239656835374e-05,
      "loss": 0.1577,
      "step": 4665
    },
    {
      "epoch": 0.23706787146555663,
      "grad_norm": 0.5317492485046387,
      "learning_rate": 1.8419547523562958e-05,
      "loss": 0.1635,
      "step": 4670
    },
    {
      "epoch": 0.2373216914564191,
      "grad_norm": 0.5751121640205383,
      "learning_rate": 1.841785539029054e-05,
      "loss": 0.1619,
      "step": 4675
    },
    {
      "epoch": 0.2375755114472816,
      "grad_norm": 0.5913323163986206,
      "learning_rate": 1.8416163257018125e-05,
      "loss": 0.1438,
      "step": 4680
    },
    {
      "epoch": 0.23782933143814408,
      "grad_norm": 0.43694448471069336,
      "learning_rate": 1.841447112374571e-05,
      "loss": 0.1558,
      "step": 4685
    },
    {
      "epoch": 0.23808315142900655,
      "grad_norm": 0.580193042755127,
      "learning_rate": 1.841277899047329e-05,
      "loss": 0.1544,
      "step": 4690
    },
    {
      "epoch": 0.23833697141986904,
      "grad_norm": 0.5924519896507263,
      "learning_rate": 1.8411086857200875e-05,
      "loss": 0.1766,
      "step": 4695
    },
    {
      "epoch": 0.2385907914107315,
      "grad_norm": 0.5101874470710754,
      "learning_rate": 1.840939472392846e-05,
      "loss": 0.149,
      "step": 4700
    },
    {
      "epoch": 0.238844611401594,
      "grad_norm": 0.5736780762672424,
      "learning_rate": 1.8407702590656043e-05,
      "loss": 0.1663,
      "step": 4705
    },
    {
      "epoch": 0.23909843139245646,
      "grad_norm": 0.4370191693305969,
      "learning_rate": 1.8406010457383626e-05,
      "loss": 0.1512,
      "step": 4710
    },
    {
      "epoch": 0.23935225138331895,
      "grad_norm": 0.545307993888855,
      "learning_rate": 1.8404318324111206e-05,
      "loss": 0.1537,
      "step": 4715
    },
    {
      "epoch": 0.23960607137418144,
      "grad_norm": 0.680793285369873,
      "learning_rate": 1.8402626190838793e-05,
      "loss": 0.1798,
      "step": 4720
    },
    {
      "epoch": 0.2398598913650439,
      "grad_norm": 0.7056201100349426,
      "learning_rate": 1.8400934057566377e-05,
      "loss": 0.1549,
      "step": 4725
    },
    {
      "epoch": 0.2401137113559064,
      "grad_norm": 0.4889606535434723,
      "learning_rate": 1.8399241924293957e-05,
      "loss": 0.1701,
      "step": 4730
    },
    {
      "epoch": 0.24036753134676886,
      "grad_norm": 0.5846245884895325,
      "learning_rate": 1.8397549791021544e-05,
      "loss": 0.1456,
      "step": 4735
    },
    {
      "epoch": 0.24062135133763135,
      "grad_norm": 0.6689904928207397,
      "learning_rate": 1.8395857657749124e-05,
      "loss": 0.1648,
      "step": 4740
    },
    {
      "epoch": 0.24087517132849384,
      "grad_norm": 0.5623743534088135,
      "learning_rate": 1.839416552447671e-05,
      "loss": 0.1538,
      "step": 4745
    },
    {
      "epoch": 0.2411289913193563,
      "grad_norm": 0.8147268295288086,
      "learning_rate": 1.8392473391204294e-05,
      "loss": 0.1594,
      "step": 4750
    },
    {
      "epoch": 0.2413828113102188,
      "grad_norm": 0.49249938130378723,
      "learning_rate": 1.8390781257931875e-05,
      "loss": 0.1728,
      "step": 4755
    },
    {
      "epoch": 0.24163663130108126,
      "grad_norm": 0.4992578327655792,
      "learning_rate": 1.838908912465946e-05,
      "loss": 0.1593,
      "step": 4760
    },
    {
      "epoch": 0.24189045129194375,
      "grad_norm": 0.5691167712211609,
      "learning_rate": 1.838739699138704e-05,
      "loss": 0.1537,
      "step": 4765
    },
    {
      "epoch": 0.24214427128280624,
      "grad_norm": 0.5181019902229309,
      "learning_rate": 1.8385704858114625e-05,
      "loss": 0.148,
      "step": 4770
    },
    {
      "epoch": 0.2423980912736687,
      "grad_norm": 0.4930359423160553,
      "learning_rate": 1.8384012724842212e-05,
      "loss": 0.1518,
      "step": 4775
    },
    {
      "epoch": 0.2426519112645312,
      "grad_norm": 0.6070294380187988,
      "learning_rate": 1.8382320591569792e-05,
      "loss": 0.1812,
      "step": 4780
    },
    {
      "epoch": 0.24290573125539366,
      "grad_norm": 0.6835089921951294,
      "learning_rate": 1.8380628458297376e-05,
      "loss": 0.1767,
      "step": 4785
    },
    {
      "epoch": 0.24315955124625616,
      "grad_norm": 0.7003611326217651,
      "learning_rate": 1.837893632502496e-05,
      "loss": 0.1476,
      "step": 4790
    },
    {
      "epoch": 0.24341337123711865,
      "grad_norm": 0.4410546123981476,
      "learning_rate": 1.8377244191752543e-05,
      "loss": 0.1493,
      "step": 4795
    },
    {
      "epoch": 0.2436671912279811,
      "grad_norm": 0.6904603242874146,
      "learning_rate": 1.837555205848013e-05,
      "loss": 0.1473,
      "step": 4800
    },
    {
      "epoch": 0.2439210112188436,
      "grad_norm": 0.5898627638816833,
      "learning_rate": 1.837385992520771e-05,
      "loss": 0.1576,
      "step": 4805
    },
    {
      "epoch": 0.24417483120970607,
      "grad_norm": 0.5015605092048645,
      "learning_rate": 1.8372167791935294e-05,
      "loss": 0.1438,
      "step": 4810
    },
    {
      "epoch": 0.24442865120056856,
      "grad_norm": 0.7323523163795471,
      "learning_rate": 1.8370475658662877e-05,
      "loss": 0.1609,
      "step": 4815
    },
    {
      "epoch": 0.24468247119143105,
      "grad_norm": 0.5204626321792603,
      "learning_rate": 1.836878352539046e-05,
      "loss": 0.169,
      "step": 4820
    },
    {
      "epoch": 0.24493629118229351,
      "grad_norm": 0.5830612182617188,
      "learning_rate": 1.8367091392118044e-05,
      "loss": 0.1387,
      "step": 4825
    },
    {
      "epoch": 0.245190111173156,
      "grad_norm": 0.45259636640548706,
      "learning_rate": 1.8365399258845628e-05,
      "loss": 0.1599,
      "step": 4830
    },
    {
      "epoch": 0.24544393116401847,
      "grad_norm": 0.6329886317253113,
      "learning_rate": 1.836370712557321e-05,
      "loss": 0.1701,
      "step": 4835
    },
    {
      "epoch": 0.24569775115488096,
      "grad_norm": 0.4662386476993561,
      "learning_rate": 1.8362014992300795e-05,
      "loss": 0.1307,
      "step": 4840
    },
    {
      "epoch": 0.24595157114574343,
      "grad_norm": 0.5723447799682617,
      "learning_rate": 1.836032285902838e-05,
      "loss": 0.1635,
      "step": 4845
    },
    {
      "epoch": 0.24620539113660592,
      "grad_norm": 0.9412028789520264,
      "learning_rate": 1.8358630725755962e-05,
      "loss": 0.1704,
      "step": 4850
    },
    {
      "epoch": 0.2464592111274684,
      "grad_norm": 0.6017778515815735,
      "learning_rate": 1.8356938592483545e-05,
      "loss": 0.1658,
      "step": 4855
    },
    {
      "epoch": 0.24671303111833087,
      "grad_norm": 0.45639511942863464,
      "learning_rate": 1.835524645921113e-05,
      "loss": 0.1519,
      "step": 4860
    },
    {
      "epoch": 0.24696685110919336,
      "grad_norm": 0.5620295405387878,
      "learning_rate": 1.8353554325938713e-05,
      "loss": 0.1727,
      "step": 4865
    },
    {
      "epoch": 0.24722067110005583,
      "grad_norm": 0.5075967907905579,
      "learning_rate": 1.8351862192666296e-05,
      "loss": 0.1656,
      "step": 4870
    },
    {
      "epoch": 0.24747449109091832,
      "grad_norm": 0.7711220383644104,
      "learning_rate": 1.835017005939388e-05,
      "loss": 0.1705,
      "step": 4875
    },
    {
      "epoch": 0.2477283110817808,
      "grad_norm": 0.5555239319801331,
      "learning_rate": 1.8348477926121463e-05,
      "loss": 0.1239,
      "step": 4880
    },
    {
      "epoch": 0.24798213107264327,
      "grad_norm": 0.4874543845653534,
      "learning_rate": 1.8346785792849047e-05,
      "loss": 0.1596,
      "step": 4885
    },
    {
      "epoch": 0.24823595106350577,
      "grad_norm": 0.49628138542175293,
      "learning_rate": 1.834509365957663e-05,
      "loss": 0.142,
      "step": 4890
    },
    {
      "epoch": 0.24848977105436823,
      "grad_norm": 0.6575145125389099,
      "learning_rate": 1.8343401526304214e-05,
      "loss": 0.1602,
      "step": 4895
    },
    {
      "epoch": 0.24874359104523072,
      "grad_norm": 0.4418342709541321,
      "learning_rate": 1.8341709393031797e-05,
      "loss": 0.1574,
      "step": 4900
    },
    {
      "epoch": 0.2489974110360932,
      "grad_norm": 0.7797775864601135,
      "learning_rate": 1.834001725975938e-05,
      "loss": 0.1718,
      "step": 4905
    },
    {
      "epoch": 0.24925123102695568,
      "grad_norm": 0.87046879529953,
      "learning_rate": 1.8338325126486964e-05,
      "loss": 0.1667,
      "step": 4910
    },
    {
      "epoch": 0.24950505101781817,
      "grad_norm": 0.47170791029930115,
      "learning_rate": 1.8336632993214548e-05,
      "loss": 0.1376,
      "step": 4915
    },
    {
      "epoch": 0.24975887100868063,
      "grad_norm": 0.6200122833251953,
      "learning_rate": 1.8334940859942128e-05,
      "loss": 0.1674,
      "step": 4920
    },
    {
      "epoch": 0.2500126909995431,
      "grad_norm": 0.7150362730026245,
      "learning_rate": 1.8333248726669715e-05,
      "loss": 0.1697,
      "step": 4925
    },
    {
      "epoch": 0.2502665109904056,
      "grad_norm": 0.9218109846115112,
      "learning_rate": 1.83315565933973e-05,
      "loss": 0.1693,
      "step": 4930
    },
    {
      "epoch": 0.2505203309812681,
      "grad_norm": 0.5581763386726379,
      "learning_rate": 1.832986446012488e-05,
      "loss": 0.1584,
      "step": 4935
    },
    {
      "epoch": 0.25077415097213057,
      "grad_norm": 0.610871434211731,
      "learning_rate": 1.8328172326852466e-05,
      "loss": 0.1564,
      "step": 4940
    },
    {
      "epoch": 0.25102797096299306,
      "grad_norm": 0.5767541527748108,
      "learning_rate": 1.8326480193580046e-05,
      "loss": 0.1501,
      "step": 4945
    },
    {
      "epoch": 0.2512817909538555,
      "grad_norm": 0.5301868319511414,
      "learning_rate": 1.8324788060307633e-05,
      "loss": 0.1444,
      "step": 4950
    },
    {
      "epoch": 0.251535610944718,
      "grad_norm": 0.4195795953273773,
      "learning_rate": 1.8323095927035216e-05,
      "loss": 0.1529,
      "step": 4955
    },
    {
      "epoch": 0.2517894309355805,
      "grad_norm": 0.6302689909934998,
      "learning_rate": 1.8321403793762797e-05,
      "loss": 0.1554,
      "step": 4960
    },
    {
      "epoch": 0.252043250926443,
      "grad_norm": 0.8117844462394714,
      "learning_rate": 1.8319711660490383e-05,
      "loss": 0.1598,
      "step": 4965
    },
    {
      "epoch": 0.25229707091730547,
      "grad_norm": 0.6213078498840332,
      "learning_rate": 1.8318019527217964e-05,
      "loss": 0.1598,
      "step": 4970
    },
    {
      "epoch": 0.2525508909081679,
      "grad_norm": 0.5293858647346497,
      "learning_rate": 1.8316327393945547e-05,
      "loss": 0.144,
      "step": 4975
    },
    {
      "epoch": 0.2528047108990304,
      "grad_norm": 0.5613870620727539,
      "learning_rate": 1.8314635260673134e-05,
      "loss": 0.1496,
      "step": 4980
    },
    {
      "epoch": 0.2530585308898929,
      "grad_norm": 0.9666538834571838,
      "learning_rate": 1.8312943127400714e-05,
      "loss": 0.1718,
      "step": 4985
    },
    {
      "epoch": 0.2533123508807554,
      "grad_norm": 0.6101076006889343,
      "learning_rate": 1.83112509941283e-05,
      "loss": 0.1444,
      "step": 4990
    },
    {
      "epoch": 0.25356617087161787,
      "grad_norm": 0.5413176417350769,
      "learning_rate": 1.830955886085588e-05,
      "loss": 0.1378,
      "step": 4995
    },
    {
      "epoch": 0.2538199908624803,
      "grad_norm": 0.731587827205658,
      "learning_rate": 1.8307866727583465e-05,
      "loss": 0.1523,
      "step": 5000
    },
    {
      "epoch": 0.2540738108533428,
      "grad_norm": 0.6076200008392334,
      "learning_rate": 1.8306174594311052e-05,
      "loss": 0.1298,
      "step": 5005
    },
    {
      "epoch": 0.2543276308442053,
      "grad_norm": 0.4898158013820648,
      "learning_rate": 1.8304482461038632e-05,
      "loss": 0.1509,
      "step": 5010
    },
    {
      "epoch": 0.2545814508350678,
      "grad_norm": 0.6534063816070557,
      "learning_rate": 1.8302790327766215e-05,
      "loss": 0.1562,
      "step": 5015
    },
    {
      "epoch": 0.25483527082593027,
      "grad_norm": 0.8010299801826477,
      "learning_rate": 1.83010981944938e-05,
      "loss": 0.1422,
      "step": 5020
    },
    {
      "epoch": 0.2550890908167927,
      "grad_norm": 0.6017109751701355,
      "learning_rate": 1.8299406061221383e-05,
      "loss": 0.1592,
      "step": 5025
    },
    {
      "epoch": 0.2553429108076552,
      "grad_norm": 0.579904317855835,
      "learning_rate": 1.8297713927948966e-05,
      "loss": 0.1567,
      "step": 5030
    },
    {
      "epoch": 0.2555967307985177,
      "grad_norm": 0.94927978515625,
      "learning_rate": 1.829602179467655e-05,
      "loss": 0.1479,
      "step": 5035
    },
    {
      "epoch": 0.2558505507893802,
      "grad_norm": 0.5884600877761841,
      "learning_rate": 1.8294329661404133e-05,
      "loss": 0.1592,
      "step": 5040
    },
    {
      "epoch": 0.2561043707802427,
      "grad_norm": 0.7687221169471741,
      "learning_rate": 1.8292637528131717e-05,
      "loss": 0.1514,
      "step": 5045
    },
    {
      "epoch": 0.2563581907711051,
      "grad_norm": 0.7298230528831482,
      "learning_rate": 1.82909453948593e-05,
      "loss": 0.1746,
      "step": 5050
    },
    {
      "epoch": 0.2566120107619676,
      "grad_norm": 0.8364652395248413,
      "learning_rate": 1.8289253261586884e-05,
      "loss": 0.145,
      "step": 5055
    },
    {
      "epoch": 0.2568658307528301,
      "grad_norm": 0.799123227596283,
      "learning_rate": 1.8287561128314467e-05,
      "loss": 0.1438,
      "step": 5060
    },
    {
      "epoch": 0.2571196507436926,
      "grad_norm": 0.7807921767234802,
      "learning_rate": 1.828586899504205e-05,
      "loss": 0.1375,
      "step": 5065
    },
    {
      "epoch": 0.2573734707345551,
      "grad_norm": 0.4554001986980438,
      "learning_rate": 1.8284176861769634e-05,
      "loss": 0.1586,
      "step": 5070
    },
    {
      "epoch": 0.2576272907254175,
      "grad_norm": 0.5187448263168335,
      "learning_rate": 1.8282484728497218e-05,
      "loss": 0.1526,
      "step": 5075
    },
    {
      "epoch": 0.25788111071628,
      "grad_norm": 0.5898265242576599,
      "learning_rate": 1.82807925952248e-05,
      "loss": 0.1523,
      "step": 5080
    },
    {
      "epoch": 0.2581349307071425,
      "grad_norm": 0.4496493637561798,
      "learning_rate": 1.8279100461952385e-05,
      "loss": 0.1583,
      "step": 5085
    },
    {
      "epoch": 0.258388750698005,
      "grad_norm": 0.5907924771308899,
      "learning_rate": 1.827740832867997e-05,
      "loss": 0.1697,
      "step": 5090
    },
    {
      "epoch": 0.2586425706888675,
      "grad_norm": 0.5844322443008423,
      "learning_rate": 1.8275716195407552e-05,
      "loss": 0.1562,
      "step": 5095
    },
    {
      "epoch": 0.2588963906797299,
      "grad_norm": 0.5347046256065369,
      "learning_rate": 1.8274024062135136e-05,
      "loss": 0.1396,
      "step": 5100
    },
    {
      "epoch": 0.2591502106705924,
      "grad_norm": 0.6127249598503113,
      "learning_rate": 1.827233192886272e-05,
      "loss": 0.1442,
      "step": 5105
    },
    {
      "epoch": 0.2594040306614549,
      "grad_norm": 0.5309034585952759,
      "learning_rate": 1.8270639795590303e-05,
      "loss": 0.148,
      "step": 5110
    },
    {
      "epoch": 0.2596578506523174,
      "grad_norm": 0.6933298707008362,
      "learning_rate": 1.8268947662317886e-05,
      "loss": 0.1723,
      "step": 5115
    },
    {
      "epoch": 0.2599116706431799,
      "grad_norm": 0.6623631119728088,
      "learning_rate": 1.826725552904547e-05,
      "loss": 0.164,
      "step": 5120
    },
    {
      "epoch": 0.2601654906340423,
      "grad_norm": 0.7233380079269409,
      "learning_rate": 1.826556339577305e-05,
      "loss": 0.1501,
      "step": 5125
    },
    {
      "epoch": 0.2604193106249048,
      "grad_norm": 0.6823766231536865,
      "learning_rate": 1.8263871262500637e-05,
      "loss": 0.1508,
      "step": 5130
    },
    {
      "epoch": 0.2606731306157673,
      "grad_norm": 0.5988183617591858,
      "learning_rate": 1.826217912922822e-05,
      "loss": 0.1495,
      "step": 5135
    },
    {
      "epoch": 0.2609269506066298,
      "grad_norm": 0.48616233468055725,
      "learning_rate": 1.8260486995955804e-05,
      "loss": 0.1423,
      "step": 5140
    },
    {
      "epoch": 0.2611807705974923,
      "grad_norm": 0.4381769001483917,
      "learning_rate": 1.8258794862683388e-05,
      "loss": 0.1396,
      "step": 5145
    },
    {
      "epoch": 0.2614345905883547,
      "grad_norm": 0.4858790934085846,
      "learning_rate": 1.8257102729410968e-05,
      "loss": 0.1503,
      "step": 5150
    },
    {
      "epoch": 0.2616884105792172,
      "grad_norm": 0.5144358277320862,
      "learning_rate": 1.8255410596138555e-05,
      "loss": 0.1586,
      "step": 5155
    },
    {
      "epoch": 0.2619422305700797,
      "grad_norm": 0.46459242701530457,
      "learning_rate": 1.8253718462866138e-05,
      "loss": 0.1571,
      "step": 5160
    },
    {
      "epoch": 0.2621960505609422,
      "grad_norm": 0.8006240725517273,
      "learning_rate": 1.825202632959372e-05,
      "loss": 0.1511,
      "step": 5165
    },
    {
      "epoch": 0.2624498705518047,
      "grad_norm": 0.5427458882331848,
      "learning_rate": 1.8250334196321305e-05,
      "loss": 0.1455,
      "step": 5170
    },
    {
      "epoch": 0.2627036905426671,
      "grad_norm": 0.6491566300392151,
      "learning_rate": 1.8248642063048886e-05,
      "loss": 0.1539,
      "step": 5175
    },
    {
      "epoch": 0.2629575105335296,
      "grad_norm": 0.9627673029899597,
      "learning_rate": 1.824694992977647e-05,
      "loss": 0.1524,
      "step": 5180
    },
    {
      "epoch": 0.2632113305243921,
      "grad_norm": 0.5841239094734192,
      "learning_rate": 1.8245257796504056e-05,
      "loss": 0.1476,
      "step": 5185
    },
    {
      "epoch": 0.2634651505152546,
      "grad_norm": 0.501349925994873,
      "learning_rate": 1.8243565663231636e-05,
      "loss": 0.1262,
      "step": 5190
    },
    {
      "epoch": 0.26371897050611703,
      "grad_norm": 0.899695873260498,
      "learning_rate": 1.8241873529959223e-05,
      "loss": 0.1508,
      "step": 5195
    },
    {
      "epoch": 0.2639727904969795,
      "grad_norm": 0.4951866865158081,
      "learning_rate": 1.8240181396686803e-05,
      "loss": 0.1479,
      "step": 5200
    },
    {
      "epoch": 0.264226610487842,
      "grad_norm": 0.5623950362205505,
      "learning_rate": 1.8238489263414387e-05,
      "loss": 0.1641,
      "step": 5205
    },
    {
      "epoch": 0.2644804304787045,
      "grad_norm": 0.4208991527557373,
      "learning_rate": 1.8236797130141974e-05,
      "loss": 0.1436,
      "step": 5210
    },
    {
      "epoch": 0.264734250469567,
      "grad_norm": 0.41514652967453003,
      "learning_rate": 1.8235104996869554e-05,
      "loss": 0.1507,
      "step": 5215
    },
    {
      "epoch": 0.26498807046042944,
      "grad_norm": 0.4979642629623413,
      "learning_rate": 1.8233412863597137e-05,
      "loss": 0.1338,
      "step": 5220
    },
    {
      "epoch": 0.26524189045129193,
      "grad_norm": 0.8609977960586548,
      "learning_rate": 1.823172073032472e-05,
      "loss": 0.1624,
      "step": 5225
    },
    {
      "epoch": 0.2654957104421544,
      "grad_norm": 0.4810056984424591,
      "learning_rate": 1.8230028597052305e-05,
      "loss": 0.1511,
      "step": 5230
    },
    {
      "epoch": 0.2657495304330169,
      "grad_norm": 0.5637201070785522,
      "learning_rate": 1.822833646377989e-05,
      "loss": 0.1451,
      "step": 5235
    },
    {
      "epoch": 0.2660033504238794,
      "grad_norm": 0.77173912525177,
      "learning_rate": 1.822664433050747e-05,
      "loss": 0.1694,
      "step": 5240
    },
    {
      "epoch": 0.26625717041474184,
      "grad_norm": 0.7287624478340149,
      "learning_rate": 1.8224952197235055e-05,
      "loss": 0.1474,
      "step": 5245
    },
    {
      "epoch": 0.26651099040560433,
      "grad_norm": 0.6791149973869324,
      "learning_rate": 1.822326006396264e-05,
      "loss": 0.1321,
      "step": 5250
    },
    {
      "epoch": 0.2667648103964668,
      "grad_norm": 0.486334890127182,
      "learning_rate": 1.8221567930690222e-05,
      "loss": 0.145,
      "step": 5255
    },
    {
      "epoch": 0.2670186303873293,
      "grad_norm": 0.4670858383178711,
      "learning_rate": 1.8219875797417806e-05,
      "loss": 0.1335,
      "step": 5260
    },
    {
      "epoch": 0.2672724503781918,
      "grad_norm": 0.4310056269168854,
      "learning_rate": 1.821818366414539e-05,
      "loss": 0.1339,
      "step": 5265
    },
    {
      "epoch": 0.26752627036905424,
      "grad_norm": 0.5363441109657288,
      "learning_rate": 1.8216491530872973e-05,
      "loss": 0.1421,
      "step": 5270
    },
    {
      "epoch": 0.26778009035991673,
      "grad_norm": 0.6194841861724854,
      "learning_rate": 1.8214799397600556e-05,
      "loss": 0.1682,
      "step": 5275
    },
    {
      "epoch": 0.2680339103507792,
      "grad_norm": 0.6221051216125488,
      "learning_rate": 1.821310726432814e-05,
      "loss": 0.139,
      "step": 5280
    },
    {
      "epoch": 0.2682877303416417,
      "grad_norm": 0.5695869326591492,
      "learning_rate": 1.8211415131055724e-05,
      "loss": 0.1409,
      "step": 5285
    },
    {
      "epoch": 0.2685415503325042,
      "grad_norm": 0.4726913273334503,
      "learning_rate": 1.8209722997783307e-05,
      "loss": 0.1443,
      "step": 5290
    },
    {
      "epoch": 0.26879537032336664,
      "grad_norm": 0.5159754157066345,
      "learning_rate": 1.820803086451089e-05,
      "loss": 0.158,
      "step": 5295
    },
    {
      "epoch": 0.26904919031422914,
      "grad_norm": 1.0508424043655396,
      "learning_rate": 1.8206338731238474e-05,
      "loss": 0.1508,
      "step": 5300
    },
    {
      "epoch": 0.26930301030509163,
      "grad_norm": 0.49234750866889954,
      "learning_rate": 1.8204646597966058e-05,
      "loss": 0.1555,
      "step": 5305
    },
    {
      "epoch": 0.2695568302959541,
      "grad_norm": 0.5780052542686462,
      "learning_rate": 1.820295446469364e-05,
      "loss": 0.1449,
      "step": 5310
    },
    {
      "epoch": 0.2698106502868166,
      "grad_norm": 0.6060476303100586,
      "learning_rate": 1.8201262331421225e-05,
      "loss": 0.1535,
      "step": 5315
    },
    {
      "epoch": 0.27006447027767905,
      "grad_norm": 0.5436288118362427,
      "learning_rate": 1.819957019814881e-05,
      "loss": 0.1513,
      "step": 5320
    },
    {
      "epoch": 0.27031829026854154,
      "grad_norm": 0.6781036257743835,
      "learning_rate": 1.8197878064876392e-05,
      "loss": 0.1599,
      "step": 5325
    },
    {
      "epoch": 0.27057211025940403,
      "grad_norm": 0.6104758381843567,
      "learning_rate": 1.8196185931603972e-05,
      "loss": 0.1395,
      "step": 5330
    },
    {
      "epoch": 0.2708259302502665,
      "grad_norm": 0.5809837579727173,
      "learning_rate": 1.819449379833156e-05,
      "loss": 0.1341,
      "step": 5335
    },
    {
      "epoch": 0.271079750241129,
      "grad_norm": 0.9313431978225708,
      "learning_rate": 1.8192801665059143e-05,
      "loss": 0.1543,
      "step": 5340
    },
    {
      "epoch": 0.27133357023199145,
      "grad_norm": 0.7047086954116821,
      "learning_rate": 1.8191109531786726e-05,
      "loss": 0.1528,
      "step": 5345
    },
    {
      "epoch": 0.27158739022285394,
      "grad_norm": 0.4853482246398926,
      "learning_rate": 1.818941739851431e-05,
      "loss": 0.1488,
      "step": 5350
    },
    {
      "epoch": 0.27184121021371643,
      "grad_norm": 1.5312269926071167,
      "learning_rate": 1.818772526524189e-05,
      "loss": 0.1581,
      "step": 5355
    },
    {
      "epoch": 0.2720950302045789,
      "grad_norm": 0.6196808815002441,
      "learning_rate": 1.8186033131969477e-05,
      "loss": 0.1306,
      "step": 5360
    },
    {
      "epoch": 0.2723488501954414,
      "grad_norm": 0.7211527228355408,
      "learning_rate": 1.818434099869706e-05,
      "loss": 0.1374,
      "step": 5365
    },
    {
      "epoch": 0.27260267018630385,
      "grad_norm": 0.4848230481147766,
      "learning_rate": 1.818264886542464e-05,
      "loss": 0.1495,
      "step": 5370
    },
    {
      "epoch": 0.27285649017716634,
      "grad_norm": 0.5052759051322937,
      "learning_rate": 1.8180956732152227e-05,
      "loss": 0.1601,
      "step": 5375
    },
    {
      "epoch": 0.27311031016802884,
      "grad_norm": 0.6089451909065247,
      "learning_rate": 1.8179264598879807e-05,
      "loss": 0.1567,
      "step": 5380
    },
    {
      "epoch": 0.2733641301588913,
      "grad_norm": 0.5925761461257935,
      "learning_rate": 1.8177572465607394e-05,
      "loss": 0.1406,
      "step": 5385
    },
    {
      "epoch": 0.2736179501497538,
      "grad_norm": 0.6521849632263184,
      "learning_rate": 1.8175880332334978e-05,
      "loss": 0.1402,
      "step": 5390
    },
    {
      "epoch": 0.27387177014061626,
      "grad_norm": 0.4725498557090759,
      "learning_rate": 1.8174188199062558e-05,
      "loss": 0.1494,
      "step": 5395
    },
    {
      "epoch": 0.27412559013147875,
      "grad_norm": 0.5570566058158875,
      "learning_rate": 1.8172496065790145e-05,
      "loss": 0.1463,
      "step": 5400
    },
    {
      "epoch": 0.27437941012234124,
      "grad_norm": 0.6081807613372803,
      "learning_rate": 1.8170803932517725e-05,
      "loss": 0.1356,
      "step": 5405
    },
    {
      "epoch": 0.27463323011320373,
      "grad_norm": 0.5569767355918884,
      "learning_rate": 1.816911179924531e-05,
      "loss": 0.131,
      "step": 5410
    },
    {
      "epoch": 0.2748870501040662,
      "grad_norm": 0.6330967545509338,
      "learning_rate": 1.8167419665972896e-05,
      "loss": 0.1529,
      "step": 5415
    },
    {
      "epoch": 0.27514087009492866,
      "grad_norm": 0.5000077486038208,
      "learning_rate": 1.8165727532700476e-05,
      "loss": 0.1352,
      "step": 5420
    },
    {
      "epoch": 0.27539469008579115,
      "grad_norm": 0.6218600273132324,
      "learning_rate": 1.816403539942806e-05,
      "loss": 0.1528,
      "step": 5425
    },
    {
      "epoch": 0.27564851007665364,
      "grad_norm": 0.5702294707298279,
      "learning_rate": 1.8162343266155643e-05,
      "loss": 0.1399,
      "step": 5430
    },
    {
      "epoch": 0.27590233006751613,
      "grad_norm": 0.9165341258049011,
      "learning_rate": 1.8160651132883226e-05,
      "loss": 0.1471,
      "step": 5435
    },
    {
      "epoch": 0.2761561500583786,
      "grad_norm": 0.5732564926147461,
      "learning_rate": 1.8158958999610813e-05,
      "loss": 0.1554,
      "step": 5440
    },
    {
      "epoch": 0.27640997004924106,
      "grad_norm": 0.5652003288269043,
      "learning_rate": 1.8157266866338394e-05,
      "loss": 0.1491,
      "step": 5445
    },
    {
      "epoch": 0.27666379004010355,
      "grad_norm": 0.5051900148391724,
      "learning_rate": 1.8155574733065977e-05,
      "loss": 0.1511,
      "step": 5450
    },
    {
      "epoch": 0.27691761003096604,
      "grad_norm": 0.4586610794067383,
      "learning_rate": 1.815388259979356e-05,
      "loss": 0.1433,
      "step": 5455
    },
    {
      "epoch": 0.27717143002182854,
      "grad_norm": 0.6589603424072266,
      "learning_rate": 1.8152190466521144e-05,
      "loss": 0.1451,
      "step": 5460
    },
    {
      "epoch": 0.277425250012691,
      "grad_norm": 0.6678264737129211,
      "learning_rate": 1.8150498333248728e-05,
      "loss": 0.1481,
      "step": 5465
    },
    {
      "epoch": 0.27767907000355346,
      "grad_norm": 0.5780376195907593,
      "learning_rate": 1.814880619997631e-05,
      "loss": 0.159,
      "step": 5470
    },
    {
      "epoch": 0.27793288999441595,
      "grad_norm": 0.72762531042099,
      "learning_rate": 1.8147114066703895e-05,
      "loss": 0.1429,
      "step": 5475
    },
    {
      "epoch": 0.27818670998527845,
      "grad_norm": 0.5006431341171265,
      "learning_rate": 1.814542193343148e-05,
      "loss": 0.1635,
      "step": 5480
    },
    {
      "epoch": 0.27844052997614094,
      "grad_norm": 0.5482341647148132,
      "learning_rate": 1.8143729800159062e-05,
      "loss": 0.1551,
      "step": 5485
    },
    {
      "epoch": 0.2786943499670034,
      "grad_norm": 0.4888313114643097,
      "learning_rate": 1.8142037666886645e-05,
      "loss": 0.1414,
      "step": 5490
    },
    {
      "epoch": 0.27894816995786587,
      "grad_norm": 0.49801504611968994,
      "learning_rate": 1.814034553361423e-05,
      "loss": 0.1561,
      "step": 5495
    },
    {
      "epoch": 0.27920198994872836,
      "grad_norm": 0.42549580335617065,
      "learning_rate": 1.8138653400341813e-05,
      "loss": 0.1265,
      "step": 5500
    },
    {
      "epoch": 0.27945580993959085,
      "grad_norm": 0.582340657711029,
      "learning_rate": 1.8136961267069396e-05,
      "loss": 0.1468,
      "step": 5505
    },
    {
      "epoch": 0.27970962993045334,
      "grad_norm": 0.46748143434524536,
      "learning_rate": 1.813526913379698e-05,
      "loss": 0.143,
      "step": 5510
    },
    {
      "epoch": 0.2799634499213158,
      "grad_norm": 0.49724528193473816,
      "learning_rate": 1.8133577000524563e-05,
      "loss": 0.1523,
      "step": 5515
    },
    {
      "epoch": 0.28021726991217827,
      "grad_norm": 0.49215927720069885,
      "learning_rate": 1.8131884867252147e-05,
      "loss": 0.1504,
      "step": 5520
    },
    {
      "epoch": 0.28047108990304076,
      "grad_norm": 0.5257487297058105,
      "learning_rate": 1.813019273397973e-05,
      "loss": 0.1279,
      "step": 5525
    },
    {
      "epoch": 0.28072490989390325,
      "grad_norm": 0.6766299605369568,
      "learning_rate": 1.8128500600707314e-05,
      "loss": 0.1515,
      "step": 5530
    },
    {
      "epoch": 0.28097872988476574,
      "grad_norm": 0.531494140625,
      "learning_rate": 1.8126808467434897e-05,
      "loss": 0.1464,
      "step": 5535
    },
    {
      "epoch": 0.2812325498756282,
      "grad_norm": 0.4178050458431244,
      "learning_rate": 1.812511633416248e-05,
      "loss": 0.1271,
      "step": 5540
    },
    {
      "epoch": 0.28148636986649067,
      "grad_norm": 0.576151967048645,
      "learning_rate": 1.8123424200890064e-05,
      "loss": 0.1328,
      "step": 5545
    },
    {
      "epoch": 0.28174018985735316,
      "grad_norm": 0.6306776404380798,
      "learning_rate": 1.8121732067617648e-05,
      "loss": 0.1511,
      "step": 5550
    },
    {
      "epoch": 0.28199400984821565,
      "grad_norm": 0.634989321231842,
      "learning_rate": 1.812003993434523e-05,
      "loss": 0.1549,
      "step": 5555
    },
    {
      "epoch": 0.28224782983907815,
      "grad_norm": 0.7044651508331299,
      "learning_rate": 1.811834780107281e-05,
      "loss": 0.1609,
      "step": 5560
    },
    {
      "epoch": 0.2825016498299406,
      "grad_norm": 0.5213934183120728,
      "learning_rate": 1.81166556678004e-05,
      "loss": 0.1359,
      "step": 5565
    },
    {
      "epoch": 0.2827554698208031,
      "grad_norm": 0.5297014117240906,
      "learning_rate": 1.8114963534527982e-05,
      "loss": 0.1493,
      "step": 5570
    },
    {
      "epoch": 0.28300928981166557,
      "grad_norm": 0.4303801953792572,
      "learning_rate": 1.8113271401255562e-05,
      "loss": 0.1408,
      "step": 5575
    },
    {
      "epoch": 0.28326310980252806,
      "grad_norm": 0.7388393878936768,
      "learning_rate": 1.811157926798315e-05,
      "loss": 0.1492,
      "step": 5580
    },
    {
      "epoch": 0.28351692979339055,
      "grad_norm": 0.6850863695144653,
      "learning_rate": 1.810988713471073e-05,
      "loss": 0.1631,
      "step": 5585
    },
    {
      "epoch": 0.283770749784253,
      "grad_norm": 0.7044445872306824,
      "learning_rate": 1.8108195001438316e-05,
      "loss": 0.1582,
      "step": 5590
    },
    {
      "epoch": 0.2840245697751155,
      "grad_norm": 0.4703899621963501,
      "learning_rate": 1.81065028681659e-05,
      "loss": 0.1369,
      "step": 5595
    },
    {
      "epoch": 0.28427838976597797,
      "grad_norm": 0.5529588460922241,
      "learning_rate": 1.810481073489348e-05,
      "loss": 0.1618,
      "step": 5600
    },
    {
      "epoch": 0.28453220975684046,
      "grad_norm": 0.6327471733093262,
      "learning_rate": 1.8103118601621067e-05,
      "loss": 0.1339,
      "step": 5605
    },
    {
      "epoch": 0.28478602974770295,
      "grad_norm": 0.5310539603233337,
      "learning_rate": 1.8101426468348647e-05,
      "loss": 0.1511,
      "step": 5610
    },
    {
      "epoch": 0.2850398497385654,
      "grad_norm": 0.5089000463485718,
      "learning_rate": 1.809973433507623e-05,
      "loss": 0.1491,
      "step": 5615
    },
    {
      "epoch": 0.2852936697294279,
      "grad_norm": 0.4953418970108032,
      "learning_rate": 1.8098042201803818e-05,
      "loss": 0.1468,
      "step": 5620
    },
    {
      "epoch": 0.28554748972029037,
      "grad_norm": 0.6269605159759521,
      "learning_rate": 1.8096350068531398e-05,
      "loss": 0.1411,
      "step": 5625
    },
    {
      "epoch": 0.28580130971115286,
      "grad_norm": 0.5403065085411072,
      "learning_rate": 1.8094657935258985e-05,
      "loss": 0.139,
      "step": 5630
    },
    {
      "epoch": 0.28605512970201535,
      "grad_norm": 0.9192875623703003,
      "learning_rate": 1.8092965801986565e-05,
      "loss": 0.133,
      "step": 5635
    },
    {
      "epoch": 0.2863089496928778,
      "grad_norm": 0.5615043044090271,
      "learning_rate": 1.809127366871415e-05,
      "loss": 0.1354,
      "step": 5640
    },
    {
      "epoch": 0.2865627696837403,
      "grad_norm": 2.930060863494873,
      "learning_rate": 1.8089581535441735e-05,
      "loss": 0.1422,
      "step": 5645
    },
    {
      "epoch": 0.2868165896746028,
      "grad_norm": 0.42745351791381836,
      "learning_rate": 1.8087889402169315e-05,
      "loss": 0.1428,
      "step": 5650
    },
    {
      "epoch": 0.28707040966546526,
      "grad_norm": 0.5274901390075684,
      "learning_rate": 1.80861972688969e-05,
      "loss": 0.164,
      "step": 5655
    },
    {
      "epoch": 0.28732422965632776,
      "grad_norm": 1.1042112112045288,
      "learning_rate": 1.8084505135624483e-05,
      "loss": 0.142,
      "step": 5660
    },
    {
      "epoch": 0.2875780496471902,
      "grad_norm": 0.6478941440582275,
      "learning_rate": 1.8082813002352066e-05,
      "loss": 0.1353,
      "step": 5665
    },
    {
      "epoch": 0.2878318696380527,
      "grad_norm": 0.6474268436431885,
      "learning_rate": 1.808112086907965e-05,
      "loss": 0.1537,
      "step": 5670
    },
    {
      "epoch": 0.2880856896289152,
      "grad_norm": 0.46995773911476135,
      "learning_rate": 1.8079428735807233e-05,
      "loss": 0.1338,
      "step": 5675
    },
    {
      "epoch": 0.28833950961977767,
      "grad_norm": 0.652370274066925,
      "learning_rate": 1.8077736602534817e-05,
      "loss": 0.1532,
      "step": 5680
    },
    {
      "epoch": 0.28859332961064016,
      "grad_norm": 0.5019606351852417,
      "learning_rate": 1.80760444692624e-05,
      "loss": 0.1606,
      "step": 5685
    },
    {
      "epoch": 0.2888471496015026,
      "grad_norm": 0.44102251529693604,
      "learning_rate": 1.8074352335989984e-05,
      "loss": 0.1462,
      "step": 5690
    },
    {
      "epoch": 0.2891009695923651,
      "grad_norm": 0.4564070701599121,
      "learning_rate": 1.8072660202717567e-05,
      "loss": 0.1537,
      "step": 5695
    },
    {
      "epoch": 0.2893547895832276,
      "grad_norm": 0.42364734411239624,
      "learning_rate": 1.807096806944515e-05,
      "loss": 0.1347,
      "step": 5700
    },
    {
      "epoch": 0.28960860957409007,
      "grad_norm": 0.5539987683296204,
      "learning_rate": 1.8069275936172734e-05,
      "loss": 0.1375,
      "step": 5705
    },
    {
      "epoch": 0.28986242956495256,
      "grad_norm": 0.4797275960445404,
      "learning_rate": 1.8067583802900318e-05,
      "loss": 0.1473,
      "step": 5710
    },
    {
      "epoch": 0.290116249555815,
      "grad_norm": 0.6089901328086853,
      "learning_rate": 1.80658916696279e-05,
      "loss": 0.1346,
      "step": 5715
    },
    {
      "epoch": 0.2903700695466775,
      "grad_norm": 0.49910488724708557,
      "learning_rate": 1.8064199536355485e-05,
      "loss": 0.1322,
      "step": 5720
    },
    {
      "epoch": 0.29062388953754,
      "grad_norm": 0.45843714475631714,
      "learning_rate": 1.806250740308307e-05,
      "loss": 0.1367,
      "step": 5725
    },
    {
      "epoch": 0.2908777095284025,
      "grad_norm": 1.3286268711090088,
      "learning_rate": 1.8060815269810652e-05,
      "loss": 0.1337,
      "step": 5730
    },
    {
      "epoch": 0.29113152951926496,
      "grad_norm": 0.5683203339576721,
      "learning_rate": 1.8059123136538236e-05,
      "loss": 0.1415,
      "step": 5735
    },
    {
      "epoch": 0.2913853495101274,
      "grad_norm": 0.6591338515281677,
      "learning_rate": 1.805743100326582e-05,
      "loss": 0.1362,
      "step": 5740
    },
    {
      "epoch": 0.2916391695009899,
      "grad_norm": 0.6420497298240662,
      "learning_rate": 1.8055738869993403e-05,
      "loss": 0.1575,
      "step": 5745
    },
    {
      "epoch": 0.2918929894918524,
      "grad_norm": 0.4309554100036621,
      "learning_rate": 1.8054046736720986e-05,
      "loss": 0.1502,
      "step": 5750
    },
    {
      "epoch": 0.2921468094827149,
      "grad_norm": 1.4654325246810913,
      "learning_rate": 1.805235460344857e-05,
      "loss": 0.1285,
      "step": 5755
    },
    {
      "epoch": 0.2924006294735773,
      "grad_norm": 0.667293906211853,
      "learning_rate": 1.8050662470176153e-05,
      "loss": 0.1473,
      "step": 5760
    },
    {
      "epoch": 0.2926544494644398,
      "grad_norm": 0.4698887765407562,
      "learning_rate": 1.8048970336903734e-05,
      "loss": 0.1396,
      "step": 5765
    },
    {
      "epoch": 0.2929082694553023,
      "grad_norm": 0.5942188501358032,
      "learning_rate": 1.804727820363132e-05,
      "loss": 0.1436,
      "step": 5770
    },
    {
      "epoch": 0.2931620894461648,
      "grad_norm": 0.34591802954673767,
      "learning_rate": 1.8045586070358904e-05,
      "loss": 0.1373,
      "step": 5775
    },
    {
      "epoch": 0.2934159094370273,
      "grad_norm": 0.4438014030456543,
      "learning_rate": 1.8043893937086488e-05,
      "loss": 0.1468,
      "step": 5780
    },
    {
      "epoch": 0.2936697294278897,
      "grad_norm": 0.5076015591621399,
      "learning_rate": 1.804220180381407e-05,
      "loss": 0.1318,
      "step": 5785
    },
    {
      "epoch": 0.2939235494187522,
      "grad_norm": 0.5563536882400513,
      "learning_rate": 1.804050967054165e-05,
      "loss": 0.1402,
      "step": 5790
    },
    {
      "epoch": 0.2941773694096147,
      "grad_norm": 14.195382118225098,
      "learning_rate": 1.8038817537269238e-05,
      "loss": 0.1376,
      "step": 5795
    },
    {
      "epoch": 0.2944311894004772,
      "grad_norm": 0.40444865822792053,
      "learning_rate": 1.8037125403996822e-05,
      "loss": 0.1234,
      "step": 5800
    },
    {
      "epoch": 0.2946850093913397,
      "grad_norm": 0.5746991038322449,
      "learning_rate": 1.8035433270724402e-05,
      "loss": 0.1297,
      "step": 5805
    },
    {
      "epoch": 0.2949388293822021,
      "grad_norm": 0.31169363856315613,
      "learning_rate": 1.803374113745199e-05,
      "loss": 0.136,
      "step": 5810
    },
    {
      "epoch": 0.2951926493730646,
      "grad_norm": 0.4355963170528412,
      "learning_rate": 1.803204900417957e-05,
      "loss": 0.1352,
      "step": 5815
    },
    {
      "epoch": 0.2954464693639271,
      "grad_norm": 0.5133848786354065,
      "learning_rate": 1.8030356870907153e-05,
      "loss": 0.149,
      "step": 5820
    },
    {
      "epoch": 0.2957002893547896,
      "grad_norm": 0.5429642200469971,
      "learning_rate": 1.802866473763474e-05,
      "loss": 0.148,
      "step": 5825
    },
    {
      "epoch": 0.2959541093456521,
      "grad_norm": 0.777310848236084,
      "learning_rate": 1.802697260436232e-05,
      "loss": 0.1203,
      "step": 5830
    },
    {
      "epoch": 0.2962079293365145,
      "grad_norm": 0.5586498975753784,
      "learning_rate": 1.8025280471089907e-05,
      "loss": 0.1489,
      "step": 5835
    },
    {
      "epoch": 0.296461749327377,
      "grad_norm": 0.4667200446128845,
      "learning_rate": 1.8023588337817487e-05,
      "loss": 0.1203,
      "step": 5840
    },
    {
      "epoch": 0.2967155693182395,
      "grad_norm": 0.7403630614280701,
      "learning_rate": 1.802189620454507e-05,
      "loss": 0.1111,
      "step": 5845
    },
    {
      "epoch": 0.296969389309102,
      "grad_norm": 1.7740875482559204,
      "learning_rate": 1.8020204071272657e-05,
      "loss": 0.1303,
      "step": 5850
    },
    {
      "epoch": 0.2972232092999645,
      "grad_norm": 0.44264400005340576,
      "learning_rate": 1.8018511938000237e-05,
      "loss": 0.141,
      "step": 5855
    },
    {
      "epoch": 0.2974770292908269,
      "grad_norm": 0.6758630275726318,
      "learning_rate": 1.801681980472782e-05,
      "loss": 0.1291,
      "step": 5860
    },
    {
      "epoch": 0.2977308492816894,
      "grad_norm": 0.6458776593208313,
      "learning_rate": 1.8015127671455405e-05,
      "loss": 0.1398,
      "step": 5865
    },
    {
      "epoch": 0.2979846692725519,
      "grad_norm": 0.6976125836372375,
      "learning_rate": 1.8013435538182988e-05,
      "loss": 0.1499,
      "step": 5870
    },
    {
      "epoch": 0.2982384892634144,
      "grad_norm": 0.7639079689979553,
      "learning_rate": 1.801174340491057e-05,
      "loss": 0.1473,
      "step": 5875
    },
    {
      "epoch": 0.2984923092542769,
      "grad_norm": 0.5145253539085388,
      "learning_rate": 1.8010051271638155e-05,
      "loss": 0.127,
      "step": 5880
    },
    {
      "epoch": 0.2987461292451393,
      "grad_norm": 0.5991470217704773,
      "learning_rate": 1.800835913836574e-05,
      "loss": 0.1434,
      "step": 5885
    },
    {
      "epoch": 0.2989999492360018,
      "grad_norm": 0.5576327443122864,
      "learning_rate": 1.8006667005093322e-05,
      "loss": 0.1258,
      "step": 5890
    },
    {
      "epoch": 0.2992537692268643,
      "grad_norm": 0.8021837472915649,
      "learning_rate": 1.8004974871820906e-05,
      "loss": 0.1543,
      "step": 5895
    },
    {
      "epoch": 0.2995075892177268,
      "grad_norm": 0.6771152019500732,
      "learning_rate": 1.800328273854849e-05,
      "loss": 0.1559,
      "step": 5900
    },
    {
      "epoch": 0.2997614092085893,
      "grad_norm": 0.6033034324645996,
      "learning_rate": 1.8001590605276073e-05,
      "loss": 0.1469,
      "step": 5905
    },
    {
      "epoch": 0.3000152291994517,
      "grad_norm": 0.4367592930793762,
      "learning_rate": 1.7999898472003656e-05,
      "loss": 0.1241,
      "step": 5910
    },
    {
      "epoch": 0.3002690491903142,
      "grad_norm": 0.8468542098999023,
      "learning_rate": 1.799820633873124e-05,
      "loss": 0.1336,
      "step": 5915
    },
    {
      "epoch": 0.3005228691811767,
      "grad_norm": 0.5358647108078003,
      "learning_rate": 1.7996514205458824e-05,
      "loss": 0.1435,
      "step": 5920
    },
    {
      "epoch": 0.3007766891720392,
      "grad_norm": 0.35355257987976074,
      "learning_rate": 1.7994822072186407e-05,
      "loss": 0.1468,
      "step": 5925
    },
    {
      "epoch": 0.3010305091629017,
      "grad_norm": 0.49814465641975403,
      "learning_rate": 1.799312993891399e-05,
      "loss": 0.133,
      "step": 5930
    },
    {
      "epoch": 0.30128432915376413,
      "grad_norm": 0.505181074142456,
      "learning_rate": 1.7991437805641574e-05,
      "loss": 0.1573,
      "step": 5935
    },
    {
      "epoch": 0.3015381491446266,
      "grad_norm": 0.5839815139770508,
      "learning_rate": 1.7989745672369158e-05,
      "loss": 0.1351,
      "step": 5940
    },
    {
      "epoch": 0.3017919691354891,
      "grad_norm": 0.7596808075904846,
      "learning_rate": 1.798805353909674e-05,
      "loss": 0.1488,
      "step": 5945
    },
    {
      "epoch": 0.3020457891263516,
      "grad_norm": 1.0872101783752441,
      "learning_rate": 1.7986361405824325e-05,
      "loss": 0.1295,
      "step": 5950
    },
    {
      "epoch": 0.3022996091172141,
      "grad_norm": 0.4910557270050049,
      "learning_rate": 1.798466927255191e-05,
      "loss": 0.1453,
      "step": 5955
    },
    {
      "epoch": 0.30255342910807653,
      "grad_norm": 0.46220898628234863,
      "learning_rate": 1.7982977139279492e-05,
      "loss": 0.1402,
      "step": 5960
    },
    {
      "epoch": 0.302807249098939,
      "grad_norm": 0.6212031245231628,
      "learning_rate": 1.7981285006007075e-05,
      "loss": 0.1473,
      "step": 5965
    },
    {
      "epoch": 0.3030610690898015,
      "grad_norm": 1.0893614292144775,
      "learning_rate": 1.797959287273466e-05,
      "loss": 0.1528,
      "step": 5970
    },
    {
      "epoch": 0.303314889080664,
      "grad_norm": 0.6324992775917053,
      "learning_rate": 1.7977900739462242e-05,
      "loss": 0.1384,
      "step": 5975
    },
    {
      "epoch": 0.3035687090715265,
      "grad_norm": 0.4653703272342682,
      "learning_rate": 1.7976208606189826e-05,
      "loss": 0.1314,
      "step": 5980
    },
    {
      "epoch": 0.30382252906238894,
      "grad_norm": 0.901879608631134,
      "learning_rate": 1.797451647291741e-05,
      "loss": 0.1424,
      "step": 5985
    },
    {
      "epoch": 0.3040763490532514,
      "grad_norm": 0.7357150316238403,
      "learning_rate": 1.7972824339644993e-05,
      "loss": 0.1383,
      "step": 5990
    },
    {
      "epoch": 0.3043301690441139,
      "grad_norm": 0.4859638810157776,
      "learning_rate": 1.7971132206372573e-05,
      "loss": 0.1473,
      "step": 5995
    },
    {
      "epoch": 0.3045839890349764,
      "grad_norm": 0.5311628580093384,
      "learning_rate": 1.796944007310016e-05,
      "loss": 0.1479,
      "step": 6000
    },
    {
      "epoch": 0.3048378090258389,
      "grad_norm": 0.48468127846717834,
      "learning_rate": 1.7967747939827744e-05,
      "loss": 0.1397,
      "step": 6005
    },
    {
      "epoch": 0.30509162901670134,
      "grad_norm": 0.5624568462371826,
      "learning_rate": 1.7966055806555324e-05,
      "loss": 0.1426,
      "step": 6010
    },
    {
      "epoch": 0.30534544900756383,
      "grad_norm": 0.6200907826423645,
      "learning_rate": 1.796436367328291e-05,
      "loss": 0.1372,
      "step": 6015
    },
    {
      "epoch": 0.3055992689984263,
      "grad_norm": 0.8948644995689392,
      "learning_rate": 1.796267154001049e-05,
      "loss": 0.1448,
      "step": 6020
    },
    {
      "epoch": 0.3058530889892888,
      "grad_norm": 0.6271385550498962,
      "learning_rate": 1.7960979406738078e-05,
      "loss": 0.1432,
      "step": 6025
    },
    {
      "epoch": 0.3061069089801513,
      "grad_norm": 0.6979460120201111,
      "learning_rate": 1.795928727346566e-05,
      "loss": 0.1276,
      "step": 6030
    },
    {
      "epoch": 0.30636072897101374,
      "grad_norm": 1.2904983758926392,
      "learning_rate": 1.795759514019324e-05,
      "loss": 0.1406,
      "step": 6035
    },
    {
      "epoch": 0.30661454896187623,
      "grad_norm": 0.6335322856903076,
      "learning_rate": 1.795590300692083e-05,
      "loss": 0.1433,
      "step": 6040
    },
    {
      "epoch": 0.3068683689527387,
      "grad_norm": 0.4658793807029724,
      "learning_rate": 1.795421087364841e-05,
      "loss": 0.1318,
      "step": 6045
    },
    {
      "epoch": 0.3071221889436012,
      "grad_norm": 0.5421667098999023,
      "learning_rate": 1.7952518740375992e-05,
      "loss": 0.142,
      "step": 6050
    },
    {
      "epoch": 0.30737600893446365,
      "grad_norm": 0.6543579697608948,
      "learning_rate": 1.7950826607103576e-05,
      "loss": 0.1387,
      "step": 6055
    },
    {
      "epoch": 0.30762982892532614,
      "grad_norm": 0.4661172330379486,
      "learning_rate": 1.794913447383116e-05,
      "loss": 0.142,
      "step": 6060
    },
    {
      "epoch": 0.30788364891618863,
      "grad_norm": 0.4072614014148712,
      "learning_rate": 1.7947442340558743e-05,
      "loss": 0.133,
      "step": 6065
    },
    {
      "epoch": 0.3081374689070511,
      "grad_norm": 0.46051713824272156,
      "learning_rate": 1.7945750207286326e-05,
      "loss": 0.1249,
      "step": 6070
    },
    {
      "epoch": 0.3083912888979136,
      "grad_norm": 0.5725454092025757,
      "learning_rate": 1.794405807401391e-05,
      "loss": 0.1221,
      "step": 6075
    },
    {
      "epoch": 0.30864510888877605,
      "grad_norm": 0.5743607878684998,
      "learning_rate": 1.7942365940741494e-05,
      "loss": 0.1223,
      "step": 6080
    },
    {
      "epoch": 0.30889892887963855,
      "grad_norm": 0.7320393919944763,
      "learning_rate": 1.7940673807469077e-05,
      "loss": 0.1453,
      "step": 6085
    },
    {
      "epoch": 0.30915274887050104,
      "grad_norm": 0.4180808365345001,
      "learning_rate": 1.793898167419666e-05,
      "loss": 0.1409,
      "step": 6090
    },
    {
      "epoch": 0.30940656886136353,
      "grad_norm": 0.5129685997962952,
      "learning_rate": 1.7937289540924244e-05,
      "loss": 0.1372,
      "step": 6095
    },
    {
      "epoch": 0.309660388852226,
      "grad_norm": 0.630707859992981,
      "learning_rate": 1.7935597407651828e-05,
      "loss": 0.1319,
      "step": 6100
    },
    {
      "epoch": 0.30991420884308846,
      "grad_norm": 0.5199107527732849,
      "learning_rate": 1.793390527437941e-05,
      "loss": 0.1371,
      "step": 6105
    },
    {
      "epoch": 0.31016802883395095,
      "grad_norm": 0.5185748338699341,
      "learning_rate": 1.7932213141106995e-05,
      "loss": 0.1467,
      "step": 6110
    },
    {
      "epoch": 0.31042184882481344,
      "grad_norm": 0.7415216565132141,
      "learning_rate": 1.793052100783458e-05,
      "loss": 0.1482,
      "step": 6115
    },
    {
      "epoch": 0.31067566881567593,
      "grad_norm": 0.5409974455833435,
      "learning_rate": 1.7928828874562162e-05,
      "loss": 0.1218,
      "step": 6120
    },
    {
      "epoch": 0.3109294888065384,
      "grad_norm": 0.47021251916885376,
      "learning_rate": 1.7927136741289745e-05,
      "loss": 0.137,
      "step": 6125
    },
    {
      "epoch": 0.31118330879740086,
      "grad_norm": 0.36402925848960876,
      "learning_rate": 1.792544460801733e-05,
      "loss": 0.1197,
      "step": 6130
    },
    {
      "epoch": 0.31143712878826335,
      "grad_norm": 0.485312819480896,
      "learning_rate": 1.7923752474744913e-05,
      "loss": 0.1321,
      "step": 6135
    },
    {
      "epoch": 0.31169094877912584,
      "grad_norm": 0.7221769690513611,
      "learning_rate": 1.7922060341472496e-05,
      "loss": 0.1479,
      "step": 6140
    },
    {
      "epoch": 0.31194476876998833,
      "grad_norm": 0.722754180431366,
      "learning_rate": 1.792036820820008e-05,
      "loss": 0.136,
      "step": 6145
    },
    {
      "epoch": 0.3121985887608508,
      "grad_norm": 0.43338173627853394,
      "learning_rate": 1.7918676074927663e-05,
      "loss": 0.135,
      "step": 6150
    },
    {
      "epoch": 0.31245240875171326,
      "grad_norm": 0.7877047061920166,
      "learning_rate": 1.7916983941655247e-05,
      "loss": 0.1345,
      "step": 6155
    },
    {
      "epoch": 0.31270622874257575,
      "grad_norm": 0.8918716907501221,
      "learning_rate": 1.791529180838283e-05,
      "loss": 0.1506,
      "step": 6160
    },
    {
      "epoch": 0.31296004873343825,
      "grad_norm": 0.4334893524646759,
      "learning_rate": 1.7913599675110414e-05,
      "loss": 0.1468,
      "step": 6165
    },
    {
      "epoch": 0.31321386872430074,
      "grad_norm": 0.6378242373466492,
      "learning_rate": 1.7911907541837997e-05,
      "loss": 0.1381,
      "step": 6170
    },
    {
      "epoch": 0.31346768871516323,
      "grad_norm": 0.6442030072212219,
      "learning_rate": 1.791021540856558e-05,
      "loss": 0.1546,
      "step": 6175
    },
    {
      "epoch": 0.31372150870602566,
      "grad_norm": 0.5167520046234131,
      "learning_rate": 1.7908523275293164e-05,
      "loss": 0.1357,
      "step": 6180
    },
    {
      "epoch": 0.31397532869688816,
      "grad_norm": 0.5824690461158752,
      "learning_rate": 1.7906831142020748e-05,
      "loss": 0.1268,
      "step": 6185
    },
    {
      "epoch": 0.31422914868775065,
      "grad_norm": 0.7088459134101868,
      "learning_rate": 1.790513900874833e-05,
      "loss": 0.1586,
      "step": 6190
    },
    {
      "epoch": 0.31448296867861314,
      "grad_norm": 0.6302781701087952,
      "learning_rate": 1.7903446875475915e-05,
      "loss": 0.1298,
      "step": 6195
    },
    {
      "epoch": 0.31473678866947563,
      "grad_norm": 0.5944651365280151,
      "learning_rate": 1.7901754742203495e-05,
      "loss": 0.1448,
      "step": 6200
    },
    {
      "epoch": 0.31499060866033807,
      "grad_norm": 0.679571270942688,
      "learning_rate": 1.7900062608931082e-05,
      "loss": 0.1469,
      "step": 6205
    },
    {
      "epoch": 0.31524442865120056,
      "grad_norm": 0.522102952003479,
      "learning_rate": 1.7898370475658666e-05,
      "loss": 0.1414,
      "step": 6210
    },
    {
      "epoch": 0.31549824864206305,
      "grad_norm": 0.5716415047645569,
      "learning_rate": 1.789667834238625e-05,
      "loss": 0.1409,
      "step": 6215
    },
    {
      "epoch": 0.31575206863292554,
      "grad_norm": 0.901875913143158,
      "learning_rate": 1.7894986209113833e-05,
      "loss": 0.1292,
      "step": 6220
    },
    {
      "epoch": 0.31600588862378803,
      "grad_norm": 0.7464191317558289,
      "learning_rate": 1.7893294075841413e-05,
      "loss": 0.1411,
      "step": 6225
    },
    {
      "epoch": 0.31625970861465047,
      "grad_norm": 0.451816201210022,
      "learning_rate": 1.7891601942569e-05,
      "loss": 0.1368,
      "step": 6230
    },
    {
      "epoch": 0.31651352860551296,
      "grad_norm": 0.6408305168151855,
      "learning_rate": 1.788990980929658e-05,
      "loss": 0.1352,
      "step": 6235
    },
    {
      "epoch": 0.31676734859637545,
      "grad_norm": 0.4555467367172241,
      "learning_rate": 1.7888217676024164e-05,
      "loss": 0.1309,
      "step": 6240
    },
    {
      "epoch": 0.31702116858723794,
      "grad_norm": 0.46165990829467773,
      "learning_rate": 1.788652554275175e-05,
      "loss": 0.1435,
      "step": 6245
    },
    {
      "epoch": 0.31727498857810044,
      "grad_norm": 0.40598687529563904,
      "learning_rate": 1.788483340947933e-05,
      "loss": 0.1415,
      "step": 6250
    },
    {
      "epoch": 0.3175288085689629,
      "grad_norm": 0.6253679990768433,
      "learning_rate": 1.7883141276206914e-05,
      "loss": 0.125,
      "step": 6255
    },
    {
      "epoch": 0.31778262855982536,
      "grad_norm": 0.46861565113067627,
      "learning_rate": 1.7881449142934498e-05,
      "loss": 0.1302,
      "step": 6260
    },
    {
      "epoch": 0.31803644855068786,
      "grad_norm": 0.6676781177520752,
      "learning_rate": 1.787975700966208e-05,
      "loss": 0.1358,
      "step": 6265
    },
    {
      "epoch": 0.31829026854155035,
      "grad_norm": 0.6015679240226746,
      "learning_rate": 1.7878064876389668e-05,
      "loss": 0.1259,
      "step": 6270
    },
    {
      "epoch": 0.31854408853241284,
      "grad_norm": 0.5872548818588257,
      "learning_rate": 1.787637274311725e-05,
      "loss": 0.138,
      "step": 6275
    },
    {
      "epoch": 0.3187979085232753,
      "grad_norm": 0.4974575936794281,
      "learning_rate": 1.7874680609844832e-05,
      "loss": 0.1137,
      "step": 6280
    },
    {
      "epoch": 0.31905172851413777,
      "grad_norm": 0.5124915838241577,
      "learning_rate": 1.7872988476572415e-05,
      "loss": 0.1423,
      "step": 6285
    },
    {
      "epoch": 0.31930554850500026,
      "grad_norm": 0.5141370892524719,
      "learning_rate": 1.78712963433e-05,
      "loss": 0.1341,
      "step": 6290
    },
    {
      "epoch": 0.31955936849586275,
      "grad_norm": 0.621908962726593,
      "learning_rate": 1.7869604210027583e-05,
      "loss": 0.1241,
      "step": 6295
    },
    {
      "epoch": 0.31981318848672524,
      "grad_norm": 0.8363323211669922,
      "learning_rate": 1.7867912076755166e-05,
      "loss": 0.1406,
      "step": 6300
    },
    {
      "epoch": 0.3200670084775877,
      "grad_norm": 0.48739150166511536,
      "learning_rate": 1.786621994348275e-05,
      "loss": 0.1296,
      "step": 6305
    },
    {
      "epoch": 0.32032082846845017,
      "grad_norm": 0.629400908946991,
      "learning_rate": 1.7864527810210333e-05,
      "loss": 0.1317,
      "step": 6310
    },
    {
      "epoch": 0.32057464845931266,
      "grad_norm": 0.4386022090911865,
      "learning_rate": 1.7862835676937917e-05,
      "loss": 0.1283,
      "step": 6315
    },
    {
      "epoch": 0.32082846845017515,
      "grad_norm": 0.44304966926574707,
      "learning_rate": 1.78611435436655e-05,
      "loss": 0.129,
      "step": 6320
    },
    {
      "epoch": 0.3210822884410376,
      "grad_norm": 0.8448670506477356,
      "learning_rate": 1.7859451410393084e-05,
      "loss": 0.1452,
      "step": 6325
    },
    {
      "epoch": 0.3213361084319001,
      "grad_norm": 0.4689197838306427,
      "learning_rate": 1.7857759277120667e-05,
      "loss": 0.1127,
      "step": 6330
    },
    {
      "epoch": 0.32158992842276257,
      "grad_norm": 0.6370121240615845,
      "learning_rate": 1.785606714384825e-05,
      "loss": 0.1357,
      "step": 6335
    },
    {
      "epoch": 0.32184374841362506,
      "grad_norm": 0.5023421049118042,
      "learning_rate": 1.7854375010575834e-05,
      "loss": 0.1465,
      "step": 6340
    },
    {
      "epoch": 0.32209756840448756,
      "grad_norm": 0.6802017688751221,
      "learning_rate": 1.7852682877303418e-05,
      "loss": 0.1416,
      "step": 6345
    },
    {
      "epoch": 0.32235138839535,
      "grad_norm": 0.48663991689682007,
      "learning_rate": 1.7850990744031e-05,
      "loss": 0.1262,
      "step": 6350
    },
    {
      "epoch": 0.3226052083862125,
      "grad_norm": 0.43129396438598633,
      "learning_rate": 1.7849298610758585e-05,
      "loss": 0.1402,
      "step": 6355
    },
    {
      "epoch": 0.322859028377075,
      "grad_norm": 0.7644272446632385,
      "learning_rate": 1.784760647748617e-05,
      "loss": 0.1396,
      "step": 6360
    },
    {
      "epoch": 0.32311284836793747,
      "grad_norm": 0.6173144578933716,
      "learning_rate": 1.7845914344213752e-05,
      "loss": 0.1371,
      "step": 6365
    },
    {
      "epoch": 0.32336666835879996,
      "grad_norm": 0.6148359179496765,
      "learning_rate": 1.7844222210941336e-05,
      "loss": 0.1211,
      "step": 6370
    },
    {
      "epoch": 0.3236204883496624,
      "grad_norm": 0.5884629487991333,
      "learning_rate": 1.784253007766892e-05,
      "loss": 0.126,
      "step": 6375
    },
    {
      "epoch": 0.3238743083405249,
      "grad_norm": 0.6891257166862488,
      "learning_rate": 1.7840837944396503e-05,
      "loss": 0.1325,
      "step": 6380
    },
    {
      "epoch": 0.3241281283313874,
      "grad_norm": 0.5167293548583984,
      "learning_rate": 1.7839145811124086e-05,
      "loss": 0.1283,
      "step": 6385
    },
    {
      "epoch": 0.32438194832224987,
      "grad_norm": 0.42851531505584717,
      "learning_rate": 1.783745367785167e-05,
      "loss": 0.1353,
      "step": 6390
    },
    {
      "epoch": 0.32463576831311236,
      "grad_norm": 0.4503128230571747,
      "learning_rate": 1.7835761544579253e-05,
      "loss": 0.1259,
      "step": 6395
    },
    {
      "epoch": 0.3248895883039748,
      "grad_norm": 0.8484407067298889,
      "learning_rate": 1.7834069411306837e-05,
      "loss": 0.1467,
      "step": 6400
    },
    {
      "epoch": 0.3251434082948373,
      "grad_norm": 0.5505496263504028,
      "learning_rate": 1.7832377278034417e-05,
      "loss": 0.1503,
      "step": 6405
    },
    {
      "epoch": 0.3253972282856998,
      "grad_norm": 0.5813527703285217,
      "learning_rate": 1.7830685144762004e-05,
      "loss": 0.1375,
      "step": 6410
    },
    {
      "epoch": 0.32565104827656227,
      "grad_norm": 0.4346156120300293,
      "learning_rate": 1.7828993011489584e-05,
      "loss": 0.1548,
      "step": 6415
    },
    {
      "epoch": 0.32590486826742476,
      "grad_norm": 0.8428875207901001,
      "learning_rate": 1.782730087821717e-05,
      "loss": 0.1355,
      "step": 6420
    },
    {
      "epoch": 0.3261586882582872,
      "grad_norm": 0.5086374878883362,
      "learning_rate": 1.7825608744944755e-05,
      "loss": 0.1477,
      "step": 6425
    },
    {
      "epoch": 0.3264125082491497,
      "grad_norm": 0.49893704056739807,
      "learning_rate": 1.7823916611672335e-05,
      "loss": 0.1396,
      "step": 6430
    },
    {
      "epoch": 0.3266663282400122,
      "grad_norm": 0.4795598089694977,
      "learning_rate": 1.7822224478399922e-05,
      "loss": 0.1382,
      "step": 6435
    },
    {
      "epoch": 0.3269201482308747,
      "grad_norm": 0.5606343150138855,
      "learning_rate": 1.7820532345127502e-05,
      "loss": 0.1386,
      "step": 6440
    },
    {
      "epoch": 0.32717396822173717,
      "grad_norm": 0.48243793845176697,
      "learning_rate": 1.7818840211855086e-05,
      "loss": 0.1241,
      "step": 6445
    },
    {
      "epoch": 0.3274277882125996,
      "grad_norm": 0.5099393725395203,
      "learning_rate": 1.7817148078582672e-05,
      "loss": 0.1387,
      "step": 6450
    },
    {
      "epoch": 0.3276816082034621,
      "grad_norm": 0.3960450291633606,
      "learning_rate": 1.7815455945310253e-05,
      "loss": 0.1212,
      "step": 6455
    },
    {
      "epoch": 0.3279354281943246,
      "grad_norm": 0.7666064500808716,
      "learning_rate": 1.781376381203784e-05,
      "loss": 0.1406,
      "step": 6460
    },
    {
      "epoch": 0.3281892481851871,
      "grad_norm": 0.6125515699386597,
      "learning_rate": 1.781207167876542e-05,
      "loss": 0.1341,
      "step": 6465
    },
    {
      "epoch": 0.32844306817604957,
      "grad_norm": 0.5433800220489502,
      "learning_rate": 1.7810379545493003e-05,
      "loss": 0.1335,
      "step": 6470
    },
    {
      "epoch": 0.328696888166912,
      "grad_norm": 0.5496737957000732,
      "learning_rate": 1.780868741222059e-05,
      "loss": 0.1399,
      "step": 6475
    },
    {
      "epoch": 0.3289507081577745,
      "grad_norm": 0.549060583114624,
      "learning_rate": 1.780699527894817e-05,
      "loss": 0.1251,
      "step": 6480
    },
    {
      "epoch": 0.329204528148637,
      "grad_norm": 0.4195408821105957,
      "learning_rate": 1.7805303145675754e-05,
      "loss": 0.1245,
      "step": 6485
    },
    {
      "epoch": 0.3294583481394995,
      "grad_norm": 0.48527300357818604,
      "learning_rate": 1.7803611012403337e-05,
      "loss": 0.1325,
      "step": 6490
    },
    {
      "epoch": 0.32971216813036197,
      "grad_norm": 0.4478241801261902,
      "learning_rate": 1.780191887913092e-05,
      "loss": 0.1227,
      "step": 6495
    },
    {
      "epoch": 0.3299659881212244,
      "grad_norm": 0.3931959569454193,
      "learning_rate": 1.7800226745858505e-05,
      "loss": 0.1245,
      "step": 6500
    },
    {
      "epoch": 0.3302198081120869,
      "grad_norm": 0.33147504925727844,
      "learning_rate": 1.7798534612586088e-05,
      "loss": 0.1131,
      "step": 6505
    },
    {
      "epoch": 0.3304736281029494,
      "grad_norm": 0.6583611965179443,
      "learning_rate": 1.779684247931367e-05,
      "loss": 0.1493,
      "step": 6510
    },
    {
      "epoch": 0.3307274480938119,
      "grad_norm": 0.5396264791488647,
      "learning_rate": 1.7795150346041255e-05,
      "loss": 0.1239,
      "step": 6515
    },
    {
      "epoch": 0.3309812680846744,
      "grad_norm": 0.5870803594589233,
      "learning_rate": 1.779345821276884e-05,
      "loss": 0.1464,
      "step": 6520
    },
    {
      "epoch": 0.3312350880755368,
      "grad_norm": 0.45324239134788513,
      "learning_rate": 1.7791766079496422e-05,
      "loss": 0.1359,
      "step": 6525
    },
    {
      "epoch": 0.3314889080663993,
      "grad_norm": 0.5347620248794556,
      "learning_rate": 1.7790073946224006e-05,
      "loss": 0.1249,
      "step": 6530
    },
    {
      "epoch": 0.3317427280572618,
      "grad_norm": 0.5685572624206543,
      "learning_rate": 1.778838181295159e-05,
      "loss": 0.1421,
      "step": 6535
    },
    {
      "epoch": 0.3319965480481243,
      "grad_norm": 0.5411074161529541,
      "learning_rate": 1.7786689679679173e-05,
      "loss": 0.1233,
      "step": 6540
    },
    {
      "epoch": 0.3322503680389868,
      "grad_norm": 0.42632856965065,
      "learning_rate": 1.7784997546406756e-05,
      "loss": 0.1339,
      "step": 6545
    },
    {
      "epoch": 0.3325041880298492,
      "grad_norm": 0.45992130041122437,
      "learning_rate": 1.778330541313434e-05,
      "loss": 0.1395,
      "step": 6550
    },
    {
      "epoch": 0.3327580080207117,
      "grad_norm": 0.7072666883468628,
      "learning_rate": 1.7781613279861923e-05,
      "loss": 0.1312,
      "step": 6555
    },
    {
      "epoch": 0.3330118280115742,
      "grad_norm": 0.5420939922332764,
      "learning_rate": 1.7779921146589507e-05,
      "loss": 0.1209,
      "step": 6560
    },
    {
      "epoch": 0.3332656480024367,
      "grad_norm": 0.5253438949584961,
      "learning_rate": 1.777822901331709e-05,
      "loss": 0.1315,
      "step": 6565
    },
    {
      "epoch": 0.3335194679932992,
      "grad_norm": 0.51002436876297,
      "learning_rate": 1.7776536880044674e-05,
      "loss": 0.1467,
      "step": 6570
    },
    {
      "epoch": 0.3337732879841616,
      "grad_norm": 0.7875125408172607,
      "learning_rate": 1.7774844746772258e-05,
      "loss": 0.1201,
      "step": 6575
    },
    {
      "epoch": 0.3340271079750241,
      "grad_norm": 0.4544963240623474,
      "learning_rate": 1.777315261349984e-05,
      "loss": 0.1239,
      "step": 6580
    },
    {
      "epoch": 0.3342809279658866,
      "grad_norm": 0.579490602016449,
      "learning_rate": 1.7771460480227425e-05,
      "loss": 0.1551,
      "step": 6585
    },
    {
      "epoch": 0.3345347479567491,
      "grad_norm": 0.5468235015869141,
      "learning_rate": 1.776976834695501e-05,
      "loss": 0.1406,
      "step": 6590
    },
    {
      "epoch": 0.3347885679476115,
      "grad_norm": 0.5087330341339111,
      "learning_rate": 1.7768076213682592e-05,
      "loss": 0.14,
      "step": 6595
    },
    {
      "epoch": 0.335042387938474,
      "grad_norm": 0.48226407170295715,
      "learning_rate": 1.7766384080410175e-05,
      "loss": 0.1214,
      "step": 6600
    },
    {
      "epoch": 0.3352962079293365,
      "grad_norm": 0.4688023626804352,
      "learning_rate": 1.776469194713776e-05,
      "loss": 0.1089,
      "step": 6605
    },
    {
      "epoch": 0.335550027920199,
      "grad_norm": 0.7546516060829163,
      "learning_rate": 1.7762999813865342e-05,
      "loss": 0.1428,
      "step": 6610
    },
    {
      "epoch": 0.3358038479110615,
      "grad_norm": 0.46591129899024963,
      "learning_rate": 1.7761307680592926e-05,
      "loss": 0.126,
      "step": 6615
    },
    {
      "epoch": 0.33605766790192393,
      "grad_norm": 0.5849207043647766,
      "learning_rate": 1.7759615547320506e-05,
      "loss": 0.1322,
      "step": 6620
    },
    {
      "epoch": 0.3363114878927864,
      "grad_norm": 0.461297869682312,
      "learning_rate": 1.7757923414048093e-05,
      "loss": 0.1355,
      "step": 6625
    },
    {
      "epoch": 0.3365653078836489,
      "grad_norm": 0.4782043397426605,
      "learning_rate": 1.7756231280775677e-05,
      "loss": 0.1337,
      "step": 6630
    },
    {
      "epoch": 0.3368191278745114,
      "grad_norm": 1.3908518552780151,
      "learning_rate": 1.7754539147503257e-05,
      "loss": 0.1418,
      "step": 6635
    },
    {
      "epoch": 0.3370729478653739,
      "grad_norm": 0.5779076218605042,
      "learning_rate": 1.7752847014230844e-05,
      "loss": 0.1416,
      "step": 6640
    },
    {
      "epoch": 0.33732676785623633,
      "grad_norm": 0.4418136477470398,
      "learning_rate": 1.7751154880958424e-05,
      "loss": 0.1194,
      "step": 6645
    },
    {
      "epoch": 0.3375805878470988,
      "grad_norm": 0.47528406977653503,
      "learning_rate": 1.7749462747686007e-05,
      "loss": 0.1362,
      "step": 6650
    },
    {
      "epoch": 0.3378344078379613,
      "grad_norm": 0.6058582663536072,
      "learning_rate": 1.7747770614413594e-05,
      "loss": 0.156,
      "step": 6655
    },
    {
      "epoch": 0.3380882278288238,
      "grad_norm": 0.4911196827888489,
      "learning_rate": 1.7746078481141175e-05,
      "loss": 0.1294,
      "step": 6660
    },
    {
      "epoch": 0.3383420478196863,
      "grad_norm": 0.5730322003364563,
      "learning_rate": 1.774438634786876e-05,
      "loss": 0.1217,
      "step": 6665
    },
    {
      "epoch": 0.33859586781054873,
      "grad_norm": 0.3868119418621063,
      "learning_rate": 1.774269421459634e-05,
      "loss": 0.1237,
      "step": 6670
    },
    {
      "epoch": 0.3388496878014112,
      "grad_norm": 0.70493483543396,
      "learning_rate": 1.7741002081323925e-05,
      "loss": 0.1471,
      "step": 6675
    },
    {
      "epoch": 0.3391035077922737,
      "grad_norm": 0.6178105473518372,
      "learning_rate": 1.7739309948051512e-05,
      "loss": 0.1282,
      "step": 6680
    },
    {
      "epoch": 0.3393573277831362,
      "grad_norm": 0.5365161895751953,
      "learning_rate": 1.7737617814779092e-05,
      "loss": 0.1191,
      "step": 6685
    },
    {
      "epoch": 0.3396111477739987,
      "grad_norm": 0.4318663477897644,
      "learning_rate": 1.7735925681506676e-05,
      "loss": 0.126,
      "step": 6690
    },
    {
      "epoch": 0.33986496776486114,
      "grad_norm": 0.46456629037857056,
      "learning_rate": 1.773423354823426e-05,
      "loss": 0.1327,
      "step": 6695
    },
    {
      "epoch": 0.34011878775572363,
      "grad_norm": 0.7820452451705933,
      "learning_rate": 1.7732541414961843e-05,
      "loss": 0.131,
      "step": 6700
    },
    {
      "epoch": 0.3403726077465861,
      "grad_norm": 0.6297656297683716,
      "learning_rate": 1.7730849281689426e-05,
      "loss": 0.151,
      "step": 6705
    },
    {
      "epoch": 0.3406264277374486,
      "grad_norm": 0.4756261706352234,
      "learning_rate": 1.772915714841701e-05,
      "loss": 0.1388,
      "step": 6710
    },
    {
      "epoch": 0.3408802477283111,
      "grad_norm": 0.5708016157150269,
      "learning_rate": 1.7727465015144594e-05,
      "loss": 0.1191,
      "step": 6715
    },
    {
      "epoch": 0.34113406771917354,
      "grad_norm": 0.5044305324554443,
      "learning_rate": 1.7725772881872177e-05,
      "loss": 0.1238,
      "step": 6720
    },
    {
      "epoch": 0.34138788771003603,
      "grad_norm": 0.39757969975471497,
      "learning_rate": 1.772408074859976e-05,
      "loss": 0.1364,
      "step": 6725
    },
    {
      "epoch": 0.3416417077008985,
      "grad_norm": 0.425277441740036,
      "learning_rate": 1.7722388615327344e-05,
      "loss": 0.1233,
      "step": 6730
    },
    {
      "epoch": 0.341895527691761,
      "grad_norm": 0.5457815527915955,
      "learning_rate": 1.7720696482054928e-05,
      "loss": 0.1335,
      "step": 6735
    },
    {
      "epoch": 0.3421493476826235,
      "grad_norm": 0.9425898194313049,
      "learning_rate": 1.771900434878251e-05,
      "loss": 0.125,
      "step": 6740
    },
    {
      "epoch": 0.34240316767348594,
      "grad_norm": 0.6296150088310242,
      "learning_rate": 1.7717312215510095e-05,
      "loss": 0.1156,
      "step": 6745
    },
    {
      "epoch": 0.34265698766434843,
      "grad_norm": 0.4702848196029663,
      "learning_rate": 1.771562008223768e-05,
      "loss": 0.1213,
      "step": 6750
    },
    {
      "epoch": 0.3429108076552109,
      "grad_norm": 0.48555007576942444,
      "learning_rate": 1.7713927948965262e-05,
      "loss": 0.1297,
      "step": 6755
    },
    {
      "epoch": 0.3431646276460734,
      "grad_norm": 0.46021294593811035,
      "learning_rate": 1.7712235815692845e-05,
      "loss": 0.1325,
      "step": 6760
    },
    {
      "epoch": 0.3434184476369359,
      "grad_norm": 0.7857270836830139,
      "learning_rate": 1.771054368242043e-05,
      "loss": 0.1354,
      "step": 6765
    },
    {
      "epoch": 0.34367226762779834,
      "grad_norm": 0.538543701171875,
      "learning_rate": 1.7708851549148013e-05,
      "loss": 0.1471,
      "step": 6770
    },
    {
      "epoch": 0.34392608761866084,
      "grad_norm": 0.5883283615112305,
      "learning_rate": 1.7707159415875596e-05,
      "loss": 0.1453,
      "step": 6775
    },
    {
      "epoch": 0.34417990760952333,
      "grad_norm": 0.5997344851493835,
      "learning_rate": 1.770546728260318e-05,
      "loss": 0.1381,
      "step": 6780
    },
    {
      "epoch": 0.3444337276003858,
      "grad_norm": 0.46415236592292786,
      "learning_rate": 1.7703775149330763e-05,
      "loss": 0.1331,
      "step": 6785
    },
    {
      "epoch": 0.3446875475912483,
      "grad_norm": 0.46198391914367676,
      "learning_rate": 1.7702083016058347e-05,
      "loss": 0.129,
      "step": 6790
    },
    {
      "epoch": 0.34494136758211075,
      "grad_norm": 0.6081626415252686,
      "learning_rate": 1.770039088278593e-05,
      "loss": 0.1367,
      "step": 6795
    },
    {
      "epoch": 0.34519518757297324,
      "grad_norm": 0.5060707926750183,
      "learning_rate": 1.769869874951351e-05,
      "loss": 0.1317,
      "step": 6800
    },
    {
      "epoch": 0.34544900756383573,
      "grad_norm": 0.48941001296043396,
      "learning_rate": 1.7697006616241097e-05,
      "loss": 0.1218,
      "step": 6805
    },
    {
      "epoch": 0.3457028275546982,
      "grad_norm": 0.8074930906295776,
      "learning_rate": 1.769531448296868e-05,
      "loss": 0.1178,
      "step": 6810
    },
    {
      "epoch": 0.3459566475455607,
      "grad_norm": 0.46753403544425964,
      "learning_rate": 1.7693622349696264e-05,
      "loss": 0.1152,
      "step": 6815
    },
    {
      "epoch": 0.34621046753642315,
      "grad_norm": 0.5410972237586975,
      "learning_rate": 1.7691930216423848e-05,
      "loss": 0.1364,
      "step": 6820
    },
    {
      "epoch": 0.34646428752728564,
      "grad_norm": 0.5206438302993774,
      "learning_rate": 1.7690238083151428e-05,
      "loss": 0.1292,
      "step": 6825
    },
    {
      "epoch": 0.34671810751814813,
      "grad_norm": 0.4511585831642151,
      "learning_rate": 1.7688545949879015e-05,
      "loss": 0.1236,
      "step": 6830
    },
    {
      "epoch": 0.3469719275090106,
      "grad_norm": 0.6129817962646484,
      "learning_rate": 1.76868538166066e-05,
      "loss": 0.1209,
      "step": 6835
    },
    {
      "epoch": 0.3472257474998731,
      "grad_norm": 1.1212575435638428,
      "learning_rate": 1.768516168333418e-05,
      "loss": 0.1293,
      "step": 6840
    },
    {
      "epoch": 0.34747956749073555,
      "grad_norm": 0.760433554649353,
      "learning_rate": 1.7683469550061766e-05,
      "loss": 0.1382,
      "step": 6845
    },
    {
      "epoch": 0.34773338748159804,
      "grad_norm": 0.6482082605361938,
      "learning_rate": 1.7681777416789346e-05,
      "loss": 0.1354,
      "step": 6850
    },
    {
      "epoch": 0.34798720747246054,
      "grad_norm": 0.606609046459198,
      "learning_rate": 1.7680085283516933e-05,
      "loss": 0.1174,
      "step": 6855
    },
    {
      "epoch": 0.348241027463323,
      "grad_norm": 0.49537140130996704,
      "learning_rate": 1.7678393150244516e-05,
      "loss": 0.1272,
      "step": 6860
    },
    {
      "epoch": 0.3484948474541855,
      "grad_norm": 0.5630388259887695,
      "learning_rate": 1.7676701016972096e-05,
      "loss": 0.1455,
      "step": 6865
    },
    {
      "epoch": 0.34874866744504796,
      "grad_norm": 0.5360369086265564,
      "learning_rate": 1.7675008883699683e-05,
      "loss": 0.1206,
      "step": 6870
    },
    {
      "epoch": 0.34900248743591045,
      "grad_norm": 0.5766199231147766,
      "learning_rate": 1.7673316750427264e-05,
      "loss": 0.1311,
      "step": 6875
    },
    {
      "epoch": 0.34925630742677294,
      "grad_norm": 0.4689859449863434,
      "learning_rate": 1.7671624617154847e-05,
      "loss": 0.1259,
      "step": 6880
    },
    {
      "epoch": 0.34951012741763543,
      "grad_norm": 0.8159186840057373,
      "learning_rate": 1.7669932483882434e-05,
      "loss": 0.1237,
      "step": 6885
    },
    {
      "epoch": 0.34976394740849787,
      "grad_norm": 0.9604336619377136,
      "learning_rate": 1.7668240350610014e-05,
      "loss": 0.1288,
      "step": 6890
    },
    {
      "epoch": 0.35001776739936036,
      "grad_norm": 1.4752824306488037,
      "learning_rate": 1.7666548217337598e-05,
      "loss": 0.1251,
      "step": 6895
    },
    {
      "epoch": 0.35027158739022285,
      "grad_norm": 0.449650377035141,
      "learning_rate": 1.766485608406518e-05,
      "loss": 0.1479,
      "step": 6900
    },
    {
      "epoch": 0.35052540738108534,
      "grad_norm": 0.5203964114189148,
      "learning_rate": 1.7663163950792765e-05,
      "loss": 0.1424,
      "step": 6905
    },
    {
      "epoch": 0.35077922737194783,
      "grad_norm": 0.5282999277114868,
      "learning_rate": 1.7661471817520352e-05,
      "loss": 0.1269,
      "step": 6910
    },
    {
      "epoch": 0.35103304736281027,
      "grad_norm": 0.41369152069091797,
      "learning_rate": 1.7659779684247932e-05,
      "loss": 0.1144,
      "step": 6915
    },
    {
      "epoch": 0.35128686735367276,
      "grad_norm": 2.030303716659546,
      "learning_rate": 1.7658087550975515e-05,
      "loss": 0.1372,
      "step": 6920
    },
    {
      "epoch": 0.35154068734453525,
      "grad_norm": 0.463283896446228,
      "learning_rate": 1.76563954177031e-05,
      "loss": 0.1329,
      "step": 6925
    },
    {
      "epoch": 0.35179450733539774,
      "grad_norm": 0.4552387595176697,
      "learning_rate": 1.7654703284430683e-05,
      "loss": 0.1317,
      "step": 6930
    },
    {
      "epoch": 0.35204832732626024,
      "grad_norm": 0.9171412587165833,
      "learning_rate": 1.7653011151158266e-05,
      "loss": 0.1284,
      "step": 6935
    },
    {
      "epoch": 0.35230214731712267,
      "grad_norm": 0.43250319361686707,
      "learning_rate": 1.765131901788585e-05,
      "loss": 0.1199,
      "step": 6940
    },
    {
      "epoch": 0.35255596730798516,
      "grad_norm": 0.8429256081581116,
      "learning_rate": 1.7649626884613433e-05,
      "loss": 0.1375,
      "step": 6945
    },
    {
      "epoch": 0.35280978729884765,
      "grad_norm": 0.651530921459198,
      "learning_rate": 1.7647934751341017e-05,
      "loss": 0.1313,
      "step": 6950
    },
    {
      "epoch": 0.35306360728971015,
      "grad_norm": 0.632590115070343,
      "learning_rate": 1.76462426180686e-05,
      "loss": 0.1437,
      "step": 6955
    },
    {
      "epoch": 0.35331742728057264,
      "grad_norm": 0.5507182478904724,
      "learning_rate": 1.7644550484796184e-05,
      "loss": 0.1353,
      "step": 6960
    },
    {
      "epoch": 0.3535712472714351,
      "grad_norm": 0.5852574110031128,
      "learning_rate": 1.7642858351523767e-05,
      "loss": 0.1166,
      "step": 6965
    },
    {
      "epoch": 0.35382506726229757,
      "grad_norm": 0.4446243643760681,
      "learning_rate": 1.764116621825135e-05,
      "loss": 0.1333,
      "step": 6970
    },
    {
      "epoch": 0.35407888725316006,
      "grad_norm": 0.6388731598854065,
      "learning_rate": 1.7639474084978934e-05,
      "loss": 0.1312,
      "step": 6975
    },
    {
      "epoch": 0.35433270724402255,
      "grad_norm": 0.4913591742515564,
      "learning_rate": 1.7637781951706518e-05,
      "loss": 0.1276,
      "step": 6980
    },
    {
      "epoch": 0.35458652723488504,
      "grad_norm": 0.5663020014762878,
      "learning_rate": 1.76360898184341e-05,
      "loss": 0.1228,
      "step": 6985
    },
    {
      "epoch": 0.3548403472257475,
      "grad_norm": 0.3867751657962799,
      "learning_rate": 1.7634397685161685e-05,
      "loss": 0.1331,
      "step": 6990
    },
    {
      "epoch": 0.35509416721660997,
      "grad_norm": 1.012452483177185,
      "learning_rate": 1.763270555188927e-05,
      "loss": 0.1431,
      "step": 6995
    },
    {
      "epoch": 0.35534798720747246,
      "grad_norm": 0.5785044431686401,
      "learning_rate": 1.7631013418616852e-05,
      "loss": 0.146,
      "step": 7000
    },
    {
      "epoch": 0.35560180719833495,
      "grad_norm": 0.5155715346336365,
      "learning_rate": 1.7629321285344436e-05,
      "loss": 0.1153,
      "step": 7005
    },
    {
      "epoch": 0.35585562718919744,
      "grad_norm": 0.5367943644523621,
      "learning_rate": 1.762762915207202e-05,
      "loss": 0.125,
      "step": 7010
    },
    {
      "epoch": 0.3561094471800599,
      "grad_norm": 0.3979216516017914,
      "learning_rate": 1.7625937018799603e-05,
      "loss": 0.1232,
      "step": 7015
    },
    {
      "epoch": 0.35636326717092237,
      "grad_norm": 0.5183656215667725,
      "learning_rate": 1.7624244885527186e-05,
      "loss": 0.1181,
      "step": 7020
    },
    {
      "epoch": 0.35661708716178486,
      "grad_norm": 0.45130103826522827,
      "learning_rate": 1.762255275225477e-05,
      "loss": 0.1325,
      "step": 7025
    },
    {
      "epoch": 0.35687090715264735,
      "grad_norm": 0.7615092992782593,
      "learning_rate": 1.762086061898235e-05,
      "loss": 0.1321,
      "step": 7030
    },
    {
      "epoch": 0.35712472714350985,
      "grad_norm": 0.5867871046066284,
      "learning_rate": 1.7619168485709937e-05,
      "loss": 0.1184,
      "step": 7035
    },
    {
      "epoch": 0.3573785471343723,
      "grad_norm": 1.0698941946029663,
      "learning_rate": 1.761747635243752e-05,
      "loss": 0.1272,
      "step": 7040
    },
    {
      "epoch": 0.3576323671252348,
      "grad_norm": 0.5040280222892761,
      "learning_rate": 1.76157842191651e-05,
      "loss": 0.1323,
      "step": 7045
    },
    {
      "epoch": 0.35788618711609727,
      "grad_norm": 0.49704474210739136,
      "learning_rate": 1.7614092085892688e-05,
      "loss": 0.133,
      "step": 7050
    },
    {
      "epoch": 0.35814000710695976,
      "grad_norm": 0.7625434398651123,
      "learning_rate": 1.7612399952620268e-05,
      "loss": 0.1295,
      "step": 7055
    },
    {
      "epoch": 0.35839382709782225,
      "grad_norm": 0.5237678289413452,
      "learning_rate": 1.7610707819347855e-05,
      "loss": 0.1282,
      "step": 7060
    },
    {
      "epoch": 0.3586476470886847,
      "grad_norm": 0.640903115272522,
      "learning_rate": 1.7609015686075438e-05,
      "loss": 0.1323,
      "step": 7065
    },
    {
      "epoch": 0.3589014670795472,
      "grad_norm": 0.5553655028343201,
      "learning_rate": 1.760732355280302e-05,
      "loss": 0.1261,
      "step": 7070
    },
    {
      "epoch": 0.35915528707040967,
      "grad_norm": 0.5053536891937256,
      "learning_rate": 1.7605631419530605e-05,
      "loss": 0.1298,
      "step": 7075
    },
    {
      "epoch": 0.35940910706127216,
      "grad_norm": 0.5184898972511292,
      "learning_rate": 1.7603939286258186e-05,
      "loss": 0.1348,
      "step": 7080
    },
    {
      "epoch": 0.35966292705213465,
      "grad_norm": 0.777260959148407,
      "learning_rate": 1.760224715298577e-05,
      "loss": 0.1373,
      "step": 7085
    },
    {
      "epoch": 0.3599167470429971,
      "grad_norm": 0.4676660895347595,
      "learning_rate": 1.7600555019713356e-05,
      "loss": 0.1254,
      "step": 7090
    },
    {
      "epoch": 0.3601705670338596,
      "grad_norm": 0.6567785739898682,
      "learning_rate": 1.7598862886440936e-05,
      "loss": 0.1165,
      "step": 7095
    },
    {
      "epoch": 0.36042438702472207,
      "grad_norm": 0.7958646416664124,
      "learning_rate": 1.7597170753168523e-05,
      "loss": 0.1334,
      "step": 7100
    },
    {
      "epoch": 0.36067820701558456,
      "grad_norm": 0.43118539452552795,
      "learning_rate": 1.7595478619896103e-05,
      "loss": 0.1203,
      "step": 7105
    },
    {
      "epoch": 0.36093202700644705,
      "grad_norm": 0.44612541794776917,
      "learning_rate": 1.7593786486623687e-05,
      "loss": 0.1049,
      "step": 7110
    },
    {
      "epoch": 0.3611858469973095,
      "grad_norm": 0.43221428990364075,
      "learning_rate": 1.7592094353351274e-05,
      "loss": 0.1152,
      "step": 7115
    },
    {
      "epoch": 0.361439666988172,
      "grad_norm": 0.486248642206192,
      "learning_rate": 1.7590402220078854e-05,
      "loss": 0.1281,
      "step": 7120
    },
    {
      "epoch": 0.3616934869790345,
      "grad_norm": 0.4399028420448303,
      "learning_rate": 1.7588710086806437e-05,
      "loss": 0.1321,
      "step": 7125
    },
    {
      "epoch": 0.36194730696989696,
      "grad_norm": 0.6637834906578064,
      "learning_rate": 1.758701795353402e-05,
      "loss": 0.1285,
      "step": 7130
    },
    {
      "epoch": 0.36220112696075946,
      "grad_norm": 0.4433422386646271,
      "learning_rate": 1.7585325820261605e-05,
      "loss": 0.133,
      "step": 7135
    },
    {
      "epoch": 0.3624549469516219,
      "grad_norm": 0.7549034953117371,
      "learning_rate": 1.7583633686989188e-05,
      "loss": 0.1279,
      "step": 7140
    },
    {
      "epoch": 0.3627087669424844,
      "grad_norm": 0.48059457540512085,
      "learning_rate": 1.758194155371677e-05,
      "loss": 0.1149,
      "step": 7145
    },
    {
      "epoch": 0.3629625869333469,
      "grad_norm": 0.582513689994812,
      "learning_rate": 1.7580249420444355e-05,
      "loss": 0.1423,
      "step": 7150
    },
    {
      "epoch": 0.36321640692420937,
      "grad_norm": 0.6541072130203247,
      "learning_rate": 1.757855728717194e-05,
      "loss": 0.1229,
      "step": 7155
    },
    {
      "epoch": 0.3634702269150718,
      "grad_norm": 0.5201939344406128,
      "learning_rate": 1.7576865153899522e-05,
      "loss": 0.1194,
      "step": 7160
    },
    {
      "epoch": 0.3637240469059343,
      "grad_norm": 1.192158579826355,
      "learning_rate": 1.7575173020627106e-05,
      "loss": 0.1261,
      "step": 7165
    },
    {
      "epoch": 0.3639778668967968,
      "grad_norm": 0.5061715245246887,
      "learning_rate": 1.757348088735469e-05,
      "loss": 0.1282,
      "step": 7170
    },
    {
      "epoch": 0.3642316868876593,
      "grad_norm": 0.45553848147392273,
      "learning_rate": 1.7571788754082273e-05,
      "loss": 0.1186,
      "step": 7175
    },
    {
      "epoch": 0.36448550687852177,
      "grad_norm": 0.4875742793083191,
      "learning_rate": 1.7570096620809856e-05,
      "loss": 0.124,
      "step": 7180
    },
    {
      "epoch": 0.3647393268693842,
      "grad_norm": 0.39498981833457947,
      "learning_rate": 1.756840448753744e-05,
      "loss": 0.1223,
      "step": 7185
    },
    {
      "epoch": 0.3649931468602467,
      "grad_norm": 0.6347449421882629,
      "learning_rate": 1.7566712354265023e-05,
      "loss": 0.1164,
      "step": 7190
    },
    {
      "epoch": 0.3652469668511092,
      "grad_norm": 0.5973386764526367,
      "learning_rate": 1.7565020220992607e-05,
      "loss": 0.1207,
      "step": 7195
    },
    {
      "epoch": 0.3655007868419717,
      "grad_norm": 0.44844862818717957,
      "learning_rate": 1.756332808772019e-05,
      "loss": 0.1268,
      "step": 7200
    },
    {
      "epoch": 0.3657546068328342,
      "grad_norm": 0.44825243949890137,
      "learning_rate": 1.7561635954447774e-05,
      "loss": 0.1333,
      "step": 7205
    },
    {
      "epoch": 0.3660084268236966,
      "grad_norm": 0.664547860622406,
      "learning_rate": 1.7559943821175358e-05,
      "loss": 0.1131,
      "step": 7210
    },
    {
      "epoch": 0.3662622468145591,
      "grad_norm": 0.5262232422828674,
      "learning_rate": 1.755825168790294e-05,
      "loss": 0.1314,
      "step": 7215
    },
    {
      "epoch": 0.3665160668054216,
      "grad_norm": 0.5070418119430542,
      "learning_rate": 1.7556559554630525e-05,
      "loss": 0.1326,
      "step": 7220
    },
    {
      "epoch": 0.3667698867962841,
      "grad_norm": 0.7615456581115723,
      "learning_rate": 1.7554867421358108e-05,
      "loss": 0.1262,
      "step": 7225
    },
    {
      "epoch": 0.3670237067871466,
      "grad_norm": 0.44965729117393494,
      "learning_rate": 1.7553175288085692e-05,
      "loss": 0.1211,
      "step": 7230
    },
    {
      "epoch": 0.367277526778009,
      "grad_norm": 0.6415624022483826,
      "learning_rate": 1.7551483154813272e-05,
      "loss": 0.1188,
      "step": 7235
    },
    {
      "epoch": 0.3675313467688715,
      "grad_norm": 0.4776475131511688,
      "learning_rate": 1.754979102154086e-05,
      "loss": 0.1141,
      "step": 7240
    },
    {
      "epoch": 0.367785166759734,
      "grad_norm": 0.4545101225376129,
      "learning_rate": 1.7548098888268442e-05,
      "loss": 0.1252,
      "step": 7245
    },
    {
      "epoch": 0.3680389867505965,
      "grad_norm": 0.8184462189674377,
      "learning_rate": 1.7546406754996026e-05,
      "loss": 0.122,
      "step": 7250
    },
    {
      "epoch": 0.368292806741459,
      "grad_norm": 0.46976208686828613,
      "learning_rate": 1.754471462172361e-05,
      "loss": 0.1219,
      "step": 7255
    },
    {
      "epoch": 0.3685466267323214,
      "grad_norm": 0.41548067331314087,
      "learning_rate": 1.754302248845119e-05,
      "loss": 0.1132,
      "step": 7260
    },
    {
      "epoch": 0.3688004467231839,
      "grad_norm": 0.781281590461731,
      "learning_rate": 1.7541330355178777e-05,
      "loss": 0.1401,
      "step": 7265
    },
    {
      "epoch": 0.3690542667140464,
      "grad_norm": 0.3721882700920105,
      "learning_rate": 1.753963822190636e-05,
      "loss": 0.1154,
      "step": 7270
    },
    {
      "epoch": 0.3693080867049089,
      "grad_norm": 0.5015051960945129,
      "learning_rate": 1.753794608863394e-05,
      "loss": 0.1186,
      "step": 7275
    },
    {
      "epoch": 0.3695619066957714,
      "grad_norm": 0.49192380905151367,
      "learning_rate": 1.7536253955361527e-05,
      "loss": 0.1115,
      "step": 7280
    },
    {
      "epoch": 0.3698157266866338,
      "grad_norm": 0.4846615195274353,
      "learning_rate": 1.7534561822089107e-05,
      "loss": 0.1333,
      "step": 7285
    },
    {
      "epoch": 0.3700695466774963,
      "grad_norm": 0.40966522693634033,
      "learning_rate": 1.753286968881669e-05,
      "loss": 0.1185,
      "step": 7290
    },
    {
      "epoch": 0.3703233666683588,
      "grad_norm": 0.533069372177124,
      "learning_rate": 1.7531177555544278e-05,
      "loss": 0.1181,
      "step": 7295
    },
    {
      "epoch": 0.3705771866592213,
      "grad_norm": 0.5515578985214233,
      "learning_rate": 1.7529485422271858e-05,
      "loss": 0.1148,
      "step": 7300
    },
    {
      "epoch": 0.3708310066500838,
      "grad_norm": 0.5015854239463806,
      "learning_rate": 1.7527793288999445e-05,
      "loss": 0.1193,
      "step": 7305
    },
    {
      "epoch": 0.3710848266409462,
      "grad_norm": 0.4574815630912781,
      "learning_rate": 1.7526101155727025e-05,
      "loss": 0.1355,
      "step": 7310
    },
    {
      "epoch": 0.3713386466318087,
      "grad_norm": 0.5332877039909363,
      "learning_rate": 1.752440902245461e-05,
      "loss": 0.1175,
      "step": 7315
    },
    {
      "epoch": 0.3715924666226712,
      "grad_norm": 0.6135624647140503,
      "learning_rate": 1.7522716889182196e-05,
      "loss": 0.1268,
      "step": 7320
    },
    {
      "epoch": 0.3718462866135337,
      "grad_norm": 0.5522362589836121,
      "learning_rate": 1.7521024755909776e-05,
      "loss": 0.1205,
      "step": 7325
    },
    {
      "epoch": 0.3721001066043962,
      "grad_norm": 0.41892364621162415,
      "learning_rate": 1.751933262263736e-05,
      "loss": 0.1272,
      "step": 7330
    },
    {
      "epoch": 0.3723539265952586,
      "grad_norm": 0.5606332421302795,
      "learning_rate": 1.7517640489364943e-05,
      "loss": 0.1238,
      "step": 7335
    },
    {
      "epoch": 0.3726077465861211,
      "grad_norm": 0.4218759834766388,
      "learning_rate": 1.7515948356092526e-05,
      "loss": 0.12,
      "step": 7340
    },
    {
      "epoch": 0.3728615665769836,
      "grad_norm": 0.45138901472091675,
      "learning_rate": 1.7514256222820113e-05,
      "loss": 0.1245,
      "step": 7345
    },
    {
      "epoch": 0.3731153865678461,
      "grad_norm": 0.8056939244270325,
      "learning_rate": 1.7512564089547694e-05,
      "loss": 0.1328,
      "step": 7350
    },
    {
      "epoch": 0.3733692065587086,
      "grad_norm": 0.5511149168014526,
      "learning_rate": 1.7510871956275277e-05,
      "loss": 0.1121,
      "step": 7355
    },
    {
      "epoch": 0.373623026549571,
      "grad_norm": 0.5154682993888855,
      "learning_rate": 1.750917982300286e-05,
      "loss": 0.132,
      "step": 7360
    },
    {
      "epoch": 0.3738768465404335,
      "grad_norm": 1.9294079542160034,
      "learning_rate": 1.7507487689730444e-05,
      "loss": 0.1322,
      "step": 7365
    },
    {
      "epoch": 0.374130666531296,
      "grad_norm": 0.43836021423339844,
      "learning_rate": 1.7505795556458028e-05,
      "loss": 0.116,
      "step": 7370
    },
    {
      "epoch": 0.3743844865221585,
      "grad_norm": 0.43150243163108826,
      "learning_rate": 1.750410342318561e-05,
      "loss": 0.1197,
      "step": 7375
    },
    {
      "epoch": 0.374638306513021,
      "grad_norm": 0.606220543384552,
      "learning_rate": 1.7502411289913195e-05,
      "loss": 0.1232,
      "step": 7380
    },
    {
      "epoch": 0.3748921265038834,
      "grad_norm": 0.4498575031757355,
      "learning_rate": 1.750071915664078e-05,
      "loss": 0.1182,
      "step": 7385
    },
    {
      "epoch": 0.3751459464947459,
      "grad_norm": 0.5238727331161499,
      "learning_rate": 1.7499027023368362e-05,
      "loss": 0.1433,
      "step": 7390
    },
    {
      "epoch": 0.3753997664856084,
      "grad_norm": 0.5070326328277588,
      "learning_rate": 1.7497334890095945e-05,
      "loss": 0.1309,
      "step": 7395
    },
    {
      "epoch": 0.3756535864764709,
      "grad_norm": 0.4144206643104553,
      "learning_rate": 1.749564275682353e-05,
      "loss": 0.1259,
      "step": 7400
    },
    {
      "epoch": 0.3759074064673334,
      "grad_norm": 0.476362943649292,
      "learning_rate": 1.7493950623551113e-05,
      "loss": 0.1213,
      "step": 7405
    },
    {
      "epoch": 0.37616122645819583,
      "grad_norm": 0.41985994577407837,
      "learning_rate": 1.7492258490278696e-05,
      "loss": 0.1143,
      "step": 7410
    },
    {
      "epoch": 0.3764150464490583,
      "grad_norm": 0.6084445714950562,
      "learning_rate": 1.749056635700628e-05,
      "loss": 0.1269,
      "step": 7415
    },
    {
      "epoch": 0.3766688664399208,
      "grad_norm": 0.39930278062820435,
      "learning_rate": 1.7488874223733863e-05,
      "loss": 0.1232,
      "step": 7420
    },
    {
      "epoch": 0.3769226864307833,
      "grad_norm": 0.4924406111240387,
      "learning_rate": 1.7487182090461447e-05,
      "loss": 0.124,
      "step": 7425
    },
    {
      "epoch": 0.3771765064216458,
      "grad_norm": 0.5184084177017212,
      "learning_rate": 1.748548995718903e-05,
      "loss": 0.1227,
      "step": 7430
    },
    {
      "epoch": 0.37743032641250823,
      "grad_norm": 0.5747602581977844,
      "learning_rate": 1.7483797823916614e-05,
      "loss": 0.1309,
      "step": 7435
    },
    {
      "epoch": 0.3776841464033707,
      "grad_norm": 0.6429848670959473,
      "learning_rate": 1.7482105690644194e-05,
      "loss": 0.1202,
      "step": 7440
    },
    {
      "epoch": 0.3779379663942332,
      "grad_norm": 0.6303040981292725,
      "learning_rate": 1.748041355737178e-05,
      "loss": 0.119,
      "step": 7445
    },
    {
      "epoch": 0.3781917863850957,
      "grad_norm": 0.8696273565292358,
      "learning_rate": 1.7478721424099364e-05,
      "loss": 0.1191,
      "step": 7450
    },
    {
      "epoch": 0.37844560637595814,
      "grad_norm": 0.7249330878257751,
      "learning_rate": 1.7477029290826948e-05,
      "loss": 0.1285,
      "step": 7455
    },
    {
      "epoch": 0.37869942636682064,
      "grad_norm": 0.37654367089271545,
      "learning_rate": 1.747533715755453e-05,
      "loss": 0.1287,
      "step": 7460
    },
    {
      "epoch": 0.3789532463576831,
      "grad_norm": 0.3263428807258606,
      "learning_rate": 1.747364502428211e-05,
      "loss": 0.109,
      "step": 7465
    },
    {
      "epoch": 0.3792070663485456,
      "grad_norm": 0.5799312591552734,
      "learning_rate": 1.74719528910097e-05,
      "loss": 0.1226,
      "step": 7470
    },
    {
      "epoch": 0.3794608863394081,
      "grad_norm": 0.47346293926239014,
      "learning_rate": 1.7470260757737282e-05,
      "loss": 0.1142,
      "step": 7475
    },
    {
      "epoch": 0.37971470633027055,
      "grad_norm": 0.44059112668037415,
      "learning_rate": 1.7468568624464862e-05,
      "loss": 0.1104,
      "step": 7480
    },
    {
      "epoch": 0.37996852632113304,
      "grad_norm": 0.5601668357849121,
      "learning_rate": 1.746687649119245e-05,
      "loss": 0.1318,
      "step": 7485
    },
    {
      "epoch": 0.38022234631199553,
      "grad_norm": 0.6741954684257507,
      "learning_rate": 1.746518435792003e-05,
      "loss": 0.123,
      "step": 7490
    },
    {
      "epoch": 0.380476166302858,
      "grad_norm": 0.7247961759567261,
      "learning_rate": 1.7463492224647616e-05,
      "loss": 0.1275,
      "step": 7495
    },
    {
      "epoch": 0.3807299862937205,
      "grad_norm": 0.4358140826225281,
      "learning_rate": 1.74618000913752e-05,
      "loss": 0.1214,
      "step": 7500
    },
    {
      "epoch": 0.38098380628458295,
      "grad_norm": 0.35858985781669617,
      "learning_rate": 1.746010795810278e-05,
      "loss": 0.1377,
      "step": 7505
    },
    {
      "epoch": 0.38123762627544544,
      "grad_norm": 0.5448585152626038,
      "learning_rate": 1.7458415824830367e-05,
      "loss": 0.1244,
      "step": 7510
    },
    {
      "epoch": 0.38149144626630793,
      "grad_norm": 0.6085044145584106,
      "learning_rate": 1.7456723691557947e-05,
      "loss": 0.1137,
      "step": 7515
    },
    {
      "epoch": 0.3817452662571704,
      "grad_norm": 0.43685221672058105,
      "learning_rate": 1.745503155828553e-05,
      "loss": 0.1193,
      "step": 7520
    },
    {
      "epoch": 0.3819990862480329,
      "grad_norm": 0.4632774293422699,
      "learning_rate": 1.7453339425013118e-05,
      "loss": 0.1253,
      "step": 7525
    },
    {
      "epoch": 0.38225290623889535,
      "grad_norm": 0.67642742395401,
      "learning_rate": 1.7451647291740698e-05,
      "loss": 0.1173,
      "step": 7530
    },
    {
      "epoch": 0.38250672622975784,
      "grad_norm": 0.5087795853614807,
      "learning_rate": 1.744995515846828e-05,
      "loss": 0.1271,
      "step": 7535
    },
    {
      "epoch": 0.38276054622062033,
      "grad_norm": 0.5182584524154663,
      "learning_rate": 1.7448263025195865e-05,
      "loss": 0.1189,
      "step": 7540
    },
    {
      "epoch": 0.3830143662114828,
      "grad_norm": 0.6090805530548096,
      "learning_rate": 1.744657089192345e-05,
      "loss": 0.1135,
      "step": 7545
    },
    {
      "epoch": 0.3832681862023453,
      "grad_norm": 0.558914065361023,
      "learning_rate": 1.7444878758651035e-05,
      "loss": 0.1275,
      "step": 7550
    },
    {
      "epoch": 0.38352200619320775,
      "grad_norm": 0.4196544885635376,
      "learning_rate": 1.7443186625378615e-05,
      "loss": 0.1208,
      "step": 7555
    },
    {
      "epoch": 0.38377582618407025,
      "grad_norm": 0.6540439128875732,
      "learning_rate": 1.74414944921062e-05,
      "loss": 0.1332,
      "step": 7560
    },
    {
      "epoch": 0.38402964617493274,
      "grad_norm": 0.9480463266372681,
      "learning_rate": 1.7439802358833783e-05,
      "loss": 0.1152,
      "step": 7565
    },
    {
      "epoch": 0.38428346616579523,
      "grad_norm": 0.68366539478302,
      "learning_rate": 1.7438110225561366e-05,
      "loss": 0.1121,
      "step": 7570
    },
    {
      "epoch": 0.3845372861566577,
      "grad_norm": 0.7604976296424866,
      "learning_rate": 1.743641809228895e-05,
      "loss": 0.1317,
      "step": 7575
    },
    {
      "epoch": 0.38479110614752016,
      "grad_norm": 0.704014003276825,
      "learning_rate": 1.7434725959016533e-05,
      "loss": 0.1103,
      "step": 7580
    },
    {
      "epoch": 0.38504492613838265,
      "grad_norm": 0.5294590592384338,
      "learning_rate": 1.7433033825744117e-05,
      "loss": 0.1276,
      "step": 7585
    },
    {
      "epoch": 0.38529874612924514,
      "grad_norm": 0.3316403925418854,
      "learning_rate": 1.74313416924717e-05,
      "loss": 0.1062,
      "step": 7590
    },
    {
      "epoch": 0.38555256612010763,
      "grad_norm": 0.48089930415153503,
      "learning_rate": 1.7429649559199284e-05,
      "loss": 0.114,
      "step": 7595
    },
    {
      "epoch": 0.3858063861109701,
      "grad_norm": 0.4831618368625641,
      "learning_rate": 1.7427957425926867e-05,
      "loss": 0.1184,
      "step": 7600
    },
    {
      "epoch": 0.38606020610183256,
      "grad_norm": 0.49979323148727417,
      "learning_rate": 1.742626529265445e-05,
      "loss": 0.1335,
      "step": 7605
    },
    {
      "epoch": 0.38631402609269505,
      "grad_norm": 1.3292902708053589,
      "learning_rate": 1.7424573159382034e-05,
      "loss": 0.1143,
      "step": 7610
    },
    {
      "epoch": 0.38656784608355754,
      "grad_norm": 0.4008750915527344,
      "learning_rate": 1.7422881026109618e-05,
      "loss": 0.1229,
      "step": 7615
    },
    {
      "epoch": 0.38682166607442003,
      "grad_norm": 0.5050025582313538,
      "learning_rate": 1.74211888928372e-05,
      "loss": 0.1137,
      "step": 7620
    },
    {
      "epoch": 0.3870754860652825,
      "grad_norm": 0.7327058911323547,
      "learning_rate": 1.7419496759564785e-05,
      "loss": 0.1408,
      "step": 7625
    },
    {
      "epoch": 0.38732930605614496,
      "grad_norm": 0.5808429718017578,
      "learning_rate": 1.741780462629237e-05,
      "loss": 0.1055,
      "step": 7630
    },
    {
      "epoch": 0.38758312604700745,
      "grad_norm": 1.0923219919204712,
      "learning_rate": 1.7416112493019952e-05,
      "loss": 0.128,
      "step": 7635
    },
    {
      "epoch": 0.38783694603786995,
      "grad_norm": 0.6931188106536865,
      "learning_rate": 1.7414420359747536e-05,
      "loss": 0.119,
      "step": 7640
    },
    {
      "epoch": 0.38809076602873244,
      "grad_norm": 0.7696263790130615,
      "learning_rate": 1.741272822647512e-05,
      "loss": 0.1138,
      "step": 7645
    },
    {
      "epoch": 0.38834458601959493,
      "grad_norm": 0.5164752006530762,
      "learning_rate": 1.7411036093202703e-05,
      "loss": 0.1151,
      "step": 7650
    },
    {
      "epoch": 0.38859840601045736,
      "grad_norm": 0.4094414710998535,
      "learning_rate": 1.7409343959930286e-05,
      "loss": 0.1241,
      "step": 7655
    },
    {
      "epoch": 0.38885222600131986,
      "grad_norm": 0.43349120020866394,
      "learning_rate": 1.740765182665787e-05,
      "loss": 0.1269,
      "step": 7660
    },
    {
      "epoch": 0.38910604599218235,
      "grad_norm": 0.5583718419075012,
      "learning_rate": 1.7405959693385453e-05,
      "loss": 0.1239,
      "step": 7665
    },
    {
      "epoch": 0.38935986598304484,
      "grad_norm": 0.45432811975479126,
      "learning_rate": 1.7404267560113034e-05,
      "loss": 0.1353,
      "step": 7670
    },
    {
      "epoch": 0.38961368597390733,
      "grad_norm": 0.4887847900390625,
      "learning_rate": 1.740257542684062e-05,
      "loss": 0.1189,
      "step": 7675
    },
    {
      "epoch": 0.38986750596476977,
      "grad_norm": 0.6936730146408081,
      "learning_rate": 1.7400883293568204e-05,
      "loss": 0.1206,
      "step": 7680
    },
    {
      "epoch": 0.39012132595563226,
      "grad_norm": 0.49213695526123047,
      "learning_rate": 1.7399191160295784e-05,
      "loss": 0.1092,
      "step": 7685
    },
    {
      "epoch": 0.39037514594649475,
      "grad_norm": 0.38413718342781067,
      "learning_rate": 1.739749902702337e-05,
      "loss": 0.1195,
      "step": 7690
    },
    {
      "epoch": 0.39062896593735724,
      "grad_norm": 0.5132599472999573,
      "learning_rate": 1.739580689375095e-05,
      "loss": 0.1336,
      "step": 7695
    },
    {
      "epoch": 0.39088278592821973,
      "grad_norm": 0.5477653741836548,
      "learning_rate": 1.7394114760478538e-05,
      "loss": 0.1258,
      "step": 7700
    },
    {
      "epoch": 0.39113660591908217,
      "grad_norm": 0.3821299076080322,
      "learning_rate": 1.7392422627206122e-05,
      "loss": 0.1269,
      "step": 7705
    },
    {
      "epoch": 0.39139042590994466,
      "grad_norm": 0.4275185167789459,
      "learning_rate": 1.7390730493933702e-05,
      "loss": 0.1288,
      "step": 7710
    },
    {
      "epoch": 0.39164424590080715,
      "grad_norm": 0.6078634858131409,
      "learning_rate": 1.738903836066129e-05,
      "loss": 0.1171,
      "step": 7715
    },
    {
      "epoch": 0.39189806589166964,
      "grad_norm": 0.5919456481933594,
      "learning_rate": 1.738734622738887e-05,
      "loss": 0.1187,
      "step": 7720
    },
    {
      "epoch": 0.3921518858825321,
      "grad_norm": 0.3606504797935486,
      "learning_rate": 1.7385654094116453e-05,
      "loss": 0.1196,
      "step": 7725
    },
    {
      "epoch": 0.3924057058733946,
      "grad_norm": 0.5512848496437073,
      "learning_rate": 1.738396196084404e-05,
      "loss": 0.1143,
      "step": 7730
    },
    {
      "epoch": 0.39265952586425706,
      "grad_norm": 0.5730594396591187,
      "learning_rate": 1.738226982757162e-05,
      "loss": 0.1215,
      "step": 7735
    },
    {
      "epoch": 0.39291334585511956,
      "grad_norm": 0.4641408622264862,
      "learning_rate": 1.7380577694299207e-05,
      "loss": 0.117,
      "step": 7740
    },
    {
      "epoch": 0.39316716584598205,
      "grad_norm": 0.417926162481308,
      "learning_rate": 1.7378885561026787e-05,
      "loss": 0.1095,
      "step": 7745
    },
    {
      "epoch": 0.3934209858368445,
      "grad_norm": 0.6021568775177002,
      "learning_rate": 1.737719342775437e-05,
      "loss": 0.1163,
      "step": 7750
    },
    {
      "epoch": 0.393674805827707,
      "grad_norm": 0.5516533851623535,
      "learning_rate": 1.7375501294481957e-05,
      "loss": 0.1147,
      "step": 7755
    },
    {
      "epoch": 0.39392862581856947,
      "grad_norm": 0.49156588315963745,
      "learning_rate": 1.7373809161209537e-05,
      "loss": 0.119,
      "step": 7760
    },
    {
      "epoch": 0.39418244580943196,
      "grad_norm": 0.546325147151947,
      "learning_rate": 1.737211702793712e-05,
      "loss": 0.1385,
      "step": 7765
    },
    {
      "epoch": 0.39443626580029445,
      "grad_norm": 0.4427269399166107,
      "learning_rate": 1.7370424894664704e-05,
      "loss": 0.1237,
      "step": 7770
    },
    {
      "epoch": 0.3946900857911569,
      "grad_norm": 0.42994052171707153,
      "learning_rate": 1.7368732761392288e-05,
      "loss": 0.107,
      "step": 7775
    },
    {
      "epoch": 0.3949439057820194,
      "grad_norm": 0.4181389808654785,
      "learning_rate": 1.736704062811987e-05,
      "loss": 0.0955,
      "step": 7780
    },
    {
      "epoch": 0.39519772577288187,
      "grad_norm": 0.599090576171875,
      "learning_rate": 1.7365348494847455e-05,
      "loss": 0.1237,
      "step": 7785
    },
    {
      "epoch": 0.39545154576374436,
      "grad_norm": 0.5317444801330566,
      "learning_rate": 1.736365636157504e-05,
      "loss": 0.1143,
      "step": 7790
    },
    {
      "epoch": 0.39570536575460685,
      "grad_norm": 0.5095632672309875,
      "learning_rate": 1.7361964228302622e-05,
      "loss": 0.1297,
      "step": 7795
    },
    {
      "epoch": 0.3959591857454693,
      "grad_norm": 0.7343167662620544,
      "learning_rate": 1.7360272095030206e-05,
      "loss": 0.1296,
      "step": 7800
    },
    {
      "epoch": 0.3962130057363318,
      "grad_norm": 0.7114145755767822,
      "learning_rate": 1.735857996175779e-05,
      "loss": 0.1186,
      "step": 7805
    },
    {
      "epoch": 0.39646682572719427,
      "grad_norm": 0.52663254737854,
      "learning_rate": 1.7356887828485373e-05,
      "loss": 0.1318,
      "step": 7810
    },
    {
      "epoch": 0.39672064571805676,
      "grad_norm": 0.33781489729881287,
      "learning_rate": 1.7355195695212956e-05,
      "loss": 0.1075,
      "step": 7815
    },
    {
      "epoch": 0.39697446570891926,
      "grad_norm": 0.8520582914352417,
      "learning_rate": 1.735350356194054e-05,
      "loss": 0.1175,
      "step": 7820
    },
    {
      "epoch": 0.3972282856997817,
      "grad_norm": 0.448352187871933,
      "learning_rate": 1.7351811428668123e-05,
      "loss": 0.1189,
      "step": 7825
    },
    {
      "epoch": 0.3974821056906442,
      "grad_norm": 1.0264869928359985,
      "learning_rate": 1.7350119295395707e-05,
      "loss": 0.1208,
      "step": 7830
    },
    {
      "epoch": 0.3977359256815067,
      "grad_norm": 0.4782174527645111,
      "learning_rate": 1.734842716212329e-05,
      "loss": 0.1211,
      "step": 7835
    },
    {
      "epoch": 0.39798974567236917,
      "grad_norm": 0.41531625390052795,
      "learning_rate": 1.7346735028850874e-05,
      "loss": 0.1131,
      "step": 7840
    },
    {
      "epoch": 0.39824356566323166,
      "grad_norm": 0.5510904788970947,
      "learning_rate": 1.7345042895578458e-05,
      "loss": 0.1068,
      "step": 7845
    },
    {
      "epoch": 0.3984973856540941,
      "grad_norm": 0.6751682758331299,
      "learning_rate": 1.734335076230604e-05,
      "loss": 0.1381,
      "step": 7850
    },
    {
      "epoch": 0.3987512056449566,
      "grad_norm": 0.47615987062454224,
      "learning_rate": 1.7341658629033625e-05,
      "loss": 0.1038,
      "step": 7855
    },
    {
      "epoch": 0.3990050256358191,
      "grad_norm": 0.5570587515830994,
      "learning_rate": 1.7339966495761208e-05,
      "loss": 0.111,
      "step": 7860
    },
    {
      "epoch": 0.39925884562668157,
      "grad_norm": 0.5412005186080933,
      "learning_rate": 1.7338274362488792e-05,
      "loss": 0.1264,
      "step": 7865
    },
    {
      "epoch": 0.39951266561754406,
      "grad_norm": 1.4431235790252686,
      "learning_rate": 1.7336582229216375e-05,
      "loss": 0.1239,
      "step": 7870
    },
    {
      "epoch": 0.3997664856084065,
      "grad_norm": 0.3393210768699646,
      "learning_rate": 1.7334890095943956e-05,
      "loss": 0.1383,
      "step": 7875
    },
    {
      "epoch": 0.400020305599269,
      "grad_norm": 0.5208285450935364,
      "learning_rate": 1.7333197962671542e-05,
      "loss": 0.1172,
      "step": 7880
    },
    {
      "epoch": 0.4002741255901315,
      "grad_norm": 0.5236523747444153,
      "learning_rate": 1.7331505829399126e-05,
      "loss": 0.1098,
      "step": 7885
    },
    {
      "epoch": 0.40052794558099397,
      "grad_norm": 0.6051152944564819,
      "learning_rate": 1.732981369612671e-05,
      "loss": 0.1165,
      "step": 7890
    },
    {
      "epoch": 0.40078176557185646,
      "grad_norm": 0.8865431547164917,
      "learning_rate": 1.7328121562854293e-05,
      "loss": 0.1145,
      "step": 7895
    },
    {
      "epoch": 0.4010355855627189,
      "grad_norm": 0.5429019927978516,
      "learning_rate": 1.7326429429581873e-05,
      "loss": 0.1297,
      "step": 7900
    },
    {
      "epoch": 0.4012894055535814,
      "grad_norm": 0.7259514331817627,
      "learning_rate": 1.732473729630946e-05,
      "loss": 0.1103,
      "step": 7905
    },
    {
      "epoch": 0.4015432255444439,
      "grad_norm": 0.6104267239570618,
      "learning_rate": 1.7323045163037044e-05,
      "loss": 0.1066,
      "step": 7910
    },
    {
      "epoch": 0.4017970455353064,
      "grad_norm": 0.4790496230125427,
      "learning_rate": 1.7321353029764624e-05,
      "loss": 0.1265,
      "step": 7915
    },
    {
      "epoch": 0.40205086552616887,
      "grad_norm": 0.39838600158691406,
      "learning_rate": 1.731966089649221e-05,
      "loss": 0.105,
      "step": 7920
    },
    {
      "epoch": 0.4023046855170313,
      "grad_norm": 0.44898778200149536,
      "learning_rate": 1.731796876321979e-05,
      "loss": 0.1256,
      "step": 7925
    },
    {
      "epoch": 0.4025585055078938,
      "grad_norm": 1.1361500024795532,
      "learning_rate": 1.7316276629947375e-05,
      "loss": 0.1154,
      "step": 7930
    },
    {
      "epoch": 0.4028123254987563,
      "grad_norm": 0.44268670678138733,
      "learning_rate": 1.731458449667496e-05,
      "loss": 0.122,
      "step": 7935
    },
    {
      "epoch": 0.4030661454896188,
      "grad_norm": 0.5239094495773315,
      "learning_rate": 1.731289236340254e-05,
      "loss": 0.1109,
      "step": 7940
    },
    {
      "epoch": 0.40331996548048127,
      "grad_norm": 0.46446603536605835,
      "learning_rate": 1.731120023013013e-05,
      "loss": 0.1077,
      "step": 7945
    },
    {
      "epoch": 0.4035737854713437,
      "grad_norm": 0.36538025736808777,
      "learning_rate": 1.730950809685771e-05,
      "loss": 0.1189,
      "step": 7950
    },
    {
      "epoch": 0.4038276054622062,
      "grad_norm": 0.3762917220592499,
      "learning_rate": 1.7307815963585292e-05,
      "loss": 0.1186,
      "step": 7955
    },
    {
      "epoch": 0.4040814254530687,
      "grad_norm": 0.3559653162956238,
      "learning_rate": 1.730612383031288e-05,
      "loss": 0.1242,
      "step": 7960
    },
    {
      "epoch": 0.4043352454439312,
      "grad_norm": 0.5034398436546326,
      "learning_rate": 1.730443169704046e-05,
      "loss": 0.1222,
      "step": 7965
    },
    {
      "epoch": 0.40458906543479367,
      "grad_norm": 0.8461323976516724,
      "learning_rate": 1.7302739563768043e-05,
      "loss": 0.1281,
      "step": 7970
    },
    {
      "epoch": 0.4048428854256561,
      "grad_norm": 0.4809536337852478,
      "learning_rate": 1.7301047430495626e-05,
      "loss": 0.1085,
      "step": 7975
    },
    {
      "epoch": 0.4050967054165186,
      "grad_norm": 0.5192294716835022,
      "learning_rate": 1.729935529722321e-05,
      "loss": 0.1183,
      "step": 7980
    },
    {
      "epoch": 0.4053505254073811,
      "grad_norm": 0.45198673009872437,
      "learning_rate": 1.7297663163950797e-05,
      "loss": 0.1144,
      "step": 7985
    },
    {
      "epoch": 0.4056043453982436,
      "grad_norm": 0.4005521535873413,
      "learning_rate": 1.7295971030678377e-05,
      "loss": 0.1195,
      "step": 7990
    },
    {
      "epoch": 0.4058581653891061,
      "grad_norm": 0.5918262004852295,
      "learning_rate": 1.729427889740596e-05,
      "loss": 0.1212,
      "step": 7995
    },
    {
      "epoch": 0.4061119853799685,
      "grad_norm": 0.46647128462791443,
      "learning_rate": 1.7292586764133544e-05,
      "loss": 0.1101,
      "step": 8000
    },
    {
      "epoch": 0.406365805370831,
      "grad_norm": 0.43062543869018555,
      "learning_rate": 1.7290894630861128e-05,
      "loss": 0.1174,
      "step": 8005
    },
    {
      "epoch": 0.4066196253616935,
      "grad_norm": 0.4810847043991089,
      "learning_rate": 1.728920249758871e-05,
      "loss": 0.1212,
      "step": 8010
    },
    {
      "epoch": 0.406873445352556,
      "grad_norm": 0.5878148078918457,
      "learning_rate": 1.7287510364316295e-05,
      "loss": 0.1138,
      "step": 8015
    },
    {
      "epoch": 0.4071272653434184,
      "grad_norm": 0.5232487916946411,
      "learning_rate": 1.728581823104388e-05,
      "loss": 0.1264,
      "step": 8020
    },
    {
      "epoch": 0.4073810853342809,
      "grad_norm": 0.3923046588897705,
      "learning_rate": 1.7284126097771462e-05,
      "loss": 0.1137,
      "step": 8025
    },
    {
      "epoch": 0.4076349053251434,
      "grad_norm": 0.5523611903190613,
      "learning_rate": 1.7282433964499045e-05,
      "loss": 0.1269,
      "step": 8030
    },
    {
      "epoch": 0.4078887253160059,
      "grad_norm": 0.587213933467865,
      "learning_rate": 1.728074183122663e-05,
      "loss": 0.1182,
      "step": 8035
    },
    {
      "epoch": 0.4081425453068684,
      "grad_norm": 0.6476024985313416,
      "learning_rate": 1.7279049697954213e-05,
      "loss": 0.1283,
      "step": 8040
    },
    {
      "epoch": 0.4083963652977308,
      "grad_norm": 0.5277358889579773,
      "learning_rate": 1.7277357564681796e-05,
      "loss": 0.1178,
      "step": 8045
    },
    {
      "epoch": 0.4086501852885933,
      "grad_norm": 0.4592641294002533,
      "learning_rate": 1.727566543140938e-05,
      "loss": 0.1155,
      "step": 8050
    },
    {
      "epoch": 0.4089040052794558,
      "grad_norm": 0.5247008800506592,
      "learning_rate": 1.7273973298136963e-05,
      "loss": 0.1086,
      "step": 8055
    },
    {
      "epoch": 0.4091578252703183,
      "grad_norm": 0.5143386125564575,
      "learning_rate": 1.7272281164864547e-05,
      "loss": 0.1229,
      "step": 8060
    },
    {
      "epoch": 0.4094116452611808,
      "grad_norm": 0.40900567173957825,
      "learning_rate": 1.727058903159213e-05,
      "loss": 0.1165,
      "step": 8065
    },
    {
      "epoch": 0.4096654652520432,
      "grad_norm": 0.44880881905555725,
      "learning_rate": 1.7268896898319714e-05,
      "loss": 0.1245,
      "step": 8070
    },
    {
      "epoch": 0.4099192852429057,
      "grad_norm": 0.4040411710739136,
      "learning_rate": 1.7267204765047297e-05,
      "loss": 0.1625,
      "step": 8075
    },
    {
      "epoch": 0.4101731052337682,
      "grad_norm": 0.5253224968910217,
      "learning_rate": 1.7265512631774877e-05,
      "loss": 0.1105,
      "step": 8080
    },
    {
      "epoch": 0.4104269252246307,
      "grad_norm": 0.5281549692153931,
      "learning_rate": 1.7263820498502464e-05,
      "loss": 0.1205,
      "step": 8085
    },
    {
      "epoch": 0.4106807452154932,
      "grad_norm": 0.4668600857257843,
      "learning_rate": 1.7262128365230048e-05,
      "loss": 0.1068,
      "step": 8090
    },
    {
      "epoch": 0.41093456520635563,
      "grad_norm": 0.47776713967323303,
      "learning_rate": 1.726043623195763e-05,
      "loss": 0.1147,
      "step": 8095
    },
    {
      "epoch": 0.4111883851972181,
      "grad_norm": 0.8644548058509827,
      "learning_rate": 1.7258744098685215e-05,
      "loss": 0.1148,
      "step": 8100
    },
    {
      "epoch": 0.4114422051880806,
      "grad_norm": 0.44676366448402405,
      "learning_rate": 1.7257051965412795e-05,
      "loss": 0.1025,
      "step": 8105
    },
    {
      "epoch": 0.4116960251789431,
      "grad_norm": 0.6538465619087219,
      "learning_rate": 1.7255359832140382e-05,
      "loss": 0.1112,
      "step": 8110
    },
    {
      "epoch": 0.4119498451698056,
      "grad_norm": 0.5455487370491028,
      "learning_rate": 1.7253667698867966e-05,
      "loss": 0.1225,
      "step": 8115
    },
    {
      "epoch": 0.41220366516066803,
      "grad_norm": 0.5371622443199158,
      "learning_rate": 1.7251975565595546e-05,
      "loss": 0.1113,
      "step": 8120
    },
    {
      "epoch": 0.4124574851515305,
      "grad_norm": 0.5552743077278137,
      "learning_rate": 1.7250283432323133e-05,
      "loss": 0.1145,
      "step": 8125
    },
    {
      "epoch": 0.412711305142393,
      "grad_norm": 0.40437573194503784,
      "learning_rate": 1.7248591299050713e-05,
      "loss": 0.115,
      "step": 8130
    },
    {
      "epoch": 0.4129651251332555,
      "grad_norm": 0.8257235884666443,
      "learning_rate": 1.72468991657783e-05,
      "loss": 0.1263,
      "step": 8135
    },
    {
      "epoch": 0.413218945124118,
      "grad_norm": 0.5629894733428955,
      "learning_rate": 1.7245207032505883e-05,
      "loss": 0.1114,
      "step": 8140
    },
    {
      "epoch": 0.41347276511498043,
      "grad_norm": 0.5272229909896851,
      "learning_rate": 1.7243514899233464e-05,
      "loss": 0.1197,
      "step": 8145
    },
    {
      "epoch": 0.4137265851058429,
      "grad_norm": 0.5540295243263245,
      "learning_rate": 1.724182276596105e-05,
      "loss": 0.1073,
      "step": 8150
    },
    {
      "epoch": 0.4139804050967054,
      "grad_norm": 0.5846796631813049,
      "learning_rate": 1.724013063268863e-05,
      "loss": 0.1228,
      "step": 8155
    },
    {
      "epoch": 0.4142342250875679,
      "grad_norm": 0.5993126034736633,
      "learning_rate": 1.7238438499416214e-05,
      "loss": 0.1118,
      "step": 8160
    },
    {
      "epoch": 0.4144880450784304,
      "grad_norm": 0.47880518436431885,
      "learning_rate": 1.72367463661438e-05,
      "loss": 0.1389,
      "step": 8165
    },
    {
      "epoch": 0.41474186506929284,
      "grad_norm": 0.7932132482528687,
      "learning_rate": 1.723505423287138e-05,
      "loss": 0.1148,
      "step": 8170
    },
    {
      "epoch": 0.41499568506015533,
      "grad_norm": 0.5608577728271484,
      "learning_rate": 1.7233362099598965e-05,
      "loss": 0.1065,
      "step": 8175
    },
    {
      "epoch": 0.4152495050510178,
      "grad_norm": 0.5518209934234619,
      "learning_rate": 1.723166996632655e-05,
      "loss": 0.1186,
      "step": 8180
    },
    {
      "epoch": 0.4155033250418803,
      "grad_norm": 0.38393184542655945,
      "learning_rate": 1.7229977833054132e-05,
      "loss": 0.112,
      "step": 8185
    },
    {
      "epoch": 0.4157571450327428,
      "grad_norm": 0.5350388884544373,
      "learning_rate": 1.722828569978172e-05,
      "loss": 0.1215,
      "step": 8190
    },
    {
      "epoch": 0.41601096502360524,
      "grad_norm": 0.487732857465744,
      "learning_rate": 1.72265935665093e-05,
      "loss": 0.1232,
      "step": 8195
    },
    {
      "epoch": 0.41626478501446773,
      "grad_norm": 0.7282403707504272,
      "learning_rate": 1.7224901433236883e-05,
      "loss": 0.1112,
      "step": 8200
    },
    {
      "epoch": 0.4165186050053302,
      "grad_norm": 0.49439603090286255,
      "learning_rate": 1.7223209299964466e-05,
      "loss": 0.1123,
      "step": 8205
    },
    {
      "epoch": 0.4167724249961927,
      "grad_norm": 0.3408522307872772,
      "learning_rate": 1.722151716669205e-05,
      "loss": 0.1047,
      "step": 8210
    },
    {
      "epoch": 0.4170262449870552,
      "grad_norm": 0.5345622897148132,
      "learning_rate": 1.7219825033419633e-05,
      "loss": 0.1233,
      "step": 8215
    },
    {
      "epoch": 0.41728006497791764,
      "grad_norm": 0.4255228638648987,
      "learning_rate": 1.7218132900147217e-05,
      "loss": 0.1268,
      "step": 8220
    },
    {
      "epoch": 0.41753388496878013,
      "grad_norm": 0.6682707667350769,
      "learning_rate": 1.72164407668748e-05,
      "loss": 0.1129,
      "step": 8225
    },
    {
      "epoch": 0.4177877049596426,
      "grad_norm": 0.49467936158180237,
      "learning_rate": 1.7214748633602384e-05,
      "loss": 0.1171,
      "step": 8230
    },
    {
      "epoch": 0.4180415249505051,
      "grad_norm": 0.4055912494659424,
      "learning_rate": 1.7213056500329967e-05,
      "loss": 0.117,
      "step": 8235
    },
    {
      "epoch": 0.4182953449413676,
      "grad_norm": 0.47649750113487244,
      "learning_rate": 1.721136436705755e-05,
      "loss": 0.1015,
      "step": 8240
    },
    {
      "epoch": 0.41854916493223004,
      "grad_norm": 0.3205729126930237,
      "learning_rate": 1.7209672233785134e-05,
      "loss": 0.1138,
      "step": 8245
    },
    {
      "epoch": 0.41880298492309254,
      "grad_norm": 0.5603702068328857,
      "learning_rate": 1.7207980100512718e-05,
      "loss": 0.106,
      "step": 8250
    },
    {
      "epoch": 0.419056804913955,
      "grad_norm": 0.6263189911842346,
      "learning_rate": 1.72062879672403e-05,
      "loss": 0.1135,
      "step": 8255
    },
    {
      "epoch": 0.4193106249048175,
      "grad_norm": 0.533976674079895,
      "learning_rate": 1.7204595833967885e-05,
      "loss": 0.1269,
      "step": 8260
    },
    {
      "epoch": 0.41956444489568,
      "grad_norm": 0.38608598709106445,
      "learning_rate": 1.720290370069547e-05,
      "loss": 0.1081,
      "step": 8265
    },
    {
      "epoch": 0.41981826488654245,
      "grad_norm": 0.329917311668396,
      "learning_rate": 1.7201211567423052e-05,
      "loss": 0.1102,
      "step": 8270
    },
    {
      "epoch": 0.42007208487740494,
      "grad_norm": 0.4950959086418152,
      "learning_rate": 1.7199519434150636e-05,
      "loss": 0.1032,
      "step": 8275
    },
    {
      "epoch": 0.42032590486826743,
      "grad_norm": 0.7091923356056213,
      "learning_rate": 1.719782730087822e-05,
      "loss": 0.108,
      "step": 8280
    },
    {
      "epoch": 0.4205797248591299,
      "grad_norm": 0.41781190037727356,
      "learning_rate": 1.7196135167605803e-05,
      "loss": 0.1106,
      "step": 8285
    },
    {
      "epoch": 0.42083354484999236,
      "grad_norm": 0.745145857334137,
      "learning_rate": 1.7194443034333386e-05,
      "loss": 0.1133,
      "step": 8290
    },
    {
      "epoch": 0.42108736484085485,
      "grad_norm": 0.4343501925468445,
      "learning_rate": 1.719275090106097e-05,
      "loss": 0.11,
      "step": 8295
    },
    {
      "epoch": 0.42134118483171734,
      "grad_norm": 0.4626290202140808,
      "learning_rate": 1.7191058767788553e-05,
      "loss": 0.1079,
      "step": 8300
    },
    {
      "epoch": 0.42159500482257983,
      "grad_norm": 0.4289278984069824,
      "learning_rate": 1.7189366634516137e-05,
      "loss": 0.1084,
      "step": 8305
    },
    {
      "epoch": 0.4218488248134423,
      "grad_norm": 0.3560695946216583,
      "learning_rate": 1.7187674501243717e-05,
      "loss": 0.109,
      "step": 8310
    },
    {
      "epoch": 0.42210264480430476,
      "grad_norm": 0.9896416664123535,
      "learning_rate": 1.7185982367971304e-05,
      "loss": 0.1067,
      "step": 8315
    },
    {
      "epoch": 0.42235646479516725,
      "grad_norm": 0.45371013879776,
      "learning_rate": 1.7184290234698888e-05,
      "loss": 0.1093,
      "step": 8320
    },
    {
      "epoch": 0.42261028478602974,
      "grad_norm": 0.38966628909111023,
      "learning_rate": 1.7182598101426468e-05,
      "loss": 0.1037,
      "step": 8325
    },
    {
      "epoch": 0.42286410477689224,
      "grad_norm": 0.6311661005020142,
      "learning_rate": 1.7180905968154055e-05,
      "loss": 0.1128,
      "step": 8330
    },
    {
      "epoch": 0.4231179247677547,
      "grad_norm": 0.36234763264656067,
      "learning_rate": 1.7179213834881635e-05,
      "loss": 0.1075,
      "step": 8335
    },
    {
      "epoch": 0.42337174475861716,
      "grad_norm": 0.4338438808917999,
      "learning_rate": 1.7177521701609222e-05,
      "loss": 0.1137,
      "step": 8340
    },
    {
      "epoch": 0.42362556474947966,
      "grad_norm": 0.603247344493866,
      "learning_rate": 1.7175829568336805e-05,
      "loss": 0.1067,
      "step": 8345
    },
    {
      "epoch": 0.42387938474034215,
      "grad_norm": 0.7704381346702576,
      "learning_rate": 1.7174137435064385e-05,
      "loss": 0.1071,
      "step": 8350
    },
    {
      "epoch": 0.42413320473120464,
      "grad_norm": 0.41250476241111755,
      "learning_rate": 1.7172445301791972e-05,
      "loss": 0.115,
      "step": 8355
    },
    {
      "epoch": 0.42438702472206713,
      "grad_norm": 0.4616197943687439,
      "learning_rate": 1.7170753168519553e-05,
      "loss": 0.112,
      "step": 8360
    },
    {
      "epoch": 0.42464084471292957,
      "grad_norm": 0.45028916001319885,
      "learning_rate": 1.7169061035247136e-05,
      "loss": 0.1158,
      "step": 8365
    },
    {
      "epoch": 0.42489466470379206,
      "grad_norm": 0.6397187113761902,
      "learning_rate": 1.7167368901974723e-05,
      "loss": 0.1111,
      "step": 8370
    },
    {
      "epoch": 0.42514848469465455,
      "grad_norm": 0.4066790044307709,
      "learning_rate": 1.7165676768702303e-05,
      "loss": 0.1185,
      "step": 8375
    },
    {
      "epoch": 0.42540230468551704,
      "grad_norm": 0.7176026701927185,
      "learning_rate": 1.716398463542989e-05,
      "loss": 0.1129,
      "step": 8380
    },
    {
      "epoch": 0.42565612467637953,
      "grad_norm": 0.6438480615615845,
      "learning_rate": 1.716229250215747e-05,
      "loss": 0.1186,
      "step": 8385
    },
    {
      "epoch": 0.42590994466724197,
      "grad_norm": 0.48937782645225525,
      "learning_rate": 1.7160600368885054e-05,
      "loss": 0.1104,
      "step": 8390
    },
    {
      "epoch": 0.42616376465810446,
      "grad_norm": 0.46808022260665894,
      "learning_rate": 1.7158908235612637e-05,
      "loss": 0.11,
      "step": 8395
    },
    {
      "epoch": 0.42641758464896695,
      "grad_norm": 0.5450372099876404,
      "learning_rate": 1.715721610234022e-05,
      "loss": 0.1175,
      "step": 8400
    },
    {
      "epoch": 0.42667140463982944,
      "grad_norm": 0.5416411757469177,
      "learning_rate": 1.7155523969067804e-05,
      "loss": 0.1329,
      "step": 8405
    },
    {
      "epoch": 0.42692522463069194,
      "grad_norm": 0.5366417169570923,
      "learning_rate": 1.7153831835795388e-05,
      "loss": 0.1159,
      "step": 8410
    },
    {
      "epoch": 0.42717904462155437,
      "grad_norm": 0.5638911128044128,
      "learning_rate": 1.715213970252297e-05,
      "loss": 0.1047,
      "step": 8415
    },
    {
      "epoch": 0.42743286461241686,
      "grad_norm": 0.4426100552082062,
      "learning_rate": 1.7150447569250555e-05,
      "loss": 0.1016,
      "step": 8420
    },
    {
      "epoch": 0.42768668460327935,
      "grad_norm": 0.4157399833202362,
      "learning_rate": 1.714875543597814e-05,
      "loss": 0.129,
      "step": 8425
    },
    {
      "epoch": 0.42794050459414185,
      "grad_norm": 0.4836711287498474,
      "learning_rate": 1.7147063302705722e-05,
      "loss": 0.1071,
      "step": 8430
    },
    {
      "epoch": 0.42819432458500434,
      "grad_norm": 0.5332268476486206,
      "learning_rate": 1.7145371169433306e-05,
      "loss": 0.1251,
      "step": 8435
    },
    {
      "epoch": 0.4284481445758668,
      "grad_norm": 0.5566955208778381,
      "learning_rate": 1.714367903616089e-05,
      "loss": 0.1124,
      "step": 8440
    },
    {
      "epoch": 0.42870196456672927,
      "grad_norm": 0.6206413507461548,
      "learning_rate": 1.7141986902888473e-05,
      "loss": 0.1224,
      "step": 8445
    },
    {
      "epoch": 0.42895578455759176,
      "grad_norm": 0.44592681527137756,
      "learning_rate": 1.7140294769616056e-05,
      "loss": 0.1146,
      "step": 8450
    },
    {
      "epoch": 0.42920960454845425,
      "grad_norm": 0.5079742074012756,
      "learning_rate": 1.713860263634364e-05,
      "loss": 0.1242,
      "step": 8455
    },
    {
      "epoch": 0.42946342453931674,
      "grad_norm": 0.41701123118400574,
      "learning_rate": 1.7136910503071223e-05,
      "loss": 0.1157,
      "step": 8460
    },
    {
      "epoch": 0.4297172445301792,
      "grad_norm": 0.4740390479564667,
      "learning_rate": 1.7135218369798807e-05,
      "loss": 0.1171,
      "step": 8465
    },
    {
      "epoch": 0.42997106452104167,
      "grad_norm": 0.378922700881958,
      "learning_rate": 1.713352623652639e-05,
      "loss": 0.1159,
      "step": 8470
    },
    {
      "epoch": 0.43022488451190416,
      "grad_norm": 0.45636868476867676,
      "learning_rate": 1.7131834103253974e-05,
      "loss": 0.1064,
      "step": 8475
    },
    {
      "epoch": 0.43047870450276665,
      "grad_norm": 0.4845265746116638,
      "learning_rate": 1.7130141969981558e-05,
      "loss": 0.1087,
      "step": 8480
    },
    {
      "epoch": 0.43073252449362914,
      "grad_norm": 0.41355660557746887,
      "learning_rate": 1.712844983670914e-05,
      "loss": 0.1154,
      "step": 8485
    },
    {
      "epoch": 0.4309863444844916,
      "grad_norm": 0.42567557096481323,
      "learning_rate": 1.7126757703436725e-05,
      "loss": 0.1131,
      "step": 8490
    },
    {
      "epoch": 0.43124016447535407,
      "grad_norm": 0.46482059359550476,
      "learning_rate": 1.7125065570164308e-05,
      "loss": 0.1068,
      "step": 8495
    },
    {
      "epoch": 0.43149398446621656,
      "grad_norm": 0.7739242911338806,
      "learning_rate": 1.7123373436891892e-05,
      "loss": 0.1175,
      "step": 8500
    },
    {
      "epoch": 0.43174780445707905,
      "grad_norm": 0.4292283058166504,
      "learning_rate": 1.7121681303619475e-05,
      "loss": 0.1197,
      "step": 8505
    },
    {
      "epoch": 0.43200162444794155,
      "grad_norm": 0.4146791696548462,
      "learning_rate": 1.711998917034706e-05,
      "loss": 0.1333,
      "step": 8510
    },
    {
      "epoch": 0.432255444438804,
      "grad_norm": 0.7794067859649658,
      "learning_rate": 1.711829703707464e-05,
      "loss": 0.1176,
      "step": 8515
    },
    {
      "epoch": 0.4325092644296665,
      "grad_norm": 0.4238576889038086,
      "learning_rate": 1.7116604903802226e-05,
      "loss": 0.0959,
      "step": 8520
    },
    {
      "epoch": 0.43276308442052897,
      "grad_norm": 0.5824153423309326,
      "learning_rate": 1.711491277052981e-05,
      "loss": 0.1198,
      "step": 8525
    },
    {
      "epoch": 0.43301690441139146,
      "grad_norm": 0.6489765644073486,
      "learning_rate": 1.7113220637257393e-05,
      "loss": 0.1005,
      "step": 8530
    },
    {
      "epoch": 0.43327072440225395,
      "grad_norm": 0.5258995890617371,
      "learning_rate": 1.7111528503984977e-05,
      "loss": 0.1167,
      "step": 8535
    },
    {
      "epoch": 0.4335245443931164,
      "grad_norm": 0.5935726761817932,
      "learning_rate": 1.7109836370712557e-05,
      "loss": 0.1123,
      "step": 8540
    },
    {
      "epoch": 0.4337783643839789,
      "grad_norm": 0.47369036078453064,
      "learning_rate": 1.7108144237440144e-05,
      "loss": 0.1208,
      "step": 8545
    },
    {
      "epoch": 0.43403218437484137,
      "grad_norm": 0.5065063834190369,
      "learning_rate": 1.7106452104167727e-05,
      "loss": 0.1088,
      "step": 8550
    },
    {
      "epoch": 0.43428600436570386,
      "grad_norm": 0.4577081799507141,
      "learning_rate": 1.7104759970895307e-05,
      "loss": 0.1283,
      "step": 8555
    },
    {
      "epoch": 0.4345398243565663,
      "grad_norm": 0.5575316548347473,
      "learning_rate": 1.7103067837622894e-05,
      "loss": 0.1152,
      "step": 8560
    },
    {
      "epoch": 0.4347936443474288,
      "grad_norm": 0.43118050694465637,
      "learning_rate": 1.7101375704350475e-05,
      "loss": 0.0927,
      "step": 8565
    },
    {
      "epoch": 0.4350474643382913,
      "grad_norm": 0.6905524134635925,
      "learning_rate": 1.7099683571078058e-05,
      "loss": 0.1209,
      "step": 8570
    },
    {
      "epoch": 0.43530128432915377,
      "grad_norm": 0.46179112792015076,
      "learning_rate": 1.709799143780564e-05,
      "loss": 0.1098,
      "step": 8575
    },
    {
      "epoch": 0.43555510432001626,
      "grad_norm": 0.5374900698661804,
      "learning_rate": 1.7096299304533225e-05,
      "loss": 0.0966,
      "step": 8580
    },
    {
      "epoch": 0.4358089243108787,
      "grad_norm": 0.47844305634498596,
      "learning_rate": 1.7094607171260812e-05,
      "loss": 0.1098,
      "step": 8585
    },
    {
      "epoch": 0.4360627443017412,
      "grad_norm": 0.5935022234916687,
      "learning_rate": 1.7092915037988392e-05,
      "loss": 0.099,
      "step": 8590
    },
    {
      "epoch": 0.4363165642926037,
      "grad_norm": 0.39847415685653687,
      "learning_rate": 1.7091222904715976e-05,
      "loss": 0.1041,
      "step": 8595
    },
    {
      "epoch": 0.4365703842834662,
      "grad_norm": 0.6426656246185303,
      "learning_rate": 1.708953077144356e-05,
      "loss": 0.1143,
      "step": 8600
    },
    {
      "epoch": 0.43682420427432866,
      "grad_norm": 0.40940332412719727,
      "learning_rate": 1.7087838638171143e-05,
      "loss": 0.1117,
      "step": 8605
    },
    {
      "epoch": 0.4370780242651911,
      "grad_norm": 0.5274299383163452,
      "learning_rate": 1.7086146504898726e-05,
      "loss": 0.1132,
      "step": 8610
    },
    {
      "epoch": 0.4373318442560536,
      "grad_norm": 0.48877760767936707,
      "learning_rate": 1.708445437162631e-05,
      "loss": 0.114,
      "step": 8615
    },
    {
      "epoch": 0.4375856642469161,
      "grad_norm": 0.4619860053062439,
      "learning_rate": 1.7082762238353894e-05,
      "loss": 0.113,
      "step": 8620
    },
    {
      "epoch": 0.4378394842377786,
      "grad_norm": 0.47805002331733704,
      "learning_rate": 1.7081070105081477e-05,
      "loss": 0.1201,
      "step": 8625
    },
    {
      "epoch": 0.43809330422864107,
      "grad_norm": 0.8779380917549133,
      "learning_rate": 1.707937797180906e-05,
      "loss": 0.1059,
      "step": 8630
    },
    {
      "epoch": 0.4383471242195035,
      "grad_norm": 0.7380253076553345,
      "learning_rate": 1.7077685838536644e-05,
      "loss": 0.1225,
      "step": 8635
    },
    {
      "epoch": 0.438600944210366,
      "grad_norm": 0.4746687710285187,
      "learning_rate": 1.7075993705264228e-05,
      "loss": 0.1119,
      "step": 8640
    },
    {
      "epoch": 0.4388547642012285,
      "grad_norm": 1.3679776191711426,
      "learning_rate": 1.707430157199181e-05,
      "loss": 0.1128,
      "step": 8645
    },
    {
      "epoch": 0.439108584192091,
      "grad_norm": 0.5259309411048889,
      "learning_rate": 1.7072609438719395e-05,
      "loss": 0.128,
      "step": 8650
    },
    {
      "epoch": 0.43936240418295347,
      "grad_norm": 0.4603613018989563,
      "learning_rate": 1.707091730544698e-05,
      "loss": 0.0927,
      "step": 8655
    },
    {
      "epoch": 0.4396162241738159,
      "grad_norm": 0.5576603412628174,
      "learning_rate": 1.7069225172174562e-05,
      "loss": 0.1329,
      "step": 8660
    },
    {
      "epoch": 0.4398700441646784,
      "grad_norm": 0.5329744219779968,
      "learning_rate": 1.7067533038902145e-05,
      "loss": 0.1017,
      "step": 8665
    },
    {
      "epoch": 0.4401238641555409,
      "grad_norm": 0.4557638168334961,
      "learning_rate": 1.706584090562973e-05,
      "loss": 0.1108,
      "step": 8670
    },
    {
      "epoch": 0.4403776841464034,
      "grad_norm": 0.6890199780464172,
      "learning_rate": 1.7064148772357313e-05,
      "loss": 0.1186,
      "step": 8675
    },
    {
      "epoch": 0.4406315041372659,
      "grad_norm": 0.526990532875061,
      "learning_rate": 1.7062456639084896e-05,
      "loss": 0.1048,
      "step": 8680
    },
    {
      "epoch": 0.4408853241281283,
      "grad_norm": 1.4475257396697998,
      "learning_rate": 1.706076450581248e-05,
      "loss": 0.1213,
      "step": 8685
    },
    {
      "epoch": 0.4411391441189908,
      "grad_norm": 0.4177703261375427,
      "learning_rate": 1.7059072372540063e-05,
      "loss": 0.1076,
      "step": 8690
    },
    {
      "epoch": 0.4413929641098533,
      "grad_norm": 1.0303242206573486,
      "learning_rate": 1.7057380239267647e-05,
      "loss": 0.1099,
      "step": 8695
    },
    {
      "epoch": 0.4416467841007158,
      "grad_norm": 0.3975723683834076,
      "learning_rate": 1.705568810599523e-05,
      "loss": 0.1072,
      "step": 8700
    },
    {
      "epoch": 0.4419006040915783,
      "grad_norm": 1.3383222818374634,
      "learning_rate": 1.7053995972722814e-05,
      "loss": 0.1233,
      "step": 8705
    },
    {
      "epoch": 0.4421544240824407,
      "grad_norm": 1.025421142578125,
      "learning_rate": 1.7052303839450397e-05,
      "loss": 0.1101,
      "step": 8710
    },
    {
      "epoch": 0.4424082440733032,
      "grad_norm": 0.33774542808532715,
      "learning_rate": 1.705061170617798e-05,
      "loss": 0.1008,
      "step": 8715
    },
    {
      "epoch": 0.4426620640641657,
      "grad_norm": 0.5173442959785461,
      "learning_rate": 1.7048919572905564e-05,
      "loss": 0.1063,
      "step": 8720
    },
    {
      "epoch": 0.4429158840550282,
      "grad_norm": 0.46076419949531555,
      "learning_rate": 1.7047227439633148e-05,
      "loss": 0.1121,
      "step": 8725
    },
    {
      "epoch": 0.4431697040458907,
      "grad_norm": 0.5615339875221252,
      "learning_rate": 1.704553530636073e-05,
      "loss": 0.1144,
      "step": 8730
    },
    {
      "epoch": 0.4434235240367531,
      "grad_norm": 0.513152539730072,
      "learning_rate": 1.7043843173088315e-05,
      "loss": 0.1111,
      "step": 8735
    },
    {
      "epoch": 0.4436773440276156,
      "grad_norm": 0.6728489995002747,
      "learning_rate": 1.70421510398159e-05,
      "loss": 0.1082,
      "step": 8740
    },
    {
      "epoch": 0.4439311640184781,
      "grad_norm": 0.4770069122314453,
      "learning_rate": 1.704045890654348e-05,
      "loss": 0.1126,
      "step": 8745
    },
    {
      "epoch": 0.4441849840093406,
      "grad_norm": 0.3625965416431427,
      "learning_rate": 1.7038766773271066e-05,
      "loss": 0.1202,
      "step": 8750
    },
    {
      "epoch": 0.4444388040002031,
      "grad_norm": 0.579849123954773,
      "learning_rate": 1.703707463999865e-05,
      "loss": 0.116,
      "step": 8755
    },
    {
      "epoch": 0.4446926239910655,
      "grad_norm": 0.5784683227539062,
      "learning_rate": 1.703538250672623e-05,
      "loss": 0.1351,
      "step": 8760
    },
    {
      "epoch": 0.444946443981928,
      "grad_norm": 0.5647311210632324,
      "learning_rate": 1.7033690373453816e-05,
      "loss": 0.101,
      "step": 8765
    },
    {
      "epoch": 0.4452002639727905,
      "grad_norm": 0.49872612953186035,
      "learning_rate": 1.7031998240181396e-05,
      "loss": 0.1206,
      "step": 8770
    },
    {
      "epoch": 0.445454083963653,
      "grad_norm": 0.6362616419792175,
      "learning_rate": 1.7030306106908983e-05,
      "loss": 0.1242,
      "step": 8775
    },
    {
      "epoch": 0.4457079039545155,
      "grad_norm": 0.34307578206062317,
      "learning_rate": 1.7028613973636564e-05,
      "loss": 0.1058,
      "step": 8780
    },
    {
      "epoch": 0.4459617239453779,
      "grad_norm": 0.5416290163993835,
      "learning_rate": 1.7026921840364147e-05,
      "loss": 0.1151,
      "step": 8785
    },
    {
      "epoch": 0.4462155439362404,
      "grad_norm": 0.46642163395881653,
      "learning_rate": 1.7025229707091734e-05,
      "loss": 0.1197,
      "step": 8790
    },
    {
      "epoch": 0.4464693639271029,
      "grad_norm": 0.3453334867954254,
      "learning_rate": 1.7023537573819314e-05,
      "loss": 0.107,
      "step": 8795
    },
    {
      "epoch": 0.4467231839179654,
      "grad_norm": 0.47410887479782104,
      "learning_rate": 1.7021845440546898e-05,
      "loss": 0.113,
      "step": 8800
    },
    {
      "epoch": 0.4469770039088279,
      "grad_norm": 1.0129278898239136,
      "learning_rate": 1.702015330727448e-05,
      "loss": 0.1171,
      "step": 8805
    },
    {
      "epoch": 0.4472308238996903,
      "grad_norm": 0.6906533241271973,
      "learning_rate": 1.7018461174002065e-05,
      "loss": 0.1278,
      "step": 8810
    },
    {
      "epoch": 0.4474846438905528,
      "grad_norm": 0.4985661804676056,
      "learning_rate": 1.701676904072965e-05,
      "loss": 0.1209,
      "step": 8815
    },
    {
      "epoch": 0.4477384638814153,
      "grad_norm": 0.4819851219654083,
      "learning_rate": 1.7015076907457232e-05,
      "loss": 0.1233,
      "step": 8820
    },
    {
      "epoch": 0.4479922838722778,
      "grad_norm": 0.49135705828666687,
      "learning_rate": 1.7013384774184815e-05,
      "loss": 0.106,
      "step": 8825
    },
    {
      "epoch": 0.4482461038631403,
      "grad_norm": 0.5526263117790222,
      "learning_rate": 1.70116926409124e-05,
      "loss": 0.1196,
      "step": 8830
    },
    {
      "epoch": 0.4484999238540027,
      "grad_norm": 0.6574402451515198,
      "learning_rate": 1.7010000507639983e-05,
      "loss": 0.1052,
      "step": 8835
    },
    {
      "epoch": 0.4487537438448652,
      "grad_norm": 0.3850298225879669,
      "learning_rate": 1.7008308374367566e-05,
      "loss": 0.099,
      "step": 8840
    },
    {
      "epoch": 0.4490075638357277,
      "grad_norm": 0.5754469037055969,
      "learning_rate": 1.700661624109515e-05,
      "loss": 0.1095,
      "step": 8845
    },
    {
      "epoch": 0.4492613838265902,
      "grad_norm": 0.6519771814346313,
      "learning_rate": 1.7004924107822733e-05,
      "loss": 0.1198,
      "step": 8850
    },
    {
      "epoch": 0.44951520381745264,
      "grad_norm": 0.5165800452232361,
      "learning_rate": 1.7003231974550317e-05,
      "loss": 0.119,
      "step": 8855
    },
    {
      "epoch": 0.4497690238083151,
      "grad_norm": 0.39308440685272217,
      "learning_rate": 1.70015398412779e-05,
      "loss": 0.1108,
      "step": 8860
    },
    {
      "epoch": 0.4500228437991776,
      "grad_norm": 0.3973790407180786,
      "learning_rate": 1.6999847708005484e-05,
      "loss": 0.108,
      "step": 8865
    },
    {
      "epoch": 0.4502766637900401,
      "grad_norm": 0.571454644203186,
      "learning_rate": 1.6998155574733067e-05,
      "loss": 0.1107,
      "step": 8870
    },
    {
      "epoch": 0.4505304837809026,
      "grad_norm": 0.48195260763168335,
      "learning_rate": 1.699646344146065e-05,
      "loss": 0.118,
      "step": 8875
    },
    {
      "epoch": 0.45078430377176504,
      "grad_norm": 0.384822815656662,
      "learning_rate": 1.6994771308188234e-05,
      "loss": 0.1239,
      "step": 8880
    },
    {
      "epoch": 0.45103812376262753,
      "grad_norm": 0.6219851970672607,
      "learning_rate": 1.6993079174915818e-05,
      "loss": 0.1087,
      "step": 8885
    },
    {
      "epoch": 0.45129194375349,
      "grad_norm": 0.5098511576652527,
      "learning_rate": 1.69913870416434e-05,
      "loss": 0.1153,
      "step": 8890
    },
    {
      "epoch": 0.4515457637443525,
      "grad_norm": 0.5047430992126465,
      "learning_rate": 1.6989694908370985e-05,
      "loss": 0.0957,
      "step": 8895
    },
    {
      "epoch": 0.451799583735215,
      "grad_norm": 0.5024707317352295,
      "learning_rate": 1.698800277509857e-05,
      "loss": 0.1006,
      "step": 8900
    },
    {
      "epoch": 0.45205340372607744,
      "grad_norm": 0.40146404504776,
      "learning_rate": 1.6986310641826152e-05,
      "loss": 0.1082,
      "step": 8905
    },
    {
      "epoch": 0.45230722371693993,
      "grad_norm": 1.9382678270339966,
      "learning_rate": 1.6984618508553736e-05,
      "loss": 0.098,
      "step": 8910
    },
    {
      "epoch": 0.4525610437078024,
      "grad_norm": 0.7064502835273743,
      "learning_rate": 1.698292637528132e-05,
      "loss": 0.1107,
      "step": 8915
    },
    {
      "epoch": 0.4528148636986649,
      "grad_norm": 0.4406428635120392,
      "learning_rate": 1.6981234242008903e-05,
      "loss": 0.0956,
      "step": 8920
    },
    {
      "epoch": 0.4530686836895274,
      "grad_norm": 0.39747583866119385,
      "learning_rate": 1.6979542108736486e-05,
      "loss": 0.105,
      "step": 8925
    },
    {
      "epoch": 0.45332250368038984,
      "grad_norm": 0.5317738056182861,
      "learning_rate": 1.697784997546407e-05,
      "loss": 0.1111,
      "step": 8930
    },
    {
      "epoch": 0.45357632367125233,
      "grad_norm": 0.4411675035953522,
      "learning_rate": 1.6976157842191653e-05,
      "loss": 0.1174,
      "step": 8935
    },
    {
      "epoch": 0.4538301436621148,
      "grad_norm": 0.7328295111656189,
      "learning_rate": 1.6974465708919237e-05,
      "loss": 0.1025,
      "step": 8940
    },
    {
      "epoch": 0.4540839636529773,
      "grad_norm": 0.40330472588539124,
      "learning_rate": 1.697277357564682e-05,
      "loss": 0.0993,
      "step": 8945
    },
    {
      "epoch": 0.4543377836438398,
      "grad_norm": 1.0070425271987915,
      "learning_rate": 1.69710814423744e-05,
      "loss": 0.1033,
      "step": 8950
    },
    {
      "epoch": 0.45459160363470225,
      "grad_norm": 0.5507986545562744,
      "learning_rate": 1.6969389309101988e-05,
      "loss": 0.1129,
      "step": 8955
    },
    {
      "epoch": 0.45484542362556474,
      "grad_norm": 0.48115500807762146,
      "learning_rate": 1.6967697175829568e-05,
      "loss": 0.1042,
      "step": 8960
    },
    {
      "epoch": 0.45509924361642723,
      "grad_norm": 0.4493533670902252,
      "learning_rate": 1.696600504255715e-05,
      "loss": 0.1132,
      "step": 8965
    },
    {
      "epoch": 0.4553530636072897,
      "grad_norm": 0.9694324135780334,
      "learning_rate": 1.6964312909284738e-05,
      "loss": 0.1155,
      "step": 8970
    },
    {
      "epoch": 0.4556068835981522,
      "grad_norm": 0.3891000747680664,
      "learning_rate": 1.696262077601232e-05,
      "loss": 0.0998,
      "step": 8975
    },
    {
      "epoch": 0.45586070358901465,
      "grad_norm": 0.4627774953842163,
      "learning_rate": 1.6960928642739905e-05,
      "loss": 0.1009,
      "step": 8980
    },
    {
      "epoch": 0.45611452357987714,
      "grad_norm": 0.5629536509513855,
      "learning_rate": 1.6959236509467485e-05,
      "loss": 0.108,
      "step": 8985
    },
    {
      "epoch": 0.45636834357073963,
      "grad_norm": 0.4632164239883423,
      "learning_rate": 1.695754437619507e-05,
      "loss": 0.113,
      "step": 8990
    },
    {
      "epoch": 0.4566221635616021,
      "grad_norm": 0.5252170562744141,
      "learning_rate": 1.6955852242922656e-05,
      "loss": 0.1114,
      "step": 8995
    },
    {
      "epoch": 0.4568759835524646,
      "grad_norm": 0.7564627528190613,
      "learning_rate": 1.6954160109650236e-05,
      "loss": 0.1086,
      "step": 9000
    },
    {
      "epoch": 0.45712980354332705,
      "grad_norm": 0.44219401478767395,
      "learning_rate": 1.695246797637782e-05,
      "loss": 0.1107,
      "step": 9005
    },
    {
      "epoch": 0.45738362353418954,
      "grad_norm": 0.5671635866165161,
      "learning_rate": 1.6950775843105403e-05,
      "loss": 0.1236,
      "step": 9010
    },
    {
      "epoch": 0.45763744352505203,
      "grad_norm": 0.42195212841033936,
      "learning_rate": 1.6949083709832987e-05,
      "loss": 0.1077,
      "step": 9015
    },
    {
      "epoch": 0.4578912635159145,
      "grad_norm": 0.36243775486946106,
      "learning_rate": 1.6947391576560574e-05,
      "loss": 0.0929,
      "step": 9020
    },
    {
      "epoch": 0.458145083506777,
      "grad_norm": 0.4893646240234375,
      "learning_rate": 1.6945699443288154e-05,
      "loss": 0.1182,
      "step": 9025
    },
    {
      "epoch": 0.45839890349763945,
      "grad_norm": 0.43038901686668396,
      "learning_rate": 1.6944007310015737e-05,
      "loss": 0.1121,
      "step": 9030
    },
    {
      "epoch": 0.45865272348850195,
      "grad_norm": 0.4785013198852539,
      "learning_rate": 1.694231517674332e-05,
      "loss": 0.1063,
      "step": 9035
    },
    {
      "epoch": 0.45890654347936444,
      "grad_norm": 0.4438520073890686,
      "learning_rate": 1.6940623043470904e-05,
      "loss": 0.107,
      "step": 9040
    },
    {
      "epoch": 0.45916036347022693,
      "grad_norm": 0.5882023572921753,
      "learning_rate": 1.6938930910198488e-05,
      "loss": 0.1147,
      "step": 9045
    },
    {
      "epoch": 0.4594141834610894,
      "grad_norm": 0.5411685705184937,
      "learning_rate": 1.693723877692607e-05,
      "loss": 0.1099,
      "step": 9050
    },
    {
      "epoch": 0.45966800345195186,
      "grad_norm": 0.9396799206733704,
      "learning_rate": 1.6935546643653655e-05,
      "loss": 0.1188,
      "step": 9055
    },
    {
      "epoch": 0.45992182344281435,
      "grad_norm": 0.5368686318397522,
      "learning_rate": 1.693385451038124e-05,
      "loss": 0.1063,
      "step": 9060
    },
    {
      "epoch": 0.46017564343367684,
      "grad_norm": 0.5517676472663879,
      "learning_rate": 1.6932162377108822e-05,
      "loss": 0.1013,
      "step": 9065
    },
    {
      "epoch": 0.46042946342453933,
      "grad_norm": 0.45235463976860046,
      "learning_rate": 1.6930470243836406e-05,
      "loss": 0.1123,
      "step": 9070
    },
    {
      "epoch": 0.4606832834154018,
      "grad_norm": 0.5119823217391968,
      "learning_rate": 1.692877811056399e-05,
      "loss": 0.1046,
      "step": 9075
    },
    {
      "epoch": 0.46093710340626426,
      "grad_norm": 0.48481494188308716,
      "learning_rate": 1.6927085977291573e-05,
      "loss": 0.1181,
      "step": 9080
    },
    {
      "epoch": 0.46119092339712675,
      "grad_norm": 0.8900187015533447,
      "learning_rate": 1.6925393844019156e-05,
      "loss": 0.0909,
      "step": 9085
    },
    {
      "epoch": 0.46144474338798924,
      "grad_norm": 0.5890236496925354,
      "learning_rate": 1.692370171074674e-05,
      "loss": 0.1115,
      "step": 9090
    },
    {
      "epoch": 0.46169856337885173,
      "grad_norm": 0.5484874248504639,
      "learning_rate": 1.6922009577474323e-05,
      "loss": 0.1222,
      "step": 9095
    },
    {
      "epoch": 0.4619523833697142,
      "grad_norm": 0.32354220747947693,
      "learning_rate": 1.6920317444201907e-05,
      "loss": 0.0988,
      "step": 9100
    },
    {
      "epoch": 0.46220620336057666,
      "grad_norm": 0.4846709966659546,
      "learning_rate": 1.691862531092949e-05,
      "loss": 0.1184,
      "step": 9105
    },
    {
      "epoch": 0.46246002335143915,
      "grad_norm": 0.43833622336387634,
      "learning_rate": 1.6916933177657074e-05,
      "loss": 0.1041,
      "step": 9110
    },
    {
      "epoch": 0.46271384334230165,
      "grad_norm": 0.5856930613517761,
      "learning_rate": 1.6915241044384658e-05,
      "loss": 0.1109,
      "step": 9115
    },
    {
      "epoch": 0.46296766333316414,
      "grad_norm": 0.6962338089942932,
      "learning_rate": 1.691354891111224e-05,
      "loss": 0.1146,
      "step": 9120
    },
    {
      "epoch": 0.4632214833240266,
      "grad_norm": 0.3463370203971863,
      "learning_rate": 1.6911856777839825e-05,
      "loss": 0.1099,
      "step": 9125
    },
    {
      "epoch": 0.46347530331488906,
      "grad_norm": 1.7089934349060059,
      "learning_rate": 1.6910164644567408e-05,
      "loss": 0.117,
      "step": 9130
    },
    {
      "epoch": 0.46372912330575156,
      "grad_norm": 0.47701790928840637,
      "learning_rate": 1.6908472511294992e-05,
      "loss": 0.1109,
      "step": 9135
    },
    {
      "epoch": 0.46398294329661405,
      "grad_norm": 0.7937628626823425,
      "learning_rate": 1.6906780378022572e-05,
      "loss": 0.1002,
      "step": 9140
    },
    {
      "epoch": 0.46423676328747654,
      "grad_norm": 0.608931303024292,
      "learning_rate": 1.690508824475016e-05,
      "loss": 0.1032,
      "step": 9145
    },
    {
      "epoch": 0.464490583278339,
      "grad_norm": 0.5773627161979675,
      "learning_rate": 1.6903396111477742e-05,
      "loss": 0.1204,
      "step": 9150
    },
    {
      "epoch": 0.46474440326920147,
      "grad_norm": 0.48551470041275024,
      "learning_rate": 1.6901703978205323e-05,
      "loss": 0.1023,
      "step": 9155
    },
    {
      "epoch": 0.46499822326006396,
      "grad_norm": 0.428069144487381,
      "learning_rate": 1.690001184493291e-05,
      "loss": 0.1174,
      "step": 9160
    },
    {
      "epoch": 0.46525204325092645,
      "grad_norm": 0.737295925617218,
      "learning_rate": 1.689831971166049e-05,
      "loss": 0.1024,
      "step": 9165
    },
    {
      "epoch": 0.46550586324178894,
      "grad_norm": 0.9198458194732666,
      "learning_rate": 1.6896627578388077e-05,
      "loss": 0.1045,
      "step": 9170
    },
    {
      "epoch": 0.4657596832326514,
      "grad_norm": 0.46820440888404846,
      "learning_rate": 1.689493544511566e-05,
      "loss": 0.105,
      "step": 9175
    },
    {
      "epoch": 0.46601350322351387,
      "grad_norm": 0.5066524147987366,
      "learning_rate": 1.689324331184324e-05,
      "loss": 0.0928,
      "step": 9180
    },
    {
      "epoch": 0.46626732321437636,
      "grad_norm": 0.45925962924957275,
      "learning_rate": 1.6891551178570827e-05,
      "loss": 0.1143,
      "step": 9185
    },
    {
      "epoch": 0.46652114320523885,
      "grad_norm": 0.5800960063934326,
      "learning_rate": 1.6889859045298407e-05,
      "loss": 0.1292,
      "step": 9190
    },
    {
      "epoch": 0.46677496319610134,
      "grad_norm": 0.47480955719947815,
      "learning_rate": 1.688816691202599e-05,
      "loss": 0.1098,
      "step": 9195
    },
    {
      "epoch": 0.4670287831869638,
      "grad_norm": 0.5127950310707092,
      "learning_rate": 1.6886474778753578e-05,
      "loss": 0.1229,
      "step": 9200
    },
    {
      "epoch": 0.4672826031778263,
      "grad_norm": 0.6397584676742554,
      "learning_rate": 1.6884782645481158e-05,
      "loss": 0.1075,
      "step": 9205
    },
    {
      "epoch": 0.46753642316868876,
      "grad_norm": 0.419527530670166,
      "learning_rate": 1.688309051220874e-05,
      "loss": 0.1033,
      "step": 9210
    },
    {
      "epoch": 0.46779024315955126,
      "grad_norm": 0.3961291015148163,
      "learning_rate": 1.6881398378936325e-05,
      "loss": 0.0988,
      "step": 9215
    },
    {
      "epoch": 0.46804406315041375,
      "grad_norm": 0.5905468463897705,
      "learning_rate": 1.687970624566391e-05,
      "loss": 0.1044,
      "step": 9220
    },
    {
      "epoch": 0.4682978831412762,
      "grad_norm": 0.390688419342041,
      "learning_rate": 1.6878014112391496e-05,
      "loss": 0.1111,
      "step": 9225
    },
    {
      "epoch": 0.4685517031321387,
      "grad_norm": 0.36093178391456604,
      "learning_rate": 1.6876321979119076e-05,
      "loss": 0.1002,
      "step": 9230
    },
    {
      "epoch": 0.46880552312300117,
      "grad_norm": 0.5973555445671082,
      "learning_rate": 1.687462984584666e-05,
      "loss": 0.1121,
      "step": 9235
    },
    {
      "epoch": 0.46905934311386366,
      "grad_norm": 0.5031787157058716,
      "learning_rate": 1.6872937712574243e-05,
      "loss": 0.1064,
      "step": 9240
    },
    {
      "epoch": 0.46931316310472615,
      "grad_norm": 0.844380259513855,
      "learning_rate": 1.6871245579301826e-05,
      "loss": 0.1042,
      "step": 9245
    },
    {
      "epoch": 0.4695669830955886,
      "grad_norm": 0.5998194217681885,
      "learning_rate": 1.686955344602941e-05,
      "loss": 0.1112,
      "step": 9250
    },
    {
      "epoch": 0.4698208030864511,
      "grad_norm": 0.5319337248802185,
      "learning_rate": 1.6867861312756994e-05,
      "loss": 0.1065,
      "step": 9255
    },
    {
      "epoch": 0.47007462307731357,
      "grad_norm": 0.43366721272468567,
      "learning_rate": 1.6866169179484577e-05,
      "loss": 0.1047,
      "step": 9260
    },
    {
      "epoch": 0.47032844306817606,
      "grad_norm": 0.4964733421802521,
      "learning_rate": 1.686447704621216e-05,
      "loss": 0.1027,
      "step": 9265
    },
    {
      "epoch": 0.47058226305903855,
      "grad_norm": 0.44293656945228577,
      "learning_rate": 1.6862784912939744e-05,
      "loss": 0.112,
      "step": 9270
    },
    {
      "epoch": 0.470836083049901,
      "grad_norm": 0.2671841084957123,
      "learning_rate": 1.6861092779667328e-05,
      "loss": 0.1019,
      "step": 9275
    },
    {
      "epoch": 0.4710899030407635,
      "grad_norm": 0.5117993354797363,
      "learning_rate": 1.685940064639491e-05,
      "loss": 0.1069,
      "step": 9280
    },
    {
      "epoch": 0.47134372303162597,
      "grad_norm": 0.49542441964149475,
      "learning_rate": 1.6857708513122495e-05,
      "loss": 0.1059,
      "step": 9285
    },
    {
      "epoch": 0.47159754302248846,
      "grad_norm": 0.4893092215061188,
      "learning_rate": 1.685601637985008e-05,
      "loss": 0.1054,
      "step": 9290
    },
    {
      "epoch": 0.47185136301335096,
      "grad_norm": 1.8522083759307861,
      "learning_rate": 1.6854324246577662e-05,
      "loss": 0.1002,
      "step": 9295
    },
    {
      "epoch": 0.4721051830042134,
      "grad_norm": 1.9006584882736206,
      "learning_rate": 1.6852632113305245e-05,
      "loss": 0.113,
      "step": 9300
    },
    {
      "epoch": 0.4723590029950759,
      "grad_norm": 0.43772315979003906,
      "learning_rate": 1.685093998003283e-05,
      "loss": 0.1172,
      "step": 9305
    },
    {
      "epoch": 0.4726128229859384,
      "grad_norm": 0.5562046766281128,
      "learning_rate": 1.6849247846760412e-05,
      "loss": 0.1037,
      "step": 9310
    },
    {
      "epoch": 0.47286664297680087,
      "grad_norm": 0.4431091248989105,
      "learning_rate": 1.6847555713487996e-05,
      "loss": 0.102,
      "step": 9315
    },
    {
      "epoch": 0.47312046296766336,
      "grad_norm": 0.4664026200771332,
      "learning_rate": 1.684586358021558e-05,
      "loss": 0.1014,
      "step": 9320
    },
    {
      "epoch": 0.4733742829585258,
      "grad_norm": 0.4194331765174866,
      "learning_rate": 1.6844171446943163e-05,
      "loss": 0.1094,
      "step": 9325
    },
    {
      "epoch": 0.4736281029493883,
      "grad_norm": 0.4214523732662201,
      "learning_rate": 1.6842479313670747e-05,
      "loss": 0.0985,
      "step": 9330
    },
    {
      "epoch": 0.4738819229402508,
      "grad_norm": 0.4564267098903656,
      "learning_rate": 1.684078718039833e-05,
      "loss": 0.0984,
      "step": 9335
    },
    {
      "epoch": 0.47413574293111327,
      "grad_norm": 0.6111764907836914,
      "learning_rate": 1.6839095047125914e-05,
      "loss": 0.1138,
      "step": 9340
    },
    {
      "epoch": 0.47438956292197576,
      "grad_norm": 0.8785223364830017,
      "learning_rate": 1.6837402913853494e-05,
      "loss": 0.1066,
      "step": 9345
    },
    {
      "epoch": 0.4746433829128382,
      "grad_norm": 0.5163254737854004,
      "learning_rate": 1.683571078058108e-05,
      "loss": 0.1086,
      "step": 9350
    },
    {
      "epoch": 0.4748972029037007,
      "grad_norm": 0.45714399218559265,
      "learning_rate": 1.6834018647308664e-05,
      "loss": 0.1137,
      "step": 9355
    },
    {
      "epoch": 0.4751510228945632,
      "grad_norm": 0.5146034359931946,
      "learning_rate": 1.6832326514036248e-05,
      "loss": 0.0973,
      "step": 9360
    },
    {
      "epoch": 0.47540484288542567,
      "grad_norm": 0.5823367834091187,
      "learning_rate": 1.683063438076383e-05,
      "loss": 0.1099,
      "step": 9365
    },
    {
      "epoch": 0.47565866287628816,
      "grad_norm": 0.472186416387558,
      "learning_rate": 1.682894224749141e-05,
      "loss": 0.1074,
      "step": 9370
    },
    {
      "epoch": 0.4759124828671506,
      "grad_norm": 0.817765474319458,
      "learning_rate": 1.6827250114219e-05,
      "loss": 0.1037,
      "step": 9375
    },
    {
      "epoch": 0.4761663028580131,
      "grad_norm": 0.43095797300338745,
      "learning_rate": 1.6825557980946582e-05,
      "loss": 0.1081,
      "step": 9380
    },
    {
      "epoch": 0.4764201228488756,
      "grad_norm": 0.40991702675819397,
      "learning_rate": 1.6823865847674162e-05,
      "loss": 0.1172,
      "step": 9385
    },
    {
      "epoch": 0.4766739428397381,
      "grad_norm": 0.39121660590171814,
      "learning_rate": 1.682217371440175e-05,
      "loss": 0.1066,
      "step": 9390
    },
    {
      "epoch": 0.47692776283060057,
      "grad_norm": 0.5421929359436035,
      "learning_rate": 1.682048158112933e-05,
      "loss": 0.0957,
      "step": 9395
    },
    {
      "epoch": 0.477181582821463,
      "grad_norm": 0.44107937812805176,
      "learning_rate": 1.6818789447856913e-05,
      "loss": 0.1129,
      "step": 9400
    },
    {
      "epoch": 0.4774354028123255,
      "grad_norm": 0.6062149405479431,
      "learning_rate": 1.68170973145845e-05,
      "loss": 0.1152,
      "step": 9405
    },
    {
      "epoch": 0.477689222803188,
      "grad_norm": 0.4286916255950928,
      "learning_rate": 1.681540518131208e-05,
      "loss": 0.1114,
      "step": 9410
    },
    {
      "epoch": 0.4779430427940505,
      "grad_norm": 0.42873385548591614,
      "learning_rate": 1.6813713048039667e-05,
      "loss": 0.1024,
      "step": 9415
    },
    {
      "epoch": 0.4781968627849129,
      "grad_norm": 0.5670800805091858,
      "learning_rate": 1.6812020914767247e-05,
      "loss": 0.1174,
      "step": 9420
    },
    {
      "epoch": 0.4784506827757754,
      "grad_norm": 0.4170010983943939,
      "learning_rate": 1.681032878149483e-05,
      "loss": 0.0972,
      "step": 9425
    },
    {
      "epoch": 0.4787045027666379,
      "grad_norm": 0.5890213251113892,
      "learning_rate": 1.6808636648222418e-05,
      "loss": 0.1212,
      "step": 9430
    },
    {
      "epoch": 0.4789583227575004,
      "grad_norm": 0.8447275757789612,
      "learning_rate": 1.6806944514949998e-05,
      "loss": 0.107,
      "step": 9435
    },
    {
      "epoch": 0.4792121427483629,
      "grad_norm": 0.41049903631210327,
      "learning_rate": 1.680525238167758e-05,
      "loss": 0.1117,
      "step": 9440
    },
    {
      "epoch": 0.4794659627392253,
      "grad_norm": 0.5393239855766296,
      "learning_rate": 1.6803560248405165e-05,
      "loss": 0.0978,
      "step": 9445
    },
    {
      "epoch": 0.4797197827300878,
      "grad_norm": 0.5582802295684814,
      "learning_rate": 1.680186811513275e-05,
      "loss": 0.1057,
      "step": 9450
    },
    {
      "epoch": 0.4799736027209503,
      "grad_norm": 0.6471379995346069,
      "learning_rate": 1.6800175981860332e-05,
      "loss": 0.1101,
      "step": 9455
    },
    {
      "epoch": 0.4802274227118128,
      "grad_norm": 0.5534394979476929,
      "learning_rate": 1.6798483848587915e-05,
      "loss": 0.1015,
      "step": 9460
    },
    {
      "epoch": 0.4804812427026753,
      "grad_norm": 0.43192145228385925,
      "learning_rate": 1.67967917153155e-05,
      "loss": 0.102,
      "step": 9465
    },
    {
      "epoch": 0.4807350626935377,
      "grad_norm": 0.38189658522605896,
      "learning_rate": 1.6795099582043083e-05,
      "loss": 0.1017,
      "step": 9470
    },
    {
      "epoch": 0.4809888826844002,
      "grad_norm": 0.4700341820716858,
      "learning_rate": 1.6793407448770666e-05,
      "loss": 0.1005,
      "step": 9475
    },
    {
      "epoch": 0.4812427026752627,
      "grad_norm": 0.6079494953155518,
      "learning_rate": 1.679171531549825e-05,
      "loss": 0.0978,
      "step": 9480
    },
    {
      "epoch": 0.4814965226661252,
      "grad_norm": 0.8428679704666138,
      "learning_rate": 1.6790023182225833e-05,
      "loss": 0.1251,
      "step": 9485
    },
    {
      "epoch": 0.4817503426569877,
      "grad_norm": 0.48023349046707153,
      "learning_rate": 1.6788331048953417e-05,
      "loss": 0.1109,
      "step": 9490
    },
    {
      "epoch": 0.4820041626478501,
      "grad_norm": 0.40779995918273926,
      "learning_rate": 1.6786638915681e-05,
      "loss": 0.1191,
      "step": 9495
    },
    {
      "epoch": 0.4822579826387126,
      "grad_norm": 0.4184429943561554,
      "learning_rate": 1.6784946782408584e-05,
      "loss": 0.0975,
      "step": 9500
    },
    {
      "epoch": 0.4825118026295751,
      "grad_norm": 0.39694687724113464,
      "learning_rate": 1.6783254649136167e-05,
      "loss": 0.0985,
      "step": 9505
    },
    {
      "epoch": 0.4827656226204376,
      "grad_norm": 0.503432035446167,
      "learning_rate": 1.678156251586375e-05,
      "loss": 0.1057,
      "step": 9510
    },
    {
      "epoch": 0.4830194426113001,
      "grad_norm": 0.43741822242736816,
      "learning_rate": 1.6779870382591334e-05,
      "loss": 0.1002,
      "step": 9515
    },
    {
      "epoch": 0.4832732626021625,
      "grad_norm": 0.49856293201446533,
      "learning_rate": 1.6778178249318918e-05,
      "loss": 0.1194,
      "step": 9520
    },
    {
      "epoch": 0.483527082593025,
      "grad_norm": 0.46641266345977783,
      "learning_rate": 1.67764861160465e-05,
      "loss": 0.0864,
      "step": 9525
    },
    {
      "epoch": 0.4837809025838875,
      "grad_norm": 0.4092983603477478,
      "learning_rate": 1.6774793982774085e-05,
      "loss": 0.1078,
      "step": 9530
    },
    {
      "epoch": 0.48403472257475,
      "grad_norm": 0.4319418668746948,
      "learning_rate": 1.677310184950167e-05,
      "loss": 0.1056,
      "step": 9535
    },
    {
      "epoch": 0.4842885425656125,
      "grad_norm": 0.3443421423435211,
      "learning_rate": 1.6771409716229252e-05,
      "loss": 0.098,
      "step": 9540
    },
    {
      "epoch": 0.4845423625564749,
      "grad_norm": 0.5126322507858276,
      "learning_rate": 1.6769717582956836e-05,
      "loss": 0.1052,
      "step": 9545
    },
    {
      "epoch": 0.4847961825473374,
      "grad_norm": 0.4677981436252594,
      "learning_rate": 1.6768025449684416e-05,
      "loss": 0.1123,
      "step": 9550
    },
    {
      "epoch": 0.4850500025381999,
      "grad_norm": 0.5052584409713745,
      "learning_rate": 1.6766333316412003e-05,
      "loss": 0.1084,
      "step": 9555
    },
    {
      "epoch": 0.4853038225290624,
      "grad_norm": 0.5080714821815491,
      "learning_rate": 1.6764641183139586e-05,
      "loss": 0.1052,
      "step": 9560
    },
    {
      "epoch": 0.4855576425199249,
      "grad_norm": 0.4302172064781189,
      "learning_rate": 1.676294904986717e-05,
      "loss": 0.1023,
      "step": 9565
    },
    {
      "epoch": 0.48581146251078733,
      "grad_norm": 0.4493500292301178,
      "learning_rate": 1.6761256916594753e-05,
      "loss": 0.1187,
      "step": 9570
    },
    {
      "epoch": 0.4860652825016498,
      "grad_norm": 0.3389817178249359,
      "learning_rate": 1.6759564783322334e-05,
      "loss": 0.1063,
      "step": 9575
    },
    {
      "epoch": 0.4863191024925123,
      "grad_norm": 0.4281669855117798,
      "learning_rate": 1.675787265004992e-05,
      "loss": 0.1004,
      "step": 9580
    },
    {
      "epoch": 0.4865729224833748,
      "grad_norm": 0.5047438740730286,
      "learning_rate": 1.6756180516777504e-05,
      "loss": 0.1168,
      "step": 9585
    },
    {
      "epoch": 0.4868267424742373,
      "grad_norm": 0.6686415076255798,
      "learning_rate": 1.6754488383505084e-05,
      "loss": 0.1013,
      "step": 9590
    },
    {
      "epoch": 0.48708056246509973,
      "grad_norm": 0.5081450939178467,
      "learning_rate": 1.675279625023267e-05,
      "loss": 0.1007,
      "step": 9595
    },
    {
      "epoch": 0.4873343824559622,
      "grad_norm": 0.5452947616577148,
      "learning_rate": 1.675110411696025e-05,
      "loss": 0.1072,
      "step": 9600
    },
    {
      "epoch": 0.4875882024468247,
      "grad_norm": 0.3936508595943451,
      "learning_rate": 1.6749411983687838e-05,
      "loss": 0.0975,
      "step": 9605
    },
    {
      "epoch": 0.4878420224376872,
      "grad_norm": 0.4514859914779663,
      "learning_rate": 1.6747719850415422e-05,
      "loss": 0.0993,
      "step": 9610
    },
    {
      "epoch": 0.4880958424285497,
      "grad_norm": 0.7531886696815491,
      "learning_rate": 1.6746027717143002e-05,
      "loss": 0.099,
      "step": 9615
    },
    {
      "epoch": 0.48834966241941213,
      "grad_norm": 0.5042106509208679,
      "learning_rate": 1.674433558387059e-05,
      "loss": 0.1125,
      "step": 9620
    },
    {
      "epoch": 0.4886034824102746,
      "grad_norm": 0.6559919714927673,
      "learning_rate": 1.674264345059817e-05,
      "loss": 0.1166,
      "step": 9625
    },
    {
      "epoch": 0.4888573024011371,
      "grad_norm": 0.4015331566333771,
      "learning_rate": 1.6740951317325753e-05,
      "loss": 0.1121,
      "step": 9630
    },
    {
      "epoch": 0.4891111223919996,
      "grad_norm": 0.38791272044181824,
      "learning_rate": 1.673925918405334e-05,
      "loss": 0.1058,
      "step": 9635
    },
    {
      "epoch": 0.4893649423828621,
      "grad_norm": 0.6192878484725952,
      "learning_rate": 1.673756705078092e-05,
      "loss": 0.1029,
      "step": 9640
    },
    {
      "epoch": 0.48961876237372454,
      "grad_norm": 0.3586585819721222,
      "learning_rate": 1.6735874917508503e-05,
      "loss": 0.1003,
      "step": 9645
    },
    {
      "epoch": 0.48987258236458703,
      "grad_norm": 0.5473580956459045,
      "learning_rate": 1.6734182784236087e-05,
      "loss": 0.102,
      "step": 9650
    },
    {
      "epoch": 0.4901264023554495,
      "grad_norm": 0.4406992495059967,
      "learning_rate": 1.673249065096367e-05,
      "loss": 0.1019,
      "step": 9655
    },
    {
      "epoch": 0.490380222346312,
      "grad_norm": 0.6175484657287598,
      "learning_rate": 1.6730798517691257e-05,
      "loss": 0.0987,
      "step": 9660
    },
    {
      "epoch": 0.4906340423371745,
      "grad_norm": 0.5513244271278381,
      "learning_rate": 1.6729106384418837e-05,
      "loss": 0.094,
      "step": 9665
    },
    {
      "epoch": 0.49088786232803694,
      "grad_norm": 0.5684154033660889,
      "learning_rate": 1.672741425114642e-05,
      "loss": 0.0943,
      "step": 9670
    },
    {
      "epoch": 0.49114168231889943,
      "grad_norm": 0.5040678381919861,
      "learning_rate": 1.6725722117874004e-05,
      "loss": 0.1,
      "step": 9675
    },
    {
      "epoch": 0.4913955023097619,
      "grad_norm": 0.5477058291435242,
      "learning_rate": 1.6724029984601588e-05,
      "loss": 0.098,
      "step": 9680
    },
    {
      "epoch": 0.4916493223006244,
      "grad_norm": 0.5544368028640747,
      "learning_rate": 1.672233785132917e-05,
      "loss": 0.1033,
      "step": 9685
    },
    {
      "epoch": 0.49190314229148685,
      "grad_norm": 0.3991355001926422,
      "learning_rate": 1.6720645718056755e-05,
      "loss": 0.1013,
      "step": 9690
    },
    {
      "epoch": 0.49215696228234934,
      "grad_norm": 0.39113104343414307,
      "learning_rate": 1.671895358478434e-05,
      "loss": 0.0954,
      "step": 9695
    },
    {
      "epoch": 0.49241078227321183,
      "grad_norm": 0.5807805061340332,
      "learning_rate": 1.6717261451511922e-05,
      "loss": 0.0946,
      "step": 9700
    },
    {
      "epoch": 0.4926646022640743,
      "grad_norm": 0.6040493249893188,
      "learning_rate": 1.6715569318239506e-05,
      "loss": 0.1099,
      "step": 9705
    },
    {
      "epoch": 0.4929184222549368,
      "grad_norm": 0.8760645985603333,
      "learning_rate": 1.671387718496709e-05,
      "loss": 0.1111,
      "step": 9710
    },
    {
      "epoch": 0.49317224224579925,
      "grad_norm": 0.7517929077148438,
      "learning_rate": 1.6712185051694673e-05,
      "loss": 0.1055,
      "step": 9715
    },
    {
      "epoch": 0.49342606223666174,
      "grad_norm": 0.5098727345466614,
      "learning_rate": 1.6710492918422256e-05,
      "loss": 0.0984,
      "step": 9720
    },
    {
      "epoch": 0.49367988222752424,
      "grad_norm": 0.7740623354911804,
      "learning_rate": 1.670880078514984e-05,
      "loss": 0.1011,
      "step": 9725
    },
    {
      "epoch": 0.4939337022183867,
      "grad_norm": 0.621465802192688,
      "learning_rate": 1.6707108651877423e-05,
      "loss": 0.1139,
      "step": 9730
    },
    {
      "epoch": 0.4941875222092492,
      "grad_norm": 0.5187132358551025,
      "learning_rate": 1.6705416518605007e-05,
      "loss": 0.1043,
      "step": 9735
    },
    {
      "epoch": 0.49444134220011166,
      "grad_norm": 0.42478689551353455,
      "learning_rate": 1.670372438533259e-05,
      "loss": 0.1017,
      "step": 9740
    },
    {
      "epoch": 0.49469516219097415,
      "grad_norm": 0.45251190662384033,
      "learning_rate": 1.6702032252060174e-05,
      "loss": 0.114,
      "step": 9745
    },
    {
      "epoch": 0.49494898218183664,
      "grad_norm": 0.48776471614837646,
      "learning_rate": 1.6700340118787758e-05,
      "loss": 0.107,
      "step": 9750
    },
    {
      "epoch": 0.49520280217269913,
      "grad_norm": 1.0470062494277954,
      "learning_rate": 1.669864798551534e-05,
      "loss": 0.0999,
      "step": 9755
    },
    {
      "epoch": 0.4954566221635616,
      "grad_norm": 0.5062408447265625,
      "learning_rate": 1.6696955852242925e-05,
      "loss": 0.1076,
      "step": 9760
    },
    {
      "epoch": 0.49571044215442406,
      "grad_norm": 0.35752901434898376,
      "learning_rate": 1.6695263718970508e-05,
      "loss": 0.1169,
      "step": 9765
    },
    {
      "epoch": 0.49596426214528655,
      "grad_norm": 0.4572597146034241,
      "learning_rate": 1.6693571585698092e-05,
      "loss": 0.0932,
      "step": 9770
    },
    {
      "epoch": 0.49621808213614904,
      "grad_norm": 1.1978650093078613,
      "learning_rate": 1.6691879452425675e-05,
      "loss": 0.1062,
      "step": 9775
    },
    {
      "epoch": 0.49647190212701153,
      "grad_norm": 0.5021398663520813,
      "learning_rate": 1.6690187319153256e-05,
      "loss": 0.1198,
      "step": 9780
    },
    {
      "epoch": 0.496725722117874,
      "grad_norm": 0.8845065236091614,
      "learning_rate": 1.6688495185880842e-05,
      "loss": 0.1021,
      "step": 9785
    },
    {
      "epoch": 0.49697954210873646,
      "grad_norm": 0.5153151750564575,
      "learning_rate": 1.6686803052608426e-05,
      "loss": 0.1053,
      "step": 9790
    },
    {
      "epoch": 0.49723336209959895,
      "grad_norm": 0.5133734941482544,
      "learning_rate": 1.6685110919336006e-05,
      "loss": 0.0964,
      "step": 9795
    },
    {
      "epoch": 0.49748718209046144,
      "grad_norm": 0.4482670724391937,
      "learning_rate": 1.6683418786063593e-05,
      "loss": 0.1081,
      "step": 9800
    },
    {
      "epoch": 0.49774100208132394,
      "grad_norm": 0.4168063998222351,
      "learning_rate": 1.6681726652791173e-05,
      "loss": 0.1145,
      "step": 9805
    },
    {
      "epoch": 0.4979948220721864,
      "grad_norm": 0.6683678030967712,
      "learning_rate": 1.668003451951876e-05,
      "loss": 0.0899,
      "step": 9810
    },
    {
      "epoch": 0.49824864206304886,
      "grad_norm": 0.46197474002838135,
      "learning_rate": 1.6678342386246344e-05,
      "loss": 0.1042,
      "step": 9815
    },
    {
      "epoch": 0.49850246205391135,
      "grad_norm": 0.3457307517528534,
      "learning_rate": 1.6676650252973924e-05,
      "loss": 0.1019,
      "step": 9820
    },
    {
      "epoch": 0.49875628204477385,
      "grad_norm": 0.41356801986694336,
      "learning_rate": 1.667495811970151e-05,
      "loss": 0.1043,
      "step": 9825
    },
    {
      "epoch": 0.49901010203563634,
      "grad_norm": 0.43680256605148315,
      "learning_rate": 1.667326598642909e-05,
      "loss": 0.1083,
      "step": 9830
    },
    {
      "epoch": 0.49926392202649883,
      "grad_norm": 0.48716604709625244,
      "learning_rate": 1.6671573853156675e-05,
      "loss": 0.1028,
      "step": 9835
    },
    {
      "epoch": 0.49951774201736127,
      "grad_norm": 0.5266230702400208,
      "learning_rate": 1.666988171988426e-05,
      "loss": 0.1009,
      "step": 9840
    },
    {
      "epoch": 0.49977156200822376,
      "grad_norm": 0.4376629889011383,
      "learning_rate": 1.666818958661184e-05,
      "loss": 0.1078,
      "step": 9845
    },
    {
      "epoch": 0.5000253819990862,
      "grad_norm": 0.48102936148643494,
      "learning_rate": 1.666649745333943e-05,
      "loss": 0.092,
      "step": 9850
    },
    {
      "epoch": 0.5002792019899487,
      "grad_norm": 0.4479629099369049,
      "learning_rate": 1.666480532006701e-05,
      "loss": 0.0984,
      "step": 9855
    },
    {
      "epoch": 0.5005330219808112,
      "grad_norm": 0.429806649684906,
      "learning_rate": 1.6663113186794592e-05,
      "loss": 0.09,
      "step": 9860
    },
    {
      "epoch": 0.5007868419716737,
      "grad_norm": 0.4415090084075928,
      "learning_rate": 1.666142105352218e-05,
      "loss": 0.0989,
      "step": 9865
    },
    {
      "epoch": 0.5010406619625362,
      "grad_norm": 0.4832078516483307,
      "learning_rate": 1.665972892024976e-05,
      "loss": 0.1061,
      "step": 9870
    },
    {
      "epoch": 0.5012944819533987,
      "grad_norm": 0.4944320619106293,
      "learning_rate": 1.6658036786977343e-05,
      "loss": 0.1129,
      "step": 9875
    },
    {
      "epoch": 0.5015483019442611,
      "grad_norm": 0.4951033592224121,
      "learning_rate": 1.6656344653704926e-05,
      "loss": 0.1057,
      "step": 9880
    },
    {
      "epoch": 0.5018021219351236,
      "grad_norm": 0.46183741092681885,
      "learning_rate": 1.665465252043251e-05,
      "loss": 0.103,
      "step": 9885
    },
    {
      "epoch": 0.5020559419259861,
      "grad_norm": 0.4078376293182373,
      "learning_rate": 1.6652960387160093e-05,
      "loss": 0.1023,
      "step": 9890
    },
    {
      "epoch": 0.5023097619168486,
      "grad_norm": 0.4910680949687958,
      "learning_rate": 1.6651268253887677e-05,
      "loss": 0.1045,
      "step": 9895
    },
    {
      "epoch": 0.502563581907711,
      "grad_norm": 0.5417612791061401,
      "learning_rate": 1.664957612061526e-05,
      "loss": 0.098,
      "step": 9900
    },
    {
      "epoch": 0.5028174018985735,
      "grad_norm": 0.44090452790260315,
      "learning_rate": 1.6647883987342844e-05,
      "loss": 0.1098,
      "step": 9905
    },
    {
      "epoch": 0.503071221889436,
      "grad_norm": 0.707319438457489,
      "learning_rate": 1.6646191854070428e-05,
      "loss": 0.0929,
      "step": 9910
    },
    {
      "epoch": 0.5033250418802985,
      "grad_norm": 0.4322894513607025,
      "learning_rate": 1.664449972079801e-05,
      "loss": 0.0897,
      "step": 9915
    },
    {
      "epoch": 0.503578861871161,
      "grad_norm": 0.44595420360565186,
      "learning_rate": 1.6642807587525595e-05,
      "loss": 0.1098,
      "step": 9920
    },
    {
      "epoch": 0.5038326818620235,
      "grad_norm": 0.47478431463241577,
      "learning_rate": 1.664111545425318e-05,
      "loss": 0.1152,
      "step": 9925
    },
    {
      "epoch": 0.504086501852886,
      "grad_norm": 0.4155206084251404,
      "learning_rate": 1.6639423320980762e-05,
      "loss": 0.101,
      "step": 9930
    },
    {
      "epoch": 0.5043403218437484,
      "grad_norm": 0.6963584423065186,
      "learning_rate": 1.6637731187708345e-05,
      "loss": 0.0959,
      "step": 9935
    },
    {
      "epoch": 0.5045941418346109,
      "grad_norm": 0.4272857904434204,
      "learning_rate": 1.663603905443593e-05,
      "loss": 0.0983,
      "step": 9940
    },
    {
      "epoch": 0.5048479618254734,
      "grad_norm": 0.4493730962276459,
      "learning_rate": 1.6634346921163512e-05,
      "loss": 0.1028,
      "step": 9945
    },
    {
      "epoch": 0.5051017818163358,
      "grad_norm": 0.5067510008811951,
      "learning_rate": 1.6632654787891096e-05,
      "loss": 0.0942,
      "step": 9950
    },
    {
      "epoch": 0.5053556018071983,
      "grad_norm": 0.5519894361495972,
      "learning_rate": 1.663096265461868e-05,
      "loss": 0.0949,
      "step": 9955
    },
    {
      "epoch": 0.5056094217980608,
      "grad_norm": 0.4868788719177246,
      "learning_rate": 1.6629270521346263e-05,
      "loss": 0.1023,
      "step": 9960
    },
    {
      "epoch": 0.5058632417889233,
      "grad_norm": 0.40864047408103943,
      "learning_rate": 1.6627578388073847e-05,
      "loss": 0.1036,
      "step": 9965
    },
    {
      "epoch": 0.5061170617797858,
      "grad_norm": 0.33604058623313904,
      "learning_rate": 1.662588625480143e-05,
      "loss": 0.0921,
      "step": 9970
    },
    {
      "epoch": 0.5063708817706483,
      "grad_norm": 0.39679473638534546,
      "learning_rate": 1.6624194121529014e-05,
      "loss": 0.1007,
      "step": 9975
    },
    {
      "epoch": 0.5066247017615108,
      "grad_norm": 0.4948192238807678,
      "learning_rate": 1.6622501988256597e-05,
      "loss": 0.104,
      "step": 9980
    },
    {
      "epoch": 0.5068785217523732,
      "grad_norm": 0.4956167936325073,
      "learning_rate": 1.6620809854984177e-05,
      "loss": 0.1078,
      "step": 9985
    },
    {
      "epoch": 0.5071323417432357,
      "grad_norm": 1.0531851053237915,
      "learning_rate": 1.6619117721711764e-05,
      "loss": 0.102,
      "step": 9990
    },
    {
      "epoch": 0.5073861617340982,
      "grad_norm": 0.3761901557445526,
      "learning_rate": 1.6617425588439348e-05,
      "loss": 0.108,
      "step": 9995
    },
    {
      "epoch": 0.5076399817249606,
      "grad_norm": 0.42027950286865234,
      "learning_rate": 1.661573345516693e-05,
      "loss": 0.0946,
      "step": 10000
    },
    {
      "epoch": 0.5078938017158231,
      "grad_norm": 0.44332537055015564,
      "learning_rate": 1.6614041321894515e-05,
      "loss": 0.1023,
      "step": 10005
    },
    {
      "epoch": 0.5081476217066856,
      "grad_norm": 0.35272926092147827,
      "learning_rate": 1.6612349188622095e-05,
      "loss": 0.0892,
      "step": 10010
    },
    {
      "epoch": 0.5084014416975481,
      "grad_norm": 0.4161962866783142,
      "learning_rate": 1.6610657055349682e-05,
      "loss": 0.0938,
      "step": 10015
    },
    {
      "epoch": 0.5086552616884106,
      "grad_norm": 1.0567114353179932,
      "learning_rate": 1.6608964922077266e-05,
      "loss": 0.1041,
      "step": 10020
    },
    {
      "epoch": 0.5089090816792731,
      "grad_norm": 0.40777215361595154,
      "learning_rate": 1.6607272788804846e-05,
      "loss": 0.103,
      "step": 10025
    },
    {
      "epoch": 0.5091629016701356,
      "grad_norm": 0.6142024993896484,
      "learning_rate": 1.6605580655532433e-05,
      "loss": 0.1047,
      "step": 10030
    },
    {
      "epoch": 0.509416721660998,
      "grad_norm": 0.6380098462104797,
      "learning_rate": 1.6603888522260013e-05,
      "loss": 0.1104,
      "step": 10035
    },
    {
      "epoch": 0.5096705416518605,
      "grad_norm": 0.5984404683113098,
      "learning_rate": 1.6602196388987596e-05,
      "loss": 0.097,
      "step": 10040
    },
    {
      "epoch": 0.509924361642723,
      "grad_norm": 0.4230678081512451,
      "learning_rate": 1.6600504255715183e-05,
      "loss": 0.1015,
      "step": 10045
    },
    {
      "epoch": 0.5101781816335854,
      "grad_norm": 0.38381847739219666,
      "learning_rate": 1.6598812122442764e-05,
      "loss": 0.0914,
      "step": 10050
    },
    {
      "epoch": 0.5104320016244479,
      "grad_norm": 0.44329530000686646,
      "learning_rate": 1.659711998917035e-05,
      "loss": 0.1188,
      "step": 10055
    },
    {
      "epoch": 0.5106858216153104,
      "grad_norm": 0.351794570684433,
      "learning_rate": 1.659542785589793e-05,
      "loss": 0.1054,
      "step": 10060
    },
    {
      "epoch": 0.5109396416061729,
      "grad_norm": 0.596990704536438,
      "learning_rate": 1.6593735722625514e-05,
      "loss": 0.1058,
      "step": 10065
    },
    {
      "epoch": 0.5111934615970354,
      "grad_norm": 0.53257155418396,
      "learning_rate": 1.65920435893531e-05,
      "loss": 0.0999,
      "step": 10070
    },
    {
      "epoch": 0.5114472815878979,
      "grad_norm": 0.768983781337738,
      "learning_rate": 1.659035145608068e-05,
      "loss": 0.0938,
      "step": 10075
    },
    {
      "epoch": 0.5117011015787604,
      "grad_norm": 0.6798129081726074,
      "learning_rate": 1.6588659322808265e-05,
      "loss": 0.0976,
      "step": 10080
    },
    {
      "epoch": 0.5119549215696229,
      "grad_norm": 0.4623749256134033,
      "learning_rate": 1.658696718953585e-05,
      "loss": 0.1173,
      "step": 10085
    },
    {
      "epoch": 0.5122087415604853,
      "grad_norm": 0.3781088590621948,
      "learning_rate": 1.6585275056263432e-05,
      "loss": 0.0992,
      "step": 10090
    },
    {
      "epoch": 0.5124625615513478,
      "grad_norm": 0.5837762951850891,
      "learning_rate": 1.6583582922991015e-05,
      "loss": 0.0996,
      "step": 10095
    },
    {
      "epoch": 0.5127163815422102,
      "grad_norm": 0.5101218819618225,
      "learning_rate": 1.65818907897186e-05,
      "loss": 0.0951,
      "step": 10100
    },
    {
      "epoch": 0.5129702015330727,
      "grad_norm": 0.5419501066207886,
      "learning_rate": 1.6580198656446183e-05,
      "loss": 0.1077,
      "step": 10105
    },
    {
      "epoch": 0.5132240215239352,
      "grad_norm": 0.28744372725486755,
      "learning_rate": 1.6578506523173766e-05,
      "loss": 0.0928,
      "step": 10110
    },
    {
      "epoch": 0.5134778415147977,
      "grad_norm": 0.43558162450790405,
      "learning_rate": 1.657681438990135e-05,
      "loss": 0.0963,
      "step": 10115
    },
    {
      "epoch": 0.5137316615056602,
      "grad_norm": 0.35775113105773926,
      "learning_rate": 1.6575122256628933e-05,
      "loss": 0.0956,
      "step": 10120
    },
    {
      "epoch": 0.5139854814965227,
      "grad_norm": 0.7168741822242737,
      "learning_rate": 1.6573430123356517e-05,
      "loss": 0.0902,
      "step": 10125
    },
    {
      "epoch": 0.5142393014873852,
      "grad_norm": 0.3916659355163574,
      "learning_rate": 1.65717379900841e-05,
      "loss": 0.105,
      "step": 10130
    },
    {
      "epoch": 0.5144931214782477,
      "grad_norm": 0.418158620595932,
      "learning_rate": 1.6570045856811684e-05,
      "loss": 0.0916,
      "step": 10135
    },
    {
      "epoch": 0.5147469414691102,
      "grad_norm": 0.678088903427124,
      "learning_rate": 1.6568353723539267e-05,
      "loss": 0.1082,
      "step": 10140
    },
    {
      "epoch": 0.5150007614599725,
      "grad_norm": 0.42723122239112854,
      "learning_rate": 1.656666159026685e-05,
      "loss": 0.1036,
      "step": 10145
    },
    {
      "epoch": 0.515254581450835,
      "grad_norm": 0.6436286568641663,
      "learning_rate": 1.6564969456994434e-05,
      "loss": 0.109,
      "step": 10150
    },
    {
      "epoch": 0.5155084014416975,
      "grad_norm": 0.7051729559898376,
      "learning_rate": 1.6563277323722018e-05,
      "loss": 0.1164,
      "step": 10155
    },
    {
      "epoch": 0.51576222143256,
      "grad_norm": 0.5349159836769104,
      "learning_rate": 1.65615851904496e-05,
      "loss": 0.1171,
      "step": 10160
    },
    {
      "epoch": 0.5160160414234225,
      "grad_norm": 0.6342788338661194,
      "learning_rate": 1.6559893057177185e-05,
      "loss": 0.114,
      "step": 10165
    },
    {
      "epoch": 0.516269861414285,
      "grad_norm": 0.4344607889652252,
      "learning_rate": 1.655820092390477e-05,
      "loss": 0.0888,
      "step": 10170
    },
    {
      "epoch": 0.5165236814051475,
      "grad_norm": 0.6470787525177002,
      "learning_rate": 1.6556508790632352e-05,
      "loss": 0.0952,
      "step": 10175
    },
    {
      "epoch": 0.51677750139601,
      "grad_norm": 0.6785343289375305,
      "learning_rate": 1.6554816657359936e-05,
      "loss": 0.111,
      "step": 10180
    },
    {
      "epoch": 0.5170313213868725,
      "grad_norm": 0.41532400250434875,
      "learning_rate": 1.655312452408752e-05,
      "loss": 0.0911,
      "step": 10185
    },
    {
      "epoch": 0.517285141377735,
      "grad_norm": 0.4153634309768677,
      "learning_rate": 1.65514323908151e-05,
      "loss": 0.0986,
      "step": 10190
    },
    {
      "epoch": 0.5175389613685973,
      "grad_norm": 0.49320197105407715,
      "learning_rate": 1.6549740257542686e-05,
      "loss": 0.1052,
      "step": 10195
    },
    {
      "epoch": 0.5177927813594598,
      "grad_norm": 0.44718942046165466,
      "learning_rate": 1.654804812427027e-05,
      "loss": 0.1064,
      "step": 10200
    },
    {
      "epoch": 0.5180466013503223,
      "grad_norm": 0.49730628728866577,
      "learning_rate": 1.6546355990997853e-05,
      "loss": 0.0981,
      "step": 10205
    },
    {
      "epoch": 0.5183004213411848,
      "grad_norm": 0.7655092477798462,
      "learning_rate": 1.6544663857725437e-05,
      "loss": 0.1057,
      "step": 10210
    },
    {
      "epoch": 0.5185542413320473,
      "grad_norm": 0.3895924389362335,
      "learning_rate": 1.6542971724453017e-05,
      "loss": 0.0978,
      "step": 10215
    },
    {
      "epoch": 0.5188080613229098,
      "grad_norm": 0.5672686696052551,
      "learning_rate": 1.6541279591180604e-05,
      "loss": 0.1043,
      "step": 10220
    },
    {
      "epoch": 0.5190618813137723,
      "grad_norm": 0.5215070843696594,
      "learning_rate": 1.6539587457908188e-05,
      "loss": 0.1016,
      "step": 10225
    },
    {
      "epoch": 0.5193157013046348,
      "grad_norm": 0.5295329689979553,
      "learning_rate": 1.6537895324635768e-05,
      "loss": 0.0954,
      "step": 10230
    },
    {
      "epoch": 0.5195695212954973,
      "grad_norm": 0.4239153563976288,
      "learning_rate": 1.6536203191363355e-05,
      "loss": 0.1025,
      "step": 10235
    },
    {
      "epoch": 0.5198233412863598,
      "grad_norm": 0.597144365310669,
      "learning_rate": 1.6534511058090935e-05,
      "loss": 0.0936,
      "step": 10240
    },
    {
      "epoch": 0.5200771612772221,
      "grad_norm": 0.4462467133998871,
      "learning_rate": 1.6532818924818522e-05,
      "loss": 0.1048,
      "step": 10245
    },
    {
      "epoch": 0.5203309812680846,
      "grad_norm": 0.7833221554756165,
      "learning_rate": 1.6531126791546105e-05,
      "loss": 0.1123,
      "step": 10250
    },
    {
      "epoch": 0.5205848012589471,
      "grad_norm": 0.7217886447906494,
      "learning_rate": 1.6529434658273685e-05,
      "loss": 0.1007,
      "step": 10255
    },
    {
      "epoch": 0.5208386212498096,
      "grad_norm": 0.6208773255348206,
      "learning_rate": 1.6527742525001272e-05,
      "loss": 0.088,
      "step": 10260
    },
    {
      "epoch": 0.5210924412406721,
      "grad_norm": 0.5365100502967834,
      "learning_rate": 1.6526050391728853e-05,
      "loss": 0.1052,
      "step": 10265
    },
    {
      "epoch": 0.5213462612315346,
      "grad_norm": 0.4037196636199951,
      "learning_rate": 1.6524358258456436e-05,
      "loss": 0.0958,
      "step": 10270
    },
    {
      "epoch": 0.5216000812223971,
      "grad_norm": 0.367753267288208,
      "learning_rate": 1.6522666125184023e-05,
      "loss": 0.0873,
      "step": 10275
    },
    {
      "epoch": 0.5218539012132596,
      "grad_norm": 0.5729825496673584,
      "learning_rate": 1.6520973991911603e-05,
      "loss": 0.1014,
      "step": 10280
    },
    {
      "epoch": 0.5221077212041221,
      "grad_norm": 0.3999881446361542,
      "learning_rate": 1.6519281858639187e-05,
      "loss": 0.0905,
      "step": 10285
    },
    {
      "epoch": 0.5223615411949846,
      "grad_norm": 0.5186746716499329,
      "learning_rate": 1.651758972536677e-05,
      "loss": 0.0934,
      "step": 10290
    },
    {
      "epoch": 0.522615361185847,
      "grad_norm": 0.4636486768722534,
      "learning_rate": 1.6515897592094354e-05,
      "loss": 0.0943,
      "step": 10295
    },
    {
      "epoch": 0.5228691811767094,
      "grad_norm": 0.5422764420509338,
      "learning_rate": 1.651420545882194e-05,
      "loss": 0.0939,
      "step": 10300
    },
    {
      "epoch": 0.5231230011675719,
      "grad_norm": 0.4960278570652008,
      "learning_rate": 1.651251332554952e-05,
      "loss": 0.089,
      "step": 10305
    },
    {
      "epoch": 0.5233768211584344,
      "grad_norm": 0.47158360481262207,
      "learning_rate": 1.6510821192277104e-05,
      "loss": 0.1025,
      "step": 10310
    },
    {
      "epoch": 0.5236306411492969,
      "grad_norm": 0.5370834469795227,
      "learning_rate": 1.6509129059004688e-05,
      "loss": 0.0955,
      "step": 10315
    },
    {
      "epoch": 0.5238844611401594,
      "grad_norm": 1.4522969722747803,
      "learning_rate": 1.650743692573227e-05,
      "loss": 0.1046,
      "step": 10320
    },
    {
      "epoch": 0.5241382811310219,
      "grad_norm": 0.5088288187980652,
      "learning_rate": 1.6505744792459855e-05,
      "loss": 0.1002,
      "step": 10325
    },
    {
      "epoch": 0.5243921011218844,
      "grad_norm": 0.6046572327613831,
      "learning_rate": 1.650405265918744e-05,
      "loss": 0.0896,
      "step": 10330
    },
    {
      "epoch": 0.5246459211127469,
      "grad_norm": 0.4464103877544403,
      "learning_rate": 1.6502360525915022e-05,
      "loss": 0.1115,
      "step": 10335
    },
    {
      "epoch": 0.5248997411036094,
      "grad_norm": 0.42552125453948975,
      "learning_rate": 1.6500668392642606e-05,
      "loss": 0.0953,
      "step": 10340
    },
    {
      "epoch": 0.5251535610944718,
      "grad_norm": 0.3787493407726288,
      "learning_rate": 1.649897625937019e-05,
      "loss": 0.0923,
      "step": 10345
    },
    {
      "epoch": 0.5254073810853342,
      "grad_norm": 0.7609786987304688,
      "learning_rate": 1.6497284126097773e-05,
      "loss": 0.0985,
      "step": 10350
    },
    {
      "epoch": 0.5256612010761967,
      "grad_norm": 0.41064974665641785,
      "learning_rate": 1.6495591992825356e-05,
      "loss": 0.1105,
      "step": 10355
    },
    {
      "epoch": 0.5259150210670592,
      "grad_norm": 0.4209999740123749,
      "learning_rate": 1.649389985955294e-05,
      "loss": 0.0995,
      "step": 10360
    },
    {
      "epoch": 0.5261688410579217,
      "grad_norm": 0.405991792678833,
      "learning_rate": 1.6492207726280523e-05,
      "loss": 0.0927,
      "step": 10365
    },
    {
      "epoch": 0.5264226610487842,
      "grad_norm": 0.47690391540527344,
      "learning_rate": 1.6490515593008107e-05,
      "loss": 0.1119,
      "step": 10370
    },
    {
      "epoch": 0.5266764810396467,
      "grad_norm": 0.3825083076953888,
      "learning_rate": 1.648882345973569e-05,
      "loss": 0.1069,
      "step": 10375
    },
    {
      "epoch": 0.5269303010305092,
      "grad_norm": 0.6007245779037476,
      "learning_rate": 1.6487131326463274e-05,
      "loss": 0.0949,
      "step": 10380
    },
    {
      "epoch": 0.5271841210213717,
      "grad_norm": 0.41400671005249023,
      "learning_rate": 1.6485439193190858e-05,
      "loss": 0.0959,
      "step": 10385
    },
    {
      "epoch": 0.5274379410122341,
      "grad_norm": 0.8201514482498169,
      "learning_rate": 1.648374705991844e-05,
      "loss": 0.0948,
      "step": 10390
    },
    {
      "epoch": 0.5276917610030966,
      "grad_norm": 0.4177849292755127,
      "learning_rate": 1.6482054926646025e-05,
      "loss": 0.1071,
      "step": 10395
    },
    {
      "epoch": 0.527945580993959,
      "grad_norm": 0.45180994272232056,
      "learning_rate": 1.6480362793373608e-05,
      "loss": 0.1038,
      "step": 10400
    },
    {
      "epoch": 0.5281994009848215,
      "grad_norm": 0.5408563613891602,
      "learning_rate": 1.6478670660101192e-05,
      "loss": 0.1203,
      "step": 10405
    },
    {
      "epoch": 0.528453220975684,
      "grad_norm": 0.4275490939617157,
      "learning_rate": 1.6476978526828775e-05,
      "loss": 0.1016,
      "step": 10410
    },
    {
      "epoch": 0.5287070409665465,
      "grad_norm": 0.4530520439147949,
      "learning_rate": 1.647528639355636e-05,
      "loss": 0.1036,
      "step": 10415
    },
    {
      "epoch": 0.528960860957409,
      "grad_norm": 0.49070650339126587,
      "learning_rate": 1.647359426028394e-05,
      "loss": 0.0977,
      "step": 10420
    },
    {
      "epoch": 0.5292146809482715,
      "grad_norm": 0.4170216917991638,
      "learning_rate": 1.6471902127011526e-05,
      "loss": 0.1227,
      "step": 10425
    },
    {
      "epoch": 0.529468500939134,
      "grad_norm": 0.3593534827232361,
      "learning_rate": 1.647020999373911e-05,
      "loss": 0.1125,
      "step": 10430
    },
    {
      "epoch": 0.5297223209299965,
      "grad_norm": 0.49653005599975586,
      "learning_rate": 1.646851786046669e-05,
      "loss": 0.0916,
      "step": 10435
    },
    {
      "epoch": 0.5299761409208589,
      "grad_norm": 0.536085844039917,
      "learning_rate": 1.6466825727194277e-05,
      "loss": 0.0895,
      "step": 10440
    },
    {
      "epoch": 0.5302299609117214,
      "grad_norm": 0.4254559278488159,
      "learning_rate": 1.6465133593921857e-05,
      "loss": 0.0948,
      "step": 10445
    },
    {
      "epoch": 0.5304837809025839,
      "grad_norm": 0.3751010000705719,
      "learning_rate": 1.6463441460649444e-05,
      "loss": 0.0907,
      "step": 10450
    },
    {
      "epoch": 0.5307376008934463,
      "grad_norm": 0.4354502558708191,
      "learning_rate": 1.6461749327377027e-05,
      "loss": 0.1001,
      "step": 10455
    },
    {
      "epoch": 0.5309914208843088,
      "grad_norm": 0.3564659357070923,
      "learning_rate": 1.6460057194104607e-05,
      "loss": 0.0916,
      "step": 10460
    },
    {
      "epoch": 0.5312452408751713,
      "grad_norm": 0.47137919068336487,
      "learning_rate": 1.6458365060832194e-05,
      "loss": 0.1,
      "step": 10465
    },
    {
      "epoch": 0.5314990608660338,
      "grad_norm": 0.32844093441963196,
      "learning_rate": 1.6456672927559775e-05,
      "loss": 0.0874,
      "step": 10470
    },
    {
      "epoch": 0.5317528808568963,
      "grad_norm": 0.521373450756073,
      "learning_rate": 1.6454980794287358e-05,
      "loss": 0.1061,
      "step": 10475
    },
    {
      "epoch": 0.5320067008477588,
      "grad_norm": 0.36201703548431396,
      "learning_rate": 1.6453288661014945e-05,
      "loss": 0.1006,
      "step": 10480
    },
    {
      "epoch": 0.5322605208386213,
      "grad_norm": 0.5359712839126587,
      "learning_rate": 1.6451596527742525e-05,
      "loss": 0.092,
      "step": 10485
    },
    {
      "epoch": 0.5325143408294837,
      "grad_norm": 0.38301995396614075,
      "learning_rate": 1.6449904394470112e-05,
      "loss": 0.0994,
      "step": 10490
    },
    {
      "epoch": 0.5327681608203462,
      "grad_norm": 0.3559681475162506,
      "learning_rate": 1.6448212261197692e-05,
      "loss": 0.0872,
      "step": 10495
    },
    {
      "epoch": 0.5330219808112087,
      "grad_norm": 0.4657198488712311,
      "learning_rate": 1.6446520127925276e-05,
      "loss": 0.0962,
      "step": 10500
    },
    {
      "epoch": 0.5332758008020712,
      "grad_norm": 0.8297700881958008,
      "learning_rate": 1.6444827994652863e-05,
      "loss": 0.1149,
      "step": 10505
    },
    {
      "epoch": 0.5335296207929336,
      "grad_norm": 0.4492545425891876,
      "learning_rate": 1.6443135861380443e-05,
      "loss": 0.0958,
      "step": 10510
    },
    {
      "epoch": 0.5337834407837961,
      "grad_norm": 0.41303345561027527,
      "learning_rate": 1.6441443728108026e-05,
      "loss": 0.0971,
      "step": 10515
    },
    {
      "epoch": 0.5340372607746586,
      "grad_norm": 0.4111636281013489,
      "learning_rate": 1.643975159483561e-05,
      "loss": 0.0831,
      "step": 10520
    },
    {
      "epoch": 0.5342910807655211,
      "grad_norm": 0.5072298049926758,
      "learning_rate": 1.6438059461563193e-05,
      "loss": 0.0935,
      "step": 10525
    },
    {
      "epoch": 0.5345449007563836,
      "grad_norm": 0.43646836280822754,
      "learning_rate": 1.6436367328290777e-05,
      "loss": 0.1003,
      "step": 10530
    },
    {
      "epoch": 0.5347987207472461,
      "grad_norm": 0.5552557110786438,
      "learning_rate": 1.643467519501836e-05,
      "loss": 0.0961,
      "step": 10535
    },
    {
      "epoch": 0.5350525407381085,
      "grad_norm": 0.398117333650589,
      "learning_rate": 1.6432983061745944e-05,
      "loss": 0.1038,
      "step": 10540
    },
    {
      "epoch": 0.535306360728971,
      "grad_norm": 0.301830530166626,
      "learning_rate": 1.6431290928473528e-05,
      "loss": 0.0828,
      "step": 10545
    },
    {
      "epoch": 0.5355601807198335,
      "grad_norm": 0.43275371193885803,
      "learning_rate": 1.642959879520111e-05,
      "loss": 0.1077,
      "step": 10550
    },
    {
      "epoch": 0.535814000710696,
      "grad_norm": 0.3594335913658142,
      "learning_rate": 1.6427906661928695e-05,
      "loss": 0.0974,
      "step": 10555
    },
    {
      "epoch": 0.5360678207015585,
      "grad_norm": 0.3689029812812805,
      "learning_rate": 1.6426214528656278e-05,
      "loss": 0.0762,
      "step": 10560
    },
    {
      "epoch": 0.5363216406924209,
      "grad_norm": 0.7268462777137756,
      "learning_rate": 1.6424522395383862e-05,
      "loss": 0.0943,
      "step": 10565
    },
    {
      "epoch": 0.5365754606832834,
      "grad_norm": 0.5907272696495056,
      "learning_rate": 1.6422830262111445e-05,
      "loss": 0.1044,
      "step": 10570
    },
    {
      "epoch": 0.5368292806741459,
      "grad_norm": 0.640363335609436,
      "learning_rate": 1.642113812883903e-05,
      "loss": 0.0887,
      "step": 10575
    },
    {
      "epoch": 0.5370831006650084,
      "grad_norm": 0.9156871438026428,
      "learning_rate": 1.6419445995566612e-05,
      "loss": 0.1107,
      "step": 10580
    },
    {
      "epoch": 0.5373369206558709,
      "grad_norm": 0.449605792760849,
      "learning_rate": 1.6417753862294196e-05,
      "loss": 0.0975,
      "step": 10585
    },
    {
      "epoch": 0.5375907406467333,
      "grad_norm": 0.4629546105861664,
      "learning_rate": 1.641606172902178e-05,
      "loss": 0.0946,
      "step": 10590
    },
    {
      "epoch": 0.5378445606375958,
      "grad_norm": 0.7200871109962463,
      "learning_rate": 1.6414369595749363e-05,
      "loss": 0.1041,
      "step": 10595
    },
    {
      "epoch": 0.5380983806284583,
      "grad_norm": 0.4900858402252197,
      "learning_rate": 1.6412677462476947e-05,
      "loss": 0.1005,
      "step": 10600
    },
    {
      "epoch": 0.5383522006193208,
      "grad_norm": 0.3713901937007904,
      "learning_rate": 1.641098532920453e-05,
      "loss": 0.0933,
      "step": 10605
    },
    {
      "epoch": 0.5386060206101833,
      "grad_norm": 0.4475940763950348,
      "learning_rate": 1.6409293195932114e-05,
      "loss": 0.0977,
      "step": 10610
    },
    {
      "epoch": 0.5388598406010457,
      "grad_norm": 0.5317751169204712,
      "learning_rate": 1.6407601062659697e-05,
      "loss": 0.109,
      "step": 10615
    },
    {
      "epoch": 0.5391136605919082,
      "grad_norm": 0.4223870635032654,
      "learning_rate": 1.640590892938728e-05,
      "loss": 0.0951,
      "step": 10620
    },
    {
      "epoch": 0.5393674805827707,
      "grad_norm": 0.4437798261642456,
      "learning_rate": 1.640421679611486e-05,
      "loss": 0.0971,
      "step": 10625
    },
    {
      "epoch": 0.5396213005736332,
      "grad_norm": 0.5337457060813904,
      "learning_rate": 1.6402524662842448e-05,
      "loss": 0.0893,
      "step": 10630
    },
    {
      "epoch": 0.5398751205644957,
      "grad_norm": 0.5861578583717346,
      "learning_rate": 1.640083252957003e-05,
      "loss": 0.1028,
      "step": 10635
    },
    {
      "epoch": 0.5401289405553581,
      "grad_norm": 0.46542319655418396,
      "learning_rate": 1.6399140396297615e-05,
      "loss": 0.0978,
      "step": 10640
    },
    {
      "epoch": 0.5403827605462206,
      "grad_norm": 0.5477737188339233,
      "learning_rate": 1.63974482630252e-05,
      "loss": 0.0928,
      "step": 10645
    },
    {
      "epoch": 0.5406365805370831,
      "grad_norm": 0.38435491919517517,
      "learning_rate": 1.639575612975278e-05,
      "loss": 0.1031,
      "step": 10650
    },
    {
      "epoch": 0.5408904005279456,
      "grad_norm": 0.632420003414154,
      "learning_rate": 1.6394063996480366e-05,
      "loss": 0.1089,
      "step": 10655
    },
    {
      "epoch": 0.5411442205188081,
      "grad_norm": 0.2502029240131378,
      "learning_rate": 1.639237186320795e-05,
      "loss": 0.1011,
      "step": 10660
    },
    {
      "epoch": 0.5413980405096706,
      "grad_norm": 0.7792786955833435,
      "learning_rate": 1.639067972993553e-05,
      "loss": 0.1021,
      "step": 10665
    },
    {
      "epoch": 0.541651860500533,
      "grad_norm": 0.36001554131507874,
      "learning_rate": 1.6388987596663116e-05,
      "loss": 0.1007,
      "step": 10670
    },
    {
      "epoch": 0.5419056804913955,
      "grad_norm": 0.6201646327972412,
      "learning_rate": 1.6387295463390696e-05,
      "loss": 0.1045,
      "step": 10675
    },
    {
      "epoch": 0.542159500482258,
      "grad_norm": 0.6190532445907593,
      "learning_rate": 1.638560333011828e-05,
      "loss": 0.0953,
      "step": 10680
    },
    {
      "epoch": 0.5424133204731204,
      "grad_norm": 0.7013592720031738,
      "learning_rate": 1.6383911196845867e-05,
      "loss": 0.0991,
      "step": 10685
    },
    {
      "epoch": 0.5426671404639829,
      "grad_norm": 0.3520493507385254,
      "learning_rate": 1.6382219063573447e-05,
      "loss": 0.0828,
      "step": 10690
    },
    {
      "epoch": 0.5429209604548454,
      "grad_norm": 0.788502037525177,
      "learning_rate": 1.6380526930301034e-05,
      "loss": 0.093,
      "step": 10695
    },
    {
      "epoch": 0.5431747804457079,
      "grad_norm": 0.38444724678993225,
      "learning_rate": 1.6378834797028614e-05,
      "loss": 0.0963,
      "step": 10700
    },
    {
      "epoch": 0.5434286004365704,
      "grad_norm": 0.6140927076339722,
      "learning_rate": 1.6377142663756198e-05,
      "loss": 0.103,
      "step": 10705
    },
    {
      "epoch": 0.5436824204274329,
      "grad_norm": 0.5046095848083496,
      "learning_rate": 1.6375450530483785e-05,
      "loss": 0.1056,
      "step": 10710
    },
    {
      "epoch": 0.5439362404182954,
      "grad_norm": 0.5487456917762756,
      "learning_rate": 1.6373758397211365e-05,
      "loss": 0.1221,
      "step": 10715
    },
    {
      "epoch": 0.5441900604091578,
      "grad_norm": 0.49649783968925476,
      "learning_rate": 1.637206626393895e-05,
      "loss": 0.1,
      "step": 10720
    },
    {
      "epoch": 0.5444438804000203,
      "grad_norm": 0.3561123013496399,
      "learning_rate": 1.6370374130666532e-05,
      "loss": 0.0825,
      "step": 10725
    },
    {
      "epoch": 0.5446977003908828,
      "grad_norm": 0.4641798138618469,
      "learning_rate": 1.6368681997394115e-05,
      "loss": 0.0995,
      "step": 10730
    },
    {
      "epoch": 0.5449515203817452,
      "grad_norm": 0.43982285261154175,
      "learning_rate": 1.63669898641217e-05,
      "loss": 0.1151,
      "step": 10735
    },
    {
      "epoch": 0.5452053403726077,
      "grad_norm": 0.3846445679664612,
      "learning_rate": 1.6365297730849283e-05,
      "loss": 0.1044,
      "step": 10740
    },
    {
      "epoch": 0.5454591603634702,
      "grad_norm": 0.46901729702949524,
      "learning_rate": 1.6363605597576866e-05,
      "loss": 0.1155,
      "step": 10745
    },
    {
      "epoch": 0.5457129803543327,
      "grad_norm": 0.3808036744594574,
      "learning_rate": 1.636191346430445e-05,
      "loss": 0.1055,
      "step": 10750
    },
    {
      "epoch": 0.5459668003451952,
      "grad_norm": 1.3655842542648315,
      "learning_rate": 1.6360221331032033e-05,
      "loss": 0.1032,
      "step": 10755
    },
    {
      "epoch": 0.5462206203360577,
      "grad_norm": 0.4646155834197998,
      "learning_rate": 1.6358529197759617e-05,
      "loss": 0.0991,
      "step": 10760
    },
    {
      "epoch": 0.5464744403269202,
      "grad_norm": 0.6558862924575806,
      "learning_rate": 1.63568370644872e-05,
      "loss": 0.1017,
      "step": 10765
    },
    {
      "epoch": 0.5467282603177827,
      "grad_norm": 0.4608815610408783,
      "learning_rate": 1.6355144931214784e-05,
      "loss": 0.1032,
      "step": 10770
    },
    {
      "epoch": 0.5469820803086451,
      "grad_norm": 0.3188372254371643,
      "learning_rate": 1.6353452797942367e-05,
      "loss": 0.0907,
      "step": 10775
    },
    {
      "epoch": 0.5472359002995076,
      "grad_norm": 0.31106188893318176,
      "learning_rate": 1.635176066466995e-05,
      "loss": 0.1002,
      "step": 10780
    },
    {
      "epoch": 0.54748972029037,
      "grad_norm": 0.37392106652259827,
      "learning_rate": 1.6350068531397534e-05,
      "loss": 0.0935,
      "step": 10785
    },
    {
      "epoch": 0.5477435402812325,
      "grad_norm": 0.3712967038154602,
      "learning_rate": 1.6348376398125118e-05,
      "loss": 0.1068,
      "step": 10790
    },
    {
      "epoch": 0.547997360272095,
      "grad_norm": 1.323493242263794,
      "learning_rate": 1.63466842648527e-05,
      "loss": 0.1024,
      "step": 10795
    },
    {
      "epoch": 0.5482511802629575,
      "grad_norm": 0.3808216452598572,
      "learning_rate": 1.6344992131580285e-05,
      "loss": 0.0899,
      "step": 10800
    },
    {
      "epoch": 0.54850500025382,
      "grad_norm": 0.7817515134811401,
      "learning_rate": 1.634329999830787e-05,
      "loss": 0.1175,
      "step": 10805
    },
    {
      "epoch": 0.5487588202446825,
      "grad_norm": 0.9595544934272766,
      "learning_rate": 1.6341607865035452e-05,
      "loss": 0.0989,
      "step": 10810
    },
    {
      "epoch": 0.549012640235545,
      "grad_norm": 0.5686103701591492,
      "learning_rate": 1.6339915731763036e-05,
      "loss": 0.1023,
      "step": 10815
    },
    {
      "epoch": 0.5492664602264075,
      "grad_norm": 0.457376092672348,
      "learning_rate": 1.633822359849062e-05,
      "loss": 0.1002,
      "step": 10820
    },
    {
      "epoch": 0.54952028021727,
      "grad_norm": 0.954951286315918,
      "learning_rate": 1.6336531465218203e-05,
      "loss": 0.1003,
      "step": 10825
    },
    {
      "epoch": 0.5497741002081324,
      "grad_norm": 0.40505141019821167,
      "learning_rate": 1.6334839331945783e-05,
      "loss": 0.0966,
      "step": 10830
    },
    {
      "epoch": 0.5500279201989948,
      "grad_norm": 0.41799911856651306,
      "learning_rate": 1.633314719867337e-05,
      "loss": 0.1108,
      "step": 10835
    },
    {
      "epoch": 0.5502817401898573,
      "grad_norm": 0.5992372632026672,
      "learning_rate": 1.6331455065400953e-05,
      "loss": 0.1106,
      "step": 10840
    },
    {
      "epoch": 0.5505355601807198,
      "grad_norm": 0.8904902935028076,
      "learning_rate": 1.6329762932128537e-05,
      "loss": 0.1022,
      "step": 10845
    },
    {
      "epoch": 0.5507893801715823,
      "grad_norm": 0.5283609628677368,
      "learning_rate": 1.632807079885612e-05,
      "loss": 0.0975,
      "step": 10850
    },
    {
      "epoch": 0.5510432001624448,
      "grad_norm": 0.49641087651252747,
      "learning_rate": 1.63263786655837e-05,
      "loss": 0.099,
      "step": 10855
    },
    {
      "epoch": 0.5512970201533073,
      "grad_norm": 0.5636726021766663,
      "learning_rate": 1.6324686532311288e-05,
      "loss": 0.0957,
      "step": 10860
    },
    {
      "epoch": 0.5515508401441698,
      "grad_norm": 0.6105141043663025,
      "learning_rate": 1.632299439903887e-05,
      "loss": 0.1119,
      "step": 10865
    },
    {
      "epoch": 0.5518046601350323,
      "grad_norm": 0.43738439679145813,
      "learning_rate": 1.632130226576645e-05,
      "loss": 0.0902,
      "step": 10870
    },
    {
      "epoch": 0.5520584801258948,
      "grad_norm": 0.6451889872550964,
      "learning_rate": 1.6319610132494038e-05,
      "loss": 0.0906,
      "step": 10875
    },
    {
      "epoch": 0.5523123001167572,
      "grad_norm": 0.4356880187988281,
      "learning_rate": 1.631791799922162e-05,
      "loss": 0.1015,
      "step": 10880
    },
    {
      "epoch": 0.5525661201076196,
      "grad_norm": 0.487354040145874,
      "learning_rate": 1.6316225865949205e-05,
      "loss": 0.1072,
      "step": 10885
    },
    {
      "epoch": 0.5528199400984821,
      "grad_norm": 0.41686418652534485,
      "learning_rate": 1.631453373267679e-05,
      "loss": 0.0994,
      "step": 10890
    },
    {
      "epoch": 0.5530737600893446,
      "grad_norm": 0.5132306218147278,
      "learning_rate": 1.631284159940437e-05,
      "loss": 0.0994,
      "step": 10895
    },
    {
      "epoch": 0.5533275800802071,
      "grad_norm": 0.4513254463672638,
      "learning_rate": 1.6311149466131956e-05,
      "loss": 0.0963,
      "step": 10900
    },
    {
      "epoch": 0.5535814000710696,
      "grad_norm": 0.38187047839164734,
      "learning_rate": 1.6309457332859536e-05,
      "loss": 0.0969,
      "step": 10905
    },
    {
      "epoch": 0.5538352200619321,
      "grad_norm": 0.5190130472183228,
      "learning_rate": 1.630776519958712e-05,
      "loss": 0.1029,
      "step": 10910
    },
    {
      "epoch": 0.5540890400527946,
      "grad_norm": 0.3652910888195038,
      "learning_rate": 1.6306073066314703e-05,
      "loss": 0.0953,
      "step": 10915
    },
    {
      "epoch": 0.5543428600436571,
      "grad_norm": 0.4218839406967163,
      "learning_rate": 1.6304380933042287e-05,
      "loss": 0.0997,
      "step": 10920
    },
    {
      "epoch": 0.5545966800345196,
      "grad_norm": 0.4914858043193817,
      "learning_rate": 1.630268879976987e-05,
      "loss": 0.1099,
      "step": 10925
    },
    {
      "epoch": 0.554850500025382,
      "grad_norm": 0.3577395975589752,
      "learning_rate": 1.6300996666497454e-05,
      "loss": 0.1083,
      "step": 10930
    },
    {
      "epoch": 0.5551043200162444,
      "grad_norm": 0.32556432485580444,
      "learning_rate": 1.6299304533225037e-05,
      "loss": 0.0865,
      "step": 10935
    },
    {
      "epoch": 0.5553581400071069,
      "grad_norm": 0.5093722939491272,
      "learning_rate": 1.629761239995262e-05,
      "loss": 0.0913,
      "step": 10940
    },
    {
      "epoch": 0.5556119599979694,
      "grad_norm": 0.5503299832344055,
      "learning_rate": 1.6295920266680204e-05,
      "loss": 0.1075,
      "step": 10945
    },
    {
      "epoch": 0.5558657799888319,
      "grad_norm": 0.6354576945304871,
      "learning_rate": 1.6294228133407788e-05,
      "loss": 0.103,
      "step": 10950
    },
    {
      "epoch": 0.5561195999796944,
      "grad_norm": 0.43695926666259766,
      "learning_rate": 1.629253600013537e-05,
      "loss": 0.0993,
      "step": 10955
    },
    {
      "epoch": 0.5563734199705569,
      "grad_norm": 0.36351871490478516,
      "learning_rate": 1.6290843866862955e-05,
      "loss": 0.0972,
      "step": 10960
    },
    {
      "epoch": 0.5566272399614194,
      "grad_norm": 0.6152383685112,
      "learning_rate": 1.628915173359054e-05,
      "loss": 0.1052,
      "step": 10965
    },
    {
      "epoch": 0.5568810599522819,
      "grad_norm": 0.7461835145950317,
      "learning_rate": 1.6287459600318122e-05,
      "loss": 0.0893,
      "step": 10970
    },
    {
      "epoch": 0.5571348799431444,
      "grad_norm": 0.45629680156707764,
      "learning_rate": 1.6285767467045706e-05,
      "loss": 0.0984,
      "step": 10975
    },
    {
      "epoch": 0.5573886999340067,
      "grad_norm": 0.4399794340133667,
      "learning_rate": 1.628407533377329e-05,
      "loss": 0.1036,
      "step": 10980
    },
    {
      "epoch": 0.5576425199248692,
      "grad_norm": 0.4887857437133789,
      "learning_rate": 1.6282383200500873e-05,
      "loss": 0.1022,
      "step": 10985
    },
    {
      "epoch": 0.5578963399157317,
      "grad_norm": 0.5037972927093506,
      "learning_rate": 1.6280691067228456e-05,
      "loss": 0.0961,
      "step": 10990
    },
    {
      "epoch": 0.5581501599065942,
      "grad_norm": 0.45002323389053345,
      "learning_rate": 1.627899893395604e-05,
      "loss": 0.0973,
      "step": 10995
    },
    {
      "epoch": 0.5584039798974567,
      "grad_norm": 0.6315869688987732,
      "learning_rate": 1.6277306800683623e-05,
      "loss": 0.0963,
      "step": 11000
    },
    {
      "epoch": 0.5586577998883192,
      "grad_norm": 0.3851071000099182,
      "learning_rate": 1.6275614667411207e-05,
      "loss": 0.0852,
      "step": 11005
    },
    {
      "epoch": 0.5589116198791817,
      "grad_norm": 0.6421680450439453,
      "learning_rate": 1.627392253413879e-05,
      "loss": 0.1016,
      "step": 11010
    },
    {
      "epoch": 0.5591654398700442,
      "grad_norm": 0.7815170884132385,
      "learning_rate": 1.6272230400866374e-05,
      "loss": 0.0833,
      "step": 11015
    },
    {
      "epoch": 0.5594192598609067,
      "grad_norm": 0.6230599880218506,
      "learning_rate": 1.6270538267593958e-05,
      "loss": 0.0952,
      "step": 11020
    },
    {
      "epoch": 0.5596730798517692,
      "grad_norm": 0.6255953311920166,
      "learning_rate": 1.626884613432154e-05,
      "loss": 0.0935,
      "step": 11025
    },
    {
      "epoch": 0.5599268998426316,
      "grad_norm": 0.3688150644302368,
      "learning_rate": 1.6267154001049125e-05,
      "loss": 0.1016,
      "step": 11030
    },
    {
      "epoch": 0.560180719833494,
      "grad_norm": 0.41703271865844727,
      "learning_rate": 1.6265461867776708e-05,
      "loss": 0.0937,
      "step": 11035
    },
    {
      "epoch": 0.5604345398243565,
      "grad_norm": 0.6567605137825012,
      "learning_rate": 1.6263769734504292e-05,
      "loss": 0.0937,
      "step": 11040
    },
    {
      "epoch": 0.560688359815219,
      "grad_norm": 0.5505645275115967,
      "learning_rate": 1.6262077601231875e-05,
      "loss": 0.1043,
      "step": 11045
    },
    {
      "epoch": 0.5609421798060815,
      "grad_norm": 0.42129865288734436,
      "learning_rate": 1.626038546795946e-05,
      "loss": 0.1049,
      "step": 11050
    },
    {
      "epoch": 0.561195999796944,
      "grad_norm": 0.5019480586051941,
      "learning_rate": 1.6258693334687042e-05,
      "loss": 0.0966,
      "step": 11055
    },
    {
      "epoch": 0.5614498197878065,
      "grad_norm": 0.44651737809181213,
      "learning_rate": 1.6257001201414623e-05,
      "loss": 0.1014,
      "step": 11060
    },
    {
      "epoch": 0.561703639778669,
      "grad_norm": 0.4271222651004791,
      "learning_rate": 1.625530906814221e-05,
      "loss": 0.105,
      "step": 11065
    },
    {
      "epoch": 0.5619574597695315,
      "grad_norm": 0.354856938123703,
      "learning_rate": 1.6253616934869793e-05,
      "loss": 0.0862,
      "step": 11070
    },
    {
      "epoch": 0.562211279760394,
      "grad_norm": 0.4536794424057007,
      "learning_rate": 1.6251924801597373e-05,
      "loss": 0.0963,
      "step": 11075
    },
    {
      "epoch": 0.5624650997512564,
      "grad_norm": 0.4090806245803833,
      "learning_rate": 1.625023266832496e-05,
      "loss": 0.0874,
      "step": 11080
    },
    {
      "epoch": 0.5627189197421189,
      "grad_norm": 0.7183617353439331,
      "learning_rate": 1.624854053505254e-05,
      "loss": 0.1116,
      "step": 11085
    },
    {
      "epoch": 0.5629727397329813,
      "grad_norm": 0.8949313759803772,
      "learning_rate": 1.6246848401780127e-05,
      "loss": 0.1,
      "step": 11090
    },
    {
      "epoch": 0.5632265597238438,
      "grad_norm": 0.47640174627304077,
      "learning_rate": 1.624515626850771e-05,
      "loss": 0.0952,
      "step": 11095
    },
    {
      "epoch": 0.5634803797147063,
      "grad_norm": 0.4236536920070648,
      "learning_rate": 1.624346413523529e-05,
      "loss": 0.0888,
      "step": 11100
    },
    {
      "epoch": 0.5637341997055688,
      "grad_norm": 0.4696132242679596,
      "learning_rate": 1.6241772001962878e-05,
      "loss": 0.0986,
      "step": 11105
    },
    {
      "epoch": 0.5639880196964313,
      "grad_norm": 0.48761463165283203,
      "learning_rate": 1.6240079868690458e-05,
      "loss": 0.086,
      "step": 11110
    },
    {
      "epoch": 0.5642418396872938,
      "grad_norm": 0.42676812410354614,
      "learning_rate": 1.623838773541804e-05,
      "loss": 0.0887,
      "step": 11115
    },
    {
      "epoch": 0.5644956596781563,
      "grad_norm": 0.5835752487182617,
      "learning_rate": 1.6236695602145625e-05,
      "loss": 0.0797,
      "step": 11120
    },
    {
      "epoch": 0.5647494796690188,
      "grad_norm": 0.39226260781288147,
      "learning_rate": 1.623500346887321e-05,
      "loss": 0.1086,
      "step": 11125
    },
    {
      "epoch": 0.5650032996598812,
      "grad_norm": 0.4616701006889343,
      "learning_rate": 1.6233311335600796e-05,
      "loss": 0.0845,
      "step": 11130
    },
    {
      "epoch": 0.5652571196507437,
      "grad_norm": 0.6852196455001831,
      "learning_rate": 1.6231619202328376e-05,
      "loss": 0.0953,
      "step": 11135
    },
    {
      "epoch": 0.5655109396416061,
      "grad_norm": 0.5007418990135193,
      "learning_rate": 1.622992706905596e-05,
      "loss": 0.1014,
      "step": 11140
    },
    {
      "epoch": 0.5657647596324686,
      "grad_norm": 0.43455320596694946,
      "learning_rate": 1.6228234935783543e-05,
      "loss": 0.0925,
      "step": 11145
    },
    {
      "epoch": 0.5660185796233311,
      "grad_norm": 0.41454216837882996,
      "learning_rate": 1.6226542802511126e-05,
      "loss": 0.0963,
      "step": 11150
    },
    {
      "epoch": 0.5662723996141936,
      "grad_norm": 0.5282880663871765,
      "learning_rate": 1.622485066923871e-05,
      "loss": 0.0926,
      "step": 11155
    },
    {
      "epoch": 0.5665262196050561,
      "grad_norm": 0.33719152212142944,
      "learning_rate": 1.6223158535966293e-05,
      "loss": 0.0983,
      "step": 11160
    },
    {
      "epoch": 0.5667800395959186,
      "grad_norm": 0.4163464605808258,
      "learning_rate": 1.6221466402693877e-05,
      "loss": 0.0993,
      "step": 11165
    },
    {
      "epoch": 0.5670338595867811,
      "grad_norm": 0.40644052624702454,
      "learning_rate": 1.621977426942146e-05,
      "loss": 0.0798,
      "step": 11170
    },
    {
      "epoch": 0.5672876795776436,
      "grad_norm": 0.42944255471229553,
      "learning_rate": 1.6218082136149044e-05,
      "loss": 0.1043,
      "step": 11175
    },
    {
      "epoch": 0.567541499568506,
      "grad_norm": 0.3915778398513794,
      "learning_rate": 1.6216390002876628e-05,
      "loss": 0.0828,
      "step": 11180
    },
    {
      "epoch": 0.5677953195593685,
      "grad_norm": 0.5200676321983337,
      "learning_rate": 1.621469786960421e-05,
      "loss": 0.0916,
      "step": 11185
    },
    {
      "epoch": 0.568049139550231,
      "grad_norm": 0.5898273587226868,
      "learning_rate": 1.6213005736331795e-05,
      "loss": 0.1157,
      "step": 11190
    },
    {
      "epoch": 0.5683029595410934,
      "grad_norm": 0.4564743638038635,
      "learning_rate": 1.6211313603059378e-05,
      "loss": 0.1059,
      "step": 11195
    },
    {
      "epoch": 0.5685567795319559,
      "grad_norm": 0.42315754294395447,
      "learning_rate": 1.6209621469786962e-05,
      "loss": 0.1021,
      "step": 11200
    },
    {
      "epoch": 0.5688105995228184,
      "grad_norm": 0.4259279668331146,
      "learning_rate": 1.6207929336514545e-05,
      "loss": 0.0893,
      "step": 11205
    },
    {
      "epoch": 0.5690644195136809,
      "grad_norm": 0.4402656555175781,
      "learning_rate": 1.620623720324213e-05,
      "loss": 0.085,
      "step": 11210
    },
    {
      "epoch": 0.5693182395045434,
      "grad_norm": 0.3866107165813446,
      "learning_rate": 1.6204545069969712e-05,
      "loss": 0.09,
      "step": 11215
    },
    {
      "epoch": 0.5695720594954059,
      "grad_norm": 0.4657836854457855,
      "learning_rate": 1.6202852936697296e-05,
      "loss": 0.0994,
      "step": 11220
    },
    {
      "epoch": 0.5698258794862683,
      "grad_norm": 0.40453943610191345,
      "learning_rate": 1.620116080342488e-05,
      "loss": 0.0987,
      "step": 11225
    },
    {
      "epoch": 0.5700796994771308,
      "grad_norm": 0.401993989944458,
      "learning_rate": 1.6199468670152463e-05,
      "loss": 0.0847,
      "step": 11230
    },
    {
      "epoch": 0.5703335194679933,
      "grad_norm": 0.5438871383666992,
      "learning_rate": 1.6197776536880047e-05,
      "loss": 0.0964,
      "step": 11235
    },
    {
      "epoch": 0.5705873394588558,
      "grad_norm": 0.5712496638298035,
      "learning_rate": 1.619608440360763e-05,
      "loss": 0.0966,
      "step": 11240
    },
    {
      "epoch": 0.5708411594497182,
      "grad_norm": 0.3064565658569336,
      "learning_rate": 1.6194392270335214e-05,
      "loss": 0.0909,
      "step": 11245
    },
    {
      "epoch": 0.5710949794405807,
      "grad_norm": 1.4858416318893433,
      "learning_rate": 1.6192700137062797e-05,
      "loss": 0.1,
      "step": 11250
    },
    {
      "epoch": 0.5713487994314432,
      "grad_norm": 0.36369451880455017,
      "learning_rate": 1.619100800379038e-05,
      "loss": 0.0975,
      "step": 11255
    },
    {
      "epoch": 0.5716026194223057,
      "grad_norm": 0.48280322551727295,
      "learning_rate": 1.6189315870517964e-05,
      "loss": 0.0852,
      "step": 11260
    },
    {
      "epoch": 0.5718564394131682,
      "grad_norm": 0.708537220954895,
      "learning_rate": 1.6187623737245545e-05,
      "loss": 0.1065,
      "step": 11265
    },
    {
      "epoch": 0.5721102594040307,
      "grad_norm": 0.5387185215950012,
      "learning_rate": 1.618593160397313e-05,
      "loss": 0.0971,
      "step": 11270
    },
    {
      "epoch": 0.5723640793948931,
      "grad_norm": 0.5026307106018066,
      "learning_rate": 1.6184239470700715e-05,
      "loss": 0.0887,
      "step": 11275
    },
    {
      "epoch": 0.5726178993857556,
      "grad_norm": 0.40268126130104065,
      "learning_rate": 1.61825473374283e-05,
      "loss": 0.0892,
      "step": 11280
    },
    {
      "epoch": 0.5728717193766181,
      "grad_norm": 0.5645183324813843,
      "learning_rate": 1.6180855204155882e-05,
      "loss": 0.0984,
      "step": 11285
    },
    {
      "epoch": 0.5731255393674806,
      "grad_norm": 0.35205450654029846,
      "learning_rate": 1.6179163070883462e-05,
      "loss": 0.1013,
      "step": 11290
    },
    {
      "epoch": 0.573379359358343,
      "grad_norm": 0.42374926805496216,
      "learning_rate": 1.617747093761105e-05,
      "loss": 0.0954,
      "step": 11295
    },
    {
      "epoch": 0.5736331793492055,
      "grad_norm": 0.438722163438797,
      "learning_rate": 1.617577880433863e-05,
      "loss": 0.0961,
      "step": 11300
    },
    {
      "epoch": 0.573886999340068,
      "grad_norm": 0.6137442588806152,
      "learning_rate": 1.6174086671066213e-05,
      "loss": 0.0918,
      "step": 11305
    },
    {
      "epoch": 0.5741408193309305,
      "grad_norm": 0.6508979201316833,
      "learning_rate": 1.61723945377938e-05,
      "loss": 0.0934,
      "step": 11310
    },
    {
      "epoch": 0.574394639321793,
      "grad_norm": 0.6937998533248901,
      "learning_rate": 1.617070240452138e-05,
      "loss": 0.104,
      "step": 11315
    },
    {
      "epoch": 0.5746484593126555,
      "grad_norm": 0.44844573736190796,
      "learning_rate": 1.6169010271248964e-05,
      "loss": 0.1024,
      "step": 11320
    },
    {
      "epoch": 0.5749022793035179,
      "grad_norm": 0.5491892695426941,
      "learning_rate": 1.6167318137976547e-05,
      "loss": 0.1098,
      "step": 11325
    },
    {
      "epoch": 0.5751560992943804,
      "grad_norm": 0.43779659271240234,
      "learning_rate": 1.616562600470413e-05,
      "loss": 0.0993,
      "step": 11330
    },
    {
      "epoch": 0.5754099192852429,
      "grad_norm": 0.3295450508594513,
      "learning_rate": 1.6163933871431718e-05,
      "loss": 0.0859,
      "step": 11335
    },
    {
      "epoch": 0.5756637392761054,
      "grad_norm": 0.43296727538108826,
      "learning_rate": 1.6162241738159298e-05,
      "loss": 0.0926,
      "step": 11340
    },
    {
      "epoch": 0.5759175592669679,
      "grad_norm": 0.5013937950134277,
      "learning_rate": 1.616054960488688e-05,
      "loss": 0.0917,
      "step": 11345
    },
    {
      "epoch": 0.5761713792578304,
      "grad_norm": 0.4258970022201538,
      "learning_rate": 1.6158857471614465e-05,
      "loss": 0.1014,
      "step": 11350
    },
    {
      "epoch": 0.5764251992486928,
      "grad_norm": 0.4371802806854248,
      "learning_rate": 1.615716533834205e-05,
      "loss": 0.0894,
      "step": 11355
    },
    {
      "epoch": 0.5766790192395553,
      "grad_norm": 0.34599819779396057,
      "learning_rate": 1.6155473205069632e-05,
      "loss": 0.0877,
      "step": 11360
    },
    {
      "epoch": 0.5769328392304178,
      "grad_norm": 0.3687082827091217,
      "learning_rate": 1.6153781071797215e-05,
      "loss": 0.0881,
      "step": 11365
    },
    {
      "epoch": 0.5771866592212803,
      "grad_norm": 0.45477935671806335,
      "learning_rate": 1.61520889385248e-05,
      "loss": 0.0896,
      "step": 11370
    },
    {
      "epoch": 0.5774404792121427,
      "grad_norm": 0.4007332921028137,
      "learning_rate": 1.6150396805252383e-05,
      "loss": 0.0888,
      "step": 11375
    },
    {
      "epoch": 0.5776942992030052,
      "grad_norm": 0.44187676906585693,
      "learning_rate": 1.6148704671979966e-05,
      "loss": 0.0953,
      "step": 11380
    },
    {
      "epoch": 0.5779481191938677,
      "grad_norm": 0.3938285708427429,
      "learning_rate": 1.614701253870755e-05,
      "loss": 0.0935,
      "step": 11385
    },
    {
      "epoch": 0.5782019391847302,
      "grad_norm": 0.4192904531955719,
      "learning_rate": 1.6145320405435133e-05,
      "loss": 0.0982,
      "step": 11390
    },
    {
      "epoch": 0.5784557591755927,
      "grad_norm": 0.3519038259983063,
      "learning_rate": 1.6143628272162717e-05,
      "loss": 0.095,
      "step": 11395
    },
    {
      "epoch": 0.5787095791664552,
      "grad_norm": 0.5053529143333435,
      "learning_rate": 1.61419361388903e-05,
      "loss": 0.1037,
      "step": 11400
    },
    {
      "epoch": 0.5789633991573176,
      "grad_norm": 0.5747759938240051,
      "learning_rate": 1.6140244005617884e-05,
      "loss": 0.0861,
      "step": 11405
    },
    {
      "epoch": 0.5792172191481801,
      "grad_norm": 0.5155202150344849,
      "learning_rate": 1.6138551872345467e-05,
      "loss": 0.0859,
      "step": 11410
    },
    {
      "epoch": 0.5794710391390426,
      "grad_norm": 0.799940288066864,
      "learning_rate": 1.613685973907305e-05,
      "loss": 0.0922,
      "step": 11415
    },
    {
      "epoch": 0.5797248591299051,
      "grad_norm": 0.29105183482170105,
      "learning_rate": 1.6135167605800634e-05,
      "loss": 0.0794,
      "step": 11420
    },
    {
      "epoch": 0.5799786791207675,
      "grad_norm": 0.43824175000190735,
      "learning_rate": 1.6133475472528218e-05,
      "loss": 0.1006,
      "step": 11425
    },
    {
      "epoch": 0.58023249911163,
      "grad_norm": 0.8796586990356445,
      "learning_rate": 1.61317833392558e-05,
      "loss": 0.0999,
      "step": 11430
    },
    {
      "epoch": 0.5804863191024925,
      "grad_norm": 0.4385398328304291,
      "learning_rate": 1.6130091205983385e-05,
      "loss": 0.0962,
      "step": 11435
    },
    {
      "epoch": 0.580740139093355,
      "grad_norm": 0.4407016336917877,
      "learning_rate": 1.612839907271097e-05,
      "loss": 0.0962,
      "step": 11440
    },
    {
      "epoch": 0.5809939590842175,
      "grad_norm": 0.37328964471817017,
      "learning_rate": 1.6126706939438552e-05,
      "loss": 0.0866,
      "step": 11445
    },
    {
      "epoch": 0.58124777907508,
      "grad_norm": 0.42281779646873474,
      "learning_rate": 1.6125014806166136e-05,
      "loss": 0.1012,
      "step": 11450
    },
    {
      "epoch": 0.5815015990659425,
      "grad_norm": 0.459942102432251,
      "learning_rate": 1.612332267289372e-05,
      "loss": 0.1061,
      "step": 11455
    },
    {
      "epoch": 0.581755419056805,
      "grad_norm": 0.6190848350524902,
      "learning_rate": 1.6121630539621303e-05,
      "loss": 0.1014,
      "step": 11460
    },
    {
      "epoch": 0.5820092390476674,
      "grad_norm": 0.6544129252433777,
      "learning_rate": 1.6119938406348886e-05,
      "loss": 0.0861,
      "step": 11465
    },
    {
      "epoch": 0.5822630590385299,
      "grad_norm": 0.32813870906829834,
      "learning_rate": 1.6118246273076466e-05,
      "loss": 0.0926,
      "step": 11470
    },
    {
      "epoch": 0.5825168790293923,
      "grad_norm": 0.4258780777454376,
      "learning_rate": 1.6116554139804053e-05,
      "loss": 0.0933,
      "step": 11475
    },
    {
      "epoch": 0.5827706990202548,
      "grad_norm": 0.6019533276557922,
      "learning_rate": 1.6114862006531634e-05,
      "loss": 0.0953,
      "step": 11480
    },
    {
      "epoch": 0.5830245190111173,
      "grad_norm": 0.4072204828262329,
      "learning_rate": 1.611316987325922e-05,
      "loss": 0.0951,
      "step": 11485
    },
    {
      "epoch": 0.5832783390019798,
      "grad_norm": 0.3429082930088043,
      "learning_rate": 1.6111477739986804e-05,
      "loss": 0.0879,
      "step": 11490
    },
    {
      "epoch": 0.5835321589928423,
      "grad_norm": 0.4086165428161621,
      "learning_rate": 1.6109785606714384e-05,
      "loss": 0.0996,
      "step": 11495
    },
    {
      "epoch": 0.5837859789837048,
      "grad_norm": 0.3771967589855194,
      "learning_rate": 1.610809347344197e-05,
      "loss": 0.0959,
      "step": 11500
    },
    {
      "epoch": 0.5840397989745673,
      "grad_norm": 0.4078262746334076,
      "learning_rate": 1.610640134016955e-05,
      "loss": 0.0953,
      "step": 11505
    },
    {
      "epoch": 0.5842936189654298,
      "grad_norm": 1.961870789527893,
      "learning_rate": 1.6104709206897135e-05,
      "loss": 0.0964,
      "step": 11510
    },
    {
      "epoch": 0.5845474389562922,
      "grad_norm": 0.4825584888458252,
      "learning_rate": 1.6103017073624722e-05,
      "loss": 0.0995,
      "step": 11515
    },
    {
      "epoch": 0.5848012589471546,
      "grad_norm": 0.46092280745506287,
      "learning_rate": 1.6101324940352302e-05,
      "loss": 0.085,
      "step": 11520
    },
    {
      "epoch": 0.5850550789380171,
      "grad_norm": 0.5379409193992615,
      "learning_rate": 1.609963280707989e-05,
      "loss": 0.1011,
      "step": 11525
    },
    {
      "epoch": 0.5853088989288796,
      "grad_norm": 0.9948629140853882,
      "learning_rate": 1.609794067380747e-05,
      "loss": 0.0914,
      "step": 11530
    },
    {
      "epoch": 0.5855627189197421,
      "grad_norm": 0.39759644865989685,
      "learning_rate": 1.6096248540535053e-05,
      "loss": 0.0939,
      "step": 11535
    },
    {
      "epoch": 0.5858165389106046,
      "grad_norm": 0.4798009693622589,
      "learning_rate": 1.609455640726264e-05,
      "loss": 0.0927,
      "step": 11540
    },
    {
      "epoch": 0.5860703589014671,
      "grad_norm": 0.3862641453742981,
      "learning_rate": 1.609286427399022e-05,
      "loss": 0.0899,
      "step": 11545
    },
    {
      "epoch": 0.5863241788923296,
      "grad_norm": 0.38452666997909546,
      "learning_rate": 1.6091172140717803e-05,
      "loss": 0.0959,
      "step": 11550
    },
    {
      "epoch": 0.5865779988831921,
      "grad_norm": 0.3630549907684326,
      "learning_rate": 1.6089480007445387e-05,
      "loss": 0.0995,
      "step": 11555
    },
    {
      "epoch": 0.5868318188740546,
      "grad_norm": 0.8909419178962708,
      "learning_rate": 1.608778787417297e-05,
      "loss": 0.1095,
      "step": 11560
    },
    {
      "epoch": 0.587085638864917,
      "grad_norm": 0.3947525918483734,
      "learning_rate": 1.6086095740900554e-05,
      "loss": 0.099,
      "step": 11565
    },
    {
      "epoch": 0.5873394588557794,
      "grad_norm": 0.3510633409023285,
      "learning_rate": 1.6084403607628137e-05,
      "loss": 0.0885,
      "step": 11570
    },
    {
      "epoch": 0.5875932788466419,
      "grad_norm": 0.49123838543891907,
      "learning_rate": 1.608271147435572e-05,
      "loss": 0.0867,
      "step": 11575
    },
    {
      "epoch": 0.5878470988375044,
      "grad_norm": 0.3366958200931549,
      "learning_rate": 1.6081019341083304e-05,
      "loss": 0.0974,
      "step": 11580
    },
    {
      "epoch": 0.5881009188283669,
      "grad_norm": 0.7465676665306091,
      "learning_rate": 1.6079327207810888e-05,
      "loss": 0.1032,
      "step": 11585
    },
    {
      "epoch": 0.5883547388192294,
      "grad_norm": 0.38815176486968994,
      "learning_rate": 1.607763507453847e-05,
      "loss": 0.1028,
      "step": 11590
    },
    {
      "epoch": 0.5886085588100919,
      "grad_norm": 0.42240670323371887,
      "learning_rate": 1.6075942941266055e-05,
      "loss": 0.086,
      "step": 11595
    },
    {
      "epoch": 0.5888623788009544,
      "grad_norm": 0.35436907410621643,
      "learning_rate": 1.607425080799364e-05,
      "loss": 0.095,
      "step": 11600
    },
    {
      "epoch": 0.5891161987918169,
      "grad_norm": 0.3987736999988556,
      "learning_rate": 1.6072558674721222e-05,
      "loss": 0.0994,
      "step": 11605
    },
    {
      "epoch": 0.5893700187826794,
      "grad_norm": 0.5091561079025269,
      "learning_rate": 1.6070866541448806e-05,
      "loss": 0.0942,
      "step": 11610
    },
    {
      "epoch": 0.5896238387735419,
      "grad_norm": 0.3973734974861145,
      "learning_rate": 1.606917440817639e-05,
      "loss": 0.0852,
      "step": 11615
    },
    {
      "epoch": 0.5898776587644042,
      "grad_norm": 0.38369306921958923,
      "learning_rate": 1.6067482274903973e-05,
      "loss": 0.0904,
      "step": 11620
    },
    {
      "epoch": 0.5901314787552667,
      "grad_norm": 0.41540855169296265,
      "learning_rate": 1.6065790141631556e-05,
      "loss": 0.083,
      "step": 11625
    },
    {
      "epoch": 0.5903852987461292,
      "grad_norm": 0.468839168548584,
      "learning_rate": 1.606409800835914e-05,
      "loss": 0.1061,
      "step": 11630
    },
    {
      "epoch": 0.5906391187369917,
      "grad_norm": 0.5892307162284851,
      "learning_rate": 1.6062405875086723e-05,
      "loss": 0.0913,
      "step": 11635
    },
    {
      "epoch": 0.5908929387278542,
      "grad_norm": 0.3928740322589874,
      "learning_rate": 1.6060713741814307e-05,
      "loss": 0.0921,
      "step": 11640
    },
    {
      "epoch": 0.5911467587187167,
      "grad_norm": 0.36110663414001465,
      "learning_rate": 1.605902160854189e-05,
      "loss": 0.0927,
      "step": 11645
    },
    {
      "epoch": 0.5914005787095792,
      "grad_norm": 0.699163019657135,
      "learning_rate": 1.6057329475269474e-05,
      "loss": 0.0997,
      "step": 11650
    },
    {
      "epoch": 0.5916543987004417,
      "grad_norm": 0.43201932311058044,
      "learning_rate": 1.6055637341997058e-05,
      "loss": 0.0863,
      "step": 11655
    },
    {
      "epoch": 0.5919082186913042,
      "grad_norm": 1.3616560697555542,
      "learning_rate": 1.6053945208724638e-05,
      "loss": 0.0943,
      "step": 11660
    },
    {
      "epoch": 0.5921620386821667,
      "grad_norm": 0.44065532088279724,
      "learning_rate": 1.6052253075452225e-05,
      "loss": 0.1008,
      "step": 11665
    },
    {
      "epoch": 0.592415858673029,
      "grad_norm": 0.48787134885787964,
      "learning_rate": 1.6050560942179808e-05,
      "loss": 0.0988,
      "step": 11670
    },
    {
      "epoch": 0.5926696786638915,
      "grad_norm": 0.502856969833374,
      "learning_rate": 1.6048868808907392e-05,
      "loss": 0.093,
      "step": 11675
    },
    {
      "epoch": 0.592923498654754,
      "grad_norm": 0.46943750977516174,
      "learning_rate": 1.6047176675634975e-05,
      "loss": 0.0912,
      "step": 11680
    },
    {
      "epoch": 0.5931773186456165,
      "grad_norm": 0.37175118923187256,
      "learning_rate": 1.6045484542362555e-05,
      "loss": 0.0861,
      "step": 11685
    },
    {
      "epoch": 0.593431138636479,
      "grad_norm": 0.5893447995185852,
      "learning_rate": 1.6043792409090142e-05,
      "loss": 0.0894,
      "step": 11690
    },
    {
      "epoch": 0.5936849586273415,
      "grad_norm": 0.4342266619205475,
      "learning_rate": 1.6042100275817726e-05,
      "loss": 0.0919,
      "step": 11695
    },
    {
      "epoch": 0.593938778618204,
      "grad_norm": 0.3848216235637665,
      "learning_rate": 1.6040408142545306e-05,
      "loss": 0.081,
      "step": 11700
    },
    {
      "epoch": 0.5941925986090665,
      "grad_norm": 0.4635235667228699,
      "learning_rate": 1.6038716009272893e-05,
      "loss": 0.0899,
      "step": 11705
    },
    {
      "epoch": 0.594446418599929,
      "grad_norm": 0.5500293374061584,
      "learning_rate": 1.6037023876000473e-05,
      "loss": 0.0902,
      "step": 11710
    },
    {
      "epoch": 0.5947002385907915,
      "grad_norm": 0.3731316924095154,
      "learning_rate": 1.6035331742728057e-05,
      "loss": 0.0846,
      "step": 11715
    },
    {
      "epoch": 0.5949540585816538,
      "grad_norm": 0.3542387783527374,
      "learning_rate": 1.6033639609455644e-05,
      "loss": 0.0959,
      "step": 11720
    },
    {
      "epoch": 0.5952078785725163,
      "grad_norm": 0.4823991656303406,
      "learning_rate": 1.6031947476183224e-05,
      "loss": 0.0959,
      "step": 11725
    },
    {
      "epoch": 0.5954616985633788,
      "grad_norm": 0.4015558063983917,
      "learning_rate": 1.603025534291081e-05,
      "loss": 0.0983,
      "step": 11730
    },
    {
      "epoch": 0.5957155185542413,
      "grad_norm": 0.4608984589576721,
      "learning_rate": 1.602856320963839e-05,
      "loss": 0.0885,
      "step": 11735
    },
    {
      "epoch": 0.5959693385451038,
      "grad_norm": 0.4496300220489502,
      "learning_rate": 1.6026871076365974e-05,
      "loss": 0.0854,
      "step": 11740
    },
    {
      "epoch": 0.5962231585359663,
      "grad_norm": 0.41843149065971375,
      "learning_rate": 1.602517894309356e-05,
      "loss": 0.0882,
      "step": 11745
    },
    {
      "epoch": 0.5964769785268288,
      "grad_norm": 0.3742343783378601,
      "learning_rate": 1.602348680982114e-05,
      "loss": 0.0829,
      "step": 11750
    },
    {
      "epoch": 0.5967307985176913,
      "grad_norm": 0.5669666528701782,
      "learning_rate": 1.6021794676548725e-05,
      "loss": 0.0933,
      "step": 11755
    },
    {
      "epoch": 0.5969846185085538,
      "grad_norm": 0.37627196311950684,
      "learning_rate": 1.602010254327631e-05,
      "loss": 0.0797,
      "step": 11760
    },
    {
      "epoch": 0.5972384384994163,
      "grad_norm": 0.4670741856098175,
      "learning_rate": 1.6018410410003892e-05,
      "loss": 0.0964,
      "step": 11765
    },
    {
      "epoch": 0.5974922584902786,
      "grad_norm": 0.5946935415267944,
      "learning_rate": 1.601671827673148e-05,
      "loss": 0.0848,
      "step": 11770
    },
    {
      "epoch": 0.5977460784811411,
      "grad_norm": 0.3806981146335602,
      "learning_rate": 1.601502614345906e-05,
      "loss": 0.0901,
      "step": 11775
    },
    {
      "epoch": 0.5979998984720036,
      "grad_norm": 0.4813790023326874,
      "learning_rate": 1.6013334010186643e-05,
      "loss": 0.0897,
      "step": 11780
    },
    {
      "epoch": 0.5982537184628661,
      "grad_norm": 0.33864712715148926,
      "learning_rate": 1.6011641876914226e-05,
      "loss": 0.0869,
      "step": 11785
    },
    {
      "epoch": 0.5985075384537286,
      "grad_norm": 0.4564129114151001,
      "learning_rate": 1.600994974364181e-05,
      "loss": 0.089,
      "step": 11790
    },
    {
      "epoch": 0.5987613584445911,
      "grad_norm": 0.3900724947452545,
      "learning_rate": 1.6008257610369393e-05,
      "loss": 0.0748,
      "step": 11795
    },
    {
      "epoch": 0.5990151784354536,
      "grad_norm": 0.34131884574890137,
      "learning_rate": 1.6006565477096977e-05,
      "loss": 0.0985,
      "step": 11800
    },
    {
      "epoch": 0.5992689984263161,
      "grad_norm": 0.6740922331809998,
      "learning_rate": 1.600487334382456e-05,
      "loss": 0.0969,
      "step": 11805
    },
    {
      "epoch": 0.5995228184171786,
      "grad_norm": 0.41970062255859375,
      "learning_rate": 1.6003181210552144e-05,
      "loss": 0.0924,
      "step": 11810
    },
    {
      "epoch": 0.599776638408041,
      "grad_norm": 0.40490224957466125,
      "learning_rate": 1.6001489077279728e-05,
      "loss": 0.0926,
      "step": 11815
    },
    {
      "epoch": 0.6000304583989035,
      "grad_norm": 0.9713094234466553,
      "learning_rate": 1.599979694400731e-05,
      "loss": 0.1022,
      "step": 11820
    },
    {
      "epoch": 0.600284278389766,
      "grad_norm": 0.3870560824871063,
      "learning_rate": 1.5998104810734895e-05,
      "loss": 0.0827,
      "step": 11825
    },
    {
      "epoch": 0.6005380983806284,
      "grad_norm": 0.44513827562332153,
      "learning_rate": 1.5996412677462478e-05,
      "loss": 0.0932,
      "step": 11830
    },
    {
      "epoch": 0.6007919183714909,
      "grad_norm": 0.42367058992385864,
      "learning_rate": 1.5994720544190062e-05,
      "loss": 0.0917,
      "step": 11835
    },
    {
      "epoch": 0.6010457383623534,
      "grad_norm": 0.5602151155471802,
      "learning_rate": 1.5993028410917645e-05,
      "loss": 0.099,
      "step": 11840
    },
    {
      "epoch": 0.6012995583532159,
      "grad_norm": 0.43050816655158997,
      "learning_rate": 1.599133627764523e-05,
      "loss": 0.0836,
      "step": 11845
    },
    {
      "epoch": 0.6015533783440784,
      "grad_norm": 0.34673821926116943,
      "learning_rate": 1.5989644144372812e-05,
      "loss": 0.0992,
      "step": 11850
    },
    {
      "epoch": 0.6018071983349409,
      "grad_norm": 0.3608914911746979,
      "learning_rate": 1.5987952011100396e-05,
      "loss": 0.0941,
      "step": 11855
    },
    {
      "epoch": 0.6020610183258034,
      "grad_norm": 0.4284670352935791,
      "learning_rate": 1.598625987782798e-05,
      "loss": 0.0925,
      "step": 11860
    },
    {
      "epoch": 0.6023148383166658,
      "grad_norm": 0.5360268950462341,
      "learning_rate": 1.5984567744555563e-05,
      "loss": 0.0936,
      "step": 11865
    },
    {
      "epoch": 0.6025686583075283,
      "grad_norm": 0.34779566526412964,
      "learning_rate": 1.5982875611283147e-05,
      "loss": 0.0871,
      "step": 11870
    },
    {
      "epoch": 0.6028224782983908,
      "grad_norm": 0.43765851855278015,
      "learning_rate": 1.598118347801073e-05,
      "loss": 0.0914,
      "step": 11875
    },
    {
      "epoch": 0.6030762982892532,
      "grad_norm": 0.5663272142410278,
      "learning_rate": 1.5979491344738314e-05,
      "loss": 0.084,
      "step": 11880
    },
    {
      "epoch": 0.6033301182801157,
      "grad_norm": 0.4468323290348053,
      "learning_rate": 1.5977799211465897e-05,
      "loss": 0.0927,
      "step": 11885
    },
    {
      "epoch": 0.6035839382709782,
      "grad_norm": 0.6487022042274475,
      "learning_rate": 1.5976107078193477e-05,
      "loss": 0.0985,
      "step": 11890
    },
    {
      "epoch": 0.6038377582618407,
      "grad_norm": 0.5686355233192444,
      "learning_rate": 1.5974414944921064e-05,
      "loss": 0.1071,
      "step": 11895
    },
    {
      "epoch": 0.6040915782527032,
      "grad_norm": 0.4146348237991333,
      "learning_rate": 1.5972722811648648e-05,
      "loss": 0.0844,
      "step": 11900
    },
    {
      "epoch": 0.6043453982435657,
      "grad_norm": 0.37118101119995117,
      "learning_rate": 1.5971030678376228e-05,
      "loss": 0.0891,
      "step": 11905
    },
    {
      "epoch": 0.6045992182344282,
      "grad_norm": 0.4417383670806885,
      "learning_rate": 1.5969338545103815e-05,
      "loss": 0.0932,
      "step": 11910
    },
    {
      "epoch": 0.6048530382252906,
      "grad_norm": 0.4875818192958832,
      "learning_rate": 1.5967646411831395e-05,
      "loss": 0.0851,
      "step": 11915
    },
    {
      "epoch": 0.6051068582161531,
      "grad_norm": 0.4297184944152832,
      "learning_rate": 1.5965954278558982e-05,
      "loss": 0.0887,
      "step": 11920
    },
    {
      "epoch": 0.6053606782070156,
      "grad_norm": 0.5882421135902405,
      "learning_rate": 1.5964262145286566e-05,
      "loss": 0.0828,
      "step": 11925
    },
    {
      "epoch": 0.605614498197878,
      "grad_norm": 0.4121566712856293,
      "learning_rate": 1.5962570012014146e-05,
      "loss": 0.092,
      "step": 11930
    },
    {
      "epoch": 0.6058683181887405,
      "grad_norm": 0.4164383113384247,
      "learning_rate": 1.5960877878741733e-05,
      "loss": 0.0916,
      "step": 11935
    },
    {
      "epoch": 0.606122138179603,
      "grad_norm": 0.7164746522903442,
      "learning_rate": 1.5959185745469313e-05,
      "loss": 0.0945,
      "step": 11940
    },
    {
      "epoch": 0.6063759581704655,
      "grad_norm": 0.3977768123149872,
      "learning_rate": 1.5957493612196896e-05,
      "loss": 0.0871,
      "step": 11945
    },
    {
      "epoch": 0.606629778161328,
      "grad_norm": 0.5656052827835083,
      "learning_rate": 1.5955801478924483e-05,
      "loss": 0.0728,
      "step": 11950
    },
    {
      "epoch": 0.6068835981521905,
      "grad_norm": 0.6886550188064575,
      "learning_rate": 1.5954109345652064e-05,
      "loss": 0.0971,
      "step": 11955
    },
    {
      "epoch": 0.607137418143053,
      "grad_norm": 0.30135658383369446,
      "learning_rate": 1.5952417212379647e-05,
      "loss": 0.097,
      "step": 11960
    },
    {
      "epoch": 0.6073912381339154,
      "grad_norm": 0.6407579779624939,
      "learning_rate": 1.595072507910723e-05,
      "loss": 0.0938,
      "step": 11965
    },
    {
      "epoch": 0.6076450581247779,
      "grad_norm": 0.3917185068130493,
      "learning_rate": 1.5949032945834814e-05,
      "loss": 0.0985,
      "step": 11970
    },
    {
      "epoch": 0.6078988781156404,
      "grad_norm": 0.38503170013427734,
      "learning_rate": 1.59473408125624e-05,
      "loss": 0.0808,
      "step": 11975
    },
    {
      "epoch": 0.6081526981065029,
      "grad_norm": 0.856484591960907,
      "learning_rate": 1.594564867928998e-05,
      "loss": 0.092,
      "step": 11980
    },
    {
      "epoch": 0.6084065180973653,
      "grad_norm": 0.538896381855011,
      "learning_rate": 1.5943956546017565e-05,
      "loss": 0.0954,
      "step": 11985
    },
    {
      "epoch": 0.6086603380882278,
      "grad_norm": 0.5355377793312073,
      "learning_rate": 1.594226441274515e-05,
      "loss": 0.0904,
      "step": 11990
    },
    {
      "epoch": 0.6089141580790903,
      "grad_norm": 0.5098680853843689,
      "learning_rate": 1.5940572279472732e-05,
      "loss": 0.0823,
      "step": 11995
    },
    {
      "epoch": 0.6091679780699528,
      "grad_norm": 0.4084436595439911,
      "learning_rate": 1.5938880146200315e-05,
      "loss": 0.082,
      "step": 12000
    },
    {
      "epoch": 0.6094217980608153,
      "grad_norm": 0.3876837193965912,
      "learning_rate": 1.59371880129279e-05,
      "loss": 0.1021,
      "step": 12005
    },
    {
      "epoch": 0.6096756180516778,
      "grad_norm": 0.4292503297328949,
      "learning_rate": 1.5935495879655483e-05,
      "loss": 0.079,
      "step": 12010
    },
    {
      "epoch": 0.6099294380425402,
      "grad_norm": 0.533645510673523,
      "learning_rate": 1.5933803746383066e-05,
      "loss": 0.0962,
      "step": 12015
    },
    {
      "epoch": 0.6101832580334027,
      "grad_norm": 0.36521783471107483,
      "learning_rate": 1.593211161311065e-05,
      "loss": 0.0956,
      "step": 12020
    },
    {
      "epoch": 0.6104370780242652,
      "grad_norm": 0.3349459767341614,
      "learning_rate": 1.5930419479838233e-05,
      "loss": 0.0904,
      "step": 12025
    },
    {
      "epoch": 0.6106908980151277,
      "grad_norm": 0.43317875266075134,
      "learning_rate": 1.5928727346565817e-05,
      "loss": 0.0913,
      "step": 12030
    },
    {
      "epoch": 0.6109447180059902,
      "grad_norm": 0.4873206317424774,
      "learning_rate": 1.59270352132934e-05,
      "loss": 0.0916,
      "step": 12035
    },
    {
      "epoch": 0.6111985379968526,
      "grad_norm": 0.38802623748779297,
      "learning_rate": 1.5925343080020984e-05,
      "loss": 0.0906,
      "step": 12040
    },
    {
      "epoch": 0.6114523579877151,
      "grad_norm": 0.5826549530029297,
      "learning_rate": 1.5923650946748567e-05,
      "loss": 0.0934,
      "step": 12045
    },
    {
      "epoch": 0.6117061779785776,
      "grad_norm": 0.5934155583381653,
      "learning_rate": 1.592195881347615e-05,
      "loss": 0.0948,
      "step": 12050
    },
    {
      "epoch": 0.6119599979694401,
      "grad_norm": 0.5034151077270508,
      "learning_rate": 1.5920266680203734e-05,
      "loss": 0.0793,
      "step": 12055
    },
    {
      "epoch": 0.6122138179603026,
      "grad_norm": 0.5423274040222168,
      "learning_rate": 1.5918574546931318e-05,
      "loss": 0.0899,
      "step": 12060
    },
    {
      "epoch": 0.612467637951165,
      "grad_norm": 0.6164039373397827,
      "learning_rate": 1.59168824136589e-05,
      "loss": 0.0932,
      "step": 12065
    },
    {
      "epoch": 0.6127214579420275,
      "grad_norm": 0.5863872766494751,
      "learning_rate": 1.5915190280386485e-05,
      "loss": 0.0858,
      "step": 12070
    },
    {
      "epoch": 0.61297527793289,
      "grad_norm": 0.3876388967037201,
      "learning_rate": 1.591349814711407e-05,
      "loss": 0.0785,
      "step": 12075
    },
    {
      "epoch": 0.6132290979237525,
      "grad_norm": 0.7438451051712036,
      "learning_rate": 1.5911806013841652e-05,
      "loss": 0.098,
      "step": 12080
    },
    {
      "epoch": 0.613482917914615,
      "grad_norm": 0.48748913407325745,
      "learning_rate": 1.5910113880569236e-05,
      "loss": 0.0984,
      "step": 12085
    },
    {
      "epoch": 0.6137367379054774,
      "grad_norm": 1.2986968755722046,
      "learning_rate": 1.590842174729682e-05,
      "loss": 0.0888,
      "step": 12090
    },
    {
      "epoch": 0.6139905578963399,
      "grad_norm": 0.525091826915741,
      "learning_rate": 1.59067296140244e-05,
      "loss": 0.0947,
      "step": 12095
    },
    {
      "epoch": 0.6142443778872024,
      "grad_norm": 0.42398321628570557,
      "learning_rate": 1.5905037480751986e-05,
      "loss": 0.0882,
      "step": 12100
    },
    {
      "epoch": 0.6144981978780649,
      "grad_norm": 0.5521324276924133,
      "learning_rate": 1.590334534747957e-05,
      "loss": 0.0885,
      "step": 12105
    },
    {
      "epoch": 0.6147520178689273,
      "grad_norm": 0.5781256556510925,
      "learning_rate": 1.5901653214207153e-05,
      "loss": 0.1041,
      "step": 12110
    },
    {
      "epoch": 0.6150058378597898,
      "grad_norm": 0.8404202461242676,
      "learning_rate": 1.5899961080934737e-05,
      "loss": 0.0966,
      "step": 12115
    },
    {
      "epoch": 0.6152596578506523,
      "grad_norm": 0.5177925229072571,
      "learning_rate": 1.5898268947662317e-05,
      "loss": 0.0846,
      "step": 12120
    },
    {
      "epoch": 0.6155134778415148,
      "grad_norm": 0.4168604910373688,
      "learning_rate": 1.5896576814389904e-05,
      "loss": 0.0937,
      "step": 12125
    },
    {
      "epoch": 0.6157672978323773,
      "grad_norm": 0.3698541820049286,
      "learning_rate": 1.5894884681117488e-05,
      "loss": 0.0905,
      "step": 12130
    },
    {
      "epoch": 0.6160211178232398,
      "grad_norm": 0.43006083369255066,
      "learning_rate": 1.5893192547845068e-05,
      "loss": 0.0896,
      "step": 12135
    },
    {
      "epoch": 0.6162749378141023,
      "grad_norm": 0.7461796402931213,
      "learning_rate": 1.5891500414572655e-05,
      "loss": 0.0853,
      "step": 12140
    },
    {
      "epoch": 0.6165287578049647,
      "grad_norm": 0.3485049307346344,
      "learning_rate": 1.5889808281300235e-05,
      "loss": 0.0941,
      "step": 12145
    },
    {
      "epoch": 0.6167825777958272,
      "grad_norm": 0.3577573299407959,
      "learning_rate": 1.588811614802782e-05,
      "loss": 0.0934,
      "step": 12150
    },
    {
      "epoch": 0.6170363977866897,
      "grad_norm": 0.43190285563468933,
      "learning_rate": 1.5886424014755405e-05,
      "loss": 0.0791,
      "step": 12155
    },
    {
      "epoch": 0.6172902177775521,
      "grad_norm": 0.3519067168235779,
      "learning_rate": 1.5884731881482985e-05,
      "loss": 0.0942,
      "step": 12160
    },
    {
      "epoch": 0.6175440377684146,
      "grad_norm": 0.35261034965515137,
      "learning_rate": 1.5883039748210572e-05,
      "loss": 0.0882,
      "step": 12165
    },
    {
      "epoch": 0.6177978577592771,
      "grad_norm": 0.39754313230514526,
      "learning_rate": 1.5881347614938153e-05,
      "loss": 0.0867,
      "step": 12170
    },
    {
      "epoch": 0.6180516777501396,
      "grad_norm": 0.5548686981201172,
      "learning_rate": 1.5879655481665736e-05,
      "loss": 0.099,
      "step": 12175
    },
    {
      "epoch": 0.6183054977410021,
      "grad_norm": 0.526914656162262,
      "learning_rate": 1.5877963348393323e-05,
      "loss": 0.087,
      "step": 12180
    },
    {
      "epoch": 0.6185593177318646,
      "grad_norm": 0.6457729339599609,
      "learning_rate": 1.5876271215120903e-05,
      "loss": 0.0894,
      "step": 12185
    },
    {
      "epoch": 0.6188131377227271,
      "grad_norm": 0.8887588381767273,
      "learning_rate": 1.5874579081848487e-05,
      "loss": 0.0885,
      "step": 12190
    },
    {
      "epoch": 0.6190669577135895,
      "grad_norm": 0.6914089918136597,
      "learning_rate": 1.587288694857607e-05,
      "loss": 0.0858,
      "step": 12195
    },
    {
      "epoch": 0.619320777704452,
      "grad_norm": 0.489515095949173,
      "learning_rate": 1.5871194815303654e-05,
      "loss": 0.0879,
      "step": 12200
    },
    {
      "epoch": 0.6195745976953145,
      "grad_norm": 0.5289825797080994,
      "learning_rate": 1.5869502682031237e-05,
      "loss": 0.092,
      "step": 12205
    },
    {
      "epoch": 0.6198284176861769,
      "grad_norm": 0.48612451553344727,
      "learning_rate": 1.586781054875882e-05,
      "loss": 0.0885,
      "step": 12210
    },
    {
      "epoch": 0.6200822376770394,
      "grad_norm": 0.4033755362033844,
      "learning_rate": 1.5866118415486404e-05,
      "loss": 0.0861,
      "step": 12215
    },
    {
      "epoch": 0.6203360576679019,
      "grad_norm": 0.5344930291175842,
      "learning_rate": 1.5864426282213988e-05,
      "loss": 0.1081,
      "step": 12220
    },
    {
      "epoch": 0.6205898776587644,
      "grad_norm": 0.3625963032245636,
      "learning_rate": 1.586273414894157e-05,
      "loss": 0.0916,
      "step": 12225
    },
    {
      "epoch": 0.6208436976496269,
      "grad_norm": 0.44668540358543396,
      "learning_rate": 1.5861042015669155e-05,
      "loss": 0.0906,
      "step": 12230
    },
    {
      "epoch": 0.6210975176404894,
      "grad_norm": 0.47905129194259644,
      "learning_rate": 1.585934988239674e-05,
      "loss": 0.0832,
      "step": 12235
    },
    {
      "epoch": 0.6213513376313519,
      "grad_norm": 0.5475975275039673,
      "learning_rate": 1.5857657749124322e-05,
      "loss": 0.0921,
      "step": 12240
    },
    {
      "epoch": 0.6216051576222144,
      "grad_norm": 0.4593254625797272,
      "learning_rate": 1.5855965615851906e-05,
      "loss": 0.0856,
      "step": 12245
    },
    {
      "epoch": 0.6218589776130768,
      "grad_norm": 0.7837191224098206,
      "learning_rate": 1.585427348257949e-05,
      "loss": 0.0886,
      "step": 12250
    },
    {
      "epoch": 0.6221127976039393,
      "grad_norm": 0.3772871196269989,
      "learning_rate": 1.5852581349307073e-05,
      "loss": 0.0924,
      "step": 12255
    },
    {
      "epoch": 0.6223666175948017,
      "grad_norm": 0.42365825176239014,
      "learning_rate": 1.5850889216034656e-05,
      "loss": 0.0944,
      "step": 12260
    },
    {
      "epoch": 0.6226204375856642,
      "grad_norm": 0.36323311924934387,
      "learning_rate": 1.584919708276224e-05,
      "loss": 0.0899,
      "step": 12265
    },
    {
      "epoch": 0.6228742575765267,
      "grad_norm": 0.39171355962753296,
      "learning_rate": 1.5847504949489823e-05,
      "loss": 0.0811,
      "step": 12270
    },
    {
      "epoch": 0.6231280775673892,
      "grad_norm": 0.914495587348938,
      "learning_rate": 1.5845812816217407e-05,
      "loss": 0.0969,
      "step": 12275
    },
    {
      "epoch": 0.6233818975582517,
      "grad_norm": 0.46378669142723083,
      "learning_rate": 1.584412068294499e-05,
      "loss": 0.0989,
      "step": 12280
    },
    {
      "epoch": 0.6236357175491142,
      "grad_norm": 0.33671021461486816,
      "learning_rate": 1.5842428549672574e-05,
      "loss": 0.0854,
      "step": 12285
    },
    {
      "epoch": 0.6238895375399767,
      "grad_norm": 0.6674852967262268,
      "learning_rate": 1.5840736416400158e-05,
      "loss": 0.0882,
      "step": 12290
    },
    {
      "epoch": 0.6241433575308392,
      "grad_norm": 0.3157777190208435,
      "learning_rate": 1.583904428312774e-05,
      "loss": 0.0932,
      "step": 12295
    },
    {
      "epoch": 0.6243971775217017,
      "grad_norm": 0.6090107560157776,
      "learning_rate": 1.583735214985532e-05,
      "loss": 0.0912,
      "step": 12300
    },
    {
      "epoch": 0.6246509975125641,
      "grad_norm": 0.4318656325340271,
      "learning_rate": 1.5835660016582908e-05,
      "loss": 0.0967,
      "step": 12305
    },
    {
      "epoch": 0.6249048175034265,
      "grad_norm": 0.42950794100761414,
      "learning_rate": 1.5833967883310492e-05,
      "loss": 0.0841,
      "step": 12310
    },
    {
      "epoch": 0.625158637494289,
      "grad_norm": 0.37206709384918213,
      "learning_rate": 1.5832275750038075e-05,
      "loss": 0.093,
      "step": 12315
    },
    {
      "epoch": 0.6254124574851515,
      "grad_norm": 0.5274568200111389,
      "learning_rate": 1.583058361676566e-05,
      "loss": 0.1056,
      "step": 12320
    },
    {
      "epoch": 0.625666277476014,
      "grad_norm": 0.39392906427383423,
      "learning_rate": 1.582889148349324e-05,
      "loss": 0.0845,
      "step": 12325
    },
    {
      "epoch": 0.6259200974668765,
      "grad_norm": 0.6189650297164917,
      "learning_rate": 1.5827199350220826e-05,
      "loss": 0.0961,
      "step": 12330
    },
    {
      "epoch": 0.626173917457739,
      "grad_norm": 0.5449414253234863,
      "learning_rate": 1.582550721694841e-05,
      "loss": 0.0868,
      "step": 12335
    },
    {
      "epoch": 0.6264277374486015,
      "grad_norm": 0.3943621516227722,
      "learning_rate": 1.582381508367599e-05,
      "loss": 0.0779,
      "step": 12340
    },
    {
      "epoch": 0.626681557439464,
      "grad_norm": 0.5067428946495056,
      "learning_rate": 1.5822122950403577e-05,
      "loss": 0.0941,
      "step": 12345
    },
    {
      "epoch": 0.6269353774303265,
      "grad_norm": 0.4015895426273346,
      "learning_rate": 1.5820430817131157e-05,
      "loss": 0.0895,
      "step": 12350
    },
    {
      "epoch": 0.6271891974211888,
      "grad_norm": 0.3975540101528168,
      "learning_rate": 1.5818738683858744e-05,
      "loss": 0.0934,
      "step": 12355
    },
    {
      "epoch": 0.6274430174120513,
      "grad_norm": 0.430369108915329,
      "learning_rate": 1.5817046550586327e-05,
      "loss": 0.1002,
      "step": 12360
    },
    {
      "epoch": 0.6276968374029138,
      "grad_norm": 0.6089135408401489,
      "learning_rate": 1.5815354417313907e-05,
      "loss": 0.0862,
      "step": 12365
    },
    {
      "epoch": 0.6279506573937763,
      "grad_norm": 0.43870821595191956,
      "learning_rate": 1.5813662284041494e-05,
      "loss": 0.0919,
      "step": 12370
    },
    {
      "epoch": 0.6282044773846388,
      "grad_norm": 0.49393364787101746,
      "learning_rate": 1.5811970150769074e-05,
      "loss": 0.0744,
      "step": 12375
    },
    {
      "epoch": 0.6284582973755013,
      "grad_norm": 0.3724878132343292,
      "learning_rate": 1.5810278017496658e-05,
      "loss": 0.0882,
      "step": 12380
    },
    {
      "epoch": 0.6287121173663638,
      "grad_norm": 0.43538209795951843,
      "learning_rate": 1.5808585884224245e-05,
      "loss": 0.0975,
      "step": 12385
    },
    {
      "epoch": 0.6289659373572263,
      "grad_norm": 0.4566599130630493,
      "learning_rate": 1.5806893750951825e-05,
      "loss": 0.0894,
      "step": 12390
    },
    {
      "epoch": 0.6292197573480888,
      "grad_norm": 0.5635015368461609,
      "learning_rate": 1.580520161767941e-05,
      "loss": 0.0965,
      "step": 12395
    },
    {
      "epoch": 0.6294735773389513,
      "grad_norm": 0.3993977904319763,
      "learning_rate": 1.5803509484406992e-05,
      "loss": 0.0837,
      "step": 12400
    },
    {
      "epoch": 0.6297273973298136,
      "grad_norm": 1.7102850675582886,
      "learning_rate": 1.5801817351134576e-05,
      "loss": 0.0964,
      "step": 12405
    },
    {
      "epoch": 0.6299812173206761,
      "grad_norm": 0.6190375089645386,
      "learning_rate": 1.5800125217862163e-05,
      "loss": 0.0833,
      "step": 12410
    },
    {
      "epoch": 0.6302350373115386,
      "grad_norm": 0.6037442684173584,
      "learning_rate": 1.5798433084589743e-05,
      "loss": 0.0892,
      "step": 12415
    },
    {
      "epoch": 0.6304888573024011,
      "grad_norm": 0.36666348576545715,
      "learning_rate": 1.5796740951317326e-05,
      "loss": 0.0823,
      "step": 12420
    },
    {
      "epoch": 0.6307426772932636,
      "grad_norm": 0.8224522471427917,
      "learning_rate": 1.579504881804491e-05,
      "loss": 0.0833,
      "step": 12425
    },
    {
      "epoch": 0.6309964972841261,
      "grad_norm": 0.37060612440109253,
      "learning_rate": 1.5793356684772493e-05,
      "loss": 0.09,
      "step": 12430
    },
    {
      "epoch": 0.6312503172749886,
      "grad_norm": 0.362363338470459,
      "learning_rate": 1.5791664551500077e-05,
      "loss": 0.0871,
      "step": 12435
    },
    {
      "epoch": 0.6315041372658511,
      "grad_norm": 0.4423457980155945,
      "learning_rate": 1.578997241822766e-05,
      "loss": 0.0924,
      "step": 12440
    },
    {
      "epoch": 0.6317579572567136,
      "grad_norm": 0.3518649935722351,
      "learning_rate": 1.5788280284955244e-05,
      "loss": 0.0833,
      "step": 12445
    },
    {
      "epoch": 0.6320117772475761,
      "grad_norm": 0.46395477652549744,
      "learning_rate": 1.5786588151682828e-05,
      "loss": 0.0922,
      "step": 12450
    },
    {
      "epoch": 0.6322655972384384,
      "grad_norm": 0.34381961822509766,
      "learning_rate": 1.578489601841041e-05,
      "loss": 0.0995,
      "step": 12455
    },
    {
      "epoch": 0.6325194172293009,
      "grad_norm": 0.4519914388656616,
      "learning_rate": 1.5783203885137995e-05,
      "loss": 0.0853,
      "step": 12460
    },
    {
      "epoch": 0.6327732372201634,
      "grad_norm": 0.30855950713157654,
      "learning_rate": 1.5781511751865578e-05,
      "loss": 0.0788,
      "step": 12465
    },
    {
      "epoch": 0.6330270572110259,
      "grad_norm": 0.4378630518913269,
      "learning_rate": 1.5779819618593162e-05,
      "loss": 0.0803,
      "step": 12470
    },
    {
      "epoch": 0.6332808772018884,
      "grad_norm": 0.7794264554977417,
      "learning_rate": 1.5778127485320745e-05,
      "loss": 0.0947,
      "step": 12475
    },
    {
      "epoch": 0.6335346971927509,
      "grad_norm": 0.44606897234916687,
      "learning_rate": 1.577643535204833e-05,
      "loss": 0.0966,
      "step": 12480
    },
    {
      "epoch": 0.6337885171836134,
      "grad_norm": 0.8295760750770569,
      "learning_rate": 1.5774743218775912e-05,
      "loss": 0.0973,
      "step": 12485
    },
    {
      "epoch": 0.6340423371744759,
      "grad_norm": 0.6100575923919678,
      "learning_rate": 1.5773051085503496e-05,
      "loss": 0.0763,
      "step": 12490
    },
    {
      "epoch": 0.6342961571653384,
      "grad_norm": 0.43771716952323914,
      "learning_rate": 1.577135895223108e-05,
      "loss": 0.092,
      "step": 12495
    },
    {
      "epoch": 0.6345499771562009,
      "grad_norm": 0.43595728278160095,
      "learning_rate": 1.5769666818958663e-05,
      "loss": 0.083,
      "step": 12500
    },
    {
      "epoch": 0.6348037971470633,
      "grad_norm": 0.5066717267036438,
      "learning_rate": 1.5767974685686247e-05,
      "loss": 0.0909,
      "step": 12505
    },
    {
      "epoch": 0.6350576171379257,
      "grad_norm": 0.42285871505737305,
      "learning_rate": 1.576628255241383e-05,
      "loss": 0.0894,
      "step": 12510
    },
    {
      "epoch": 0.6353114371287882,
      "grad_norm": 0.35143038630485535,
      "learning_rate": 1.5764590419141414e-05,
      "loss": 0.0883,
      "step": 12515
    },
    {
      "epoch": 0.6355652571196507,
      "grad_norm": 0.36473914980888367,
      "learning_rate": 1.5762898285868997e-05,
      "loss": 0.091,
      "step": 12520
    },
    {
      "epoch": 0.6358190771105132,
      "grad_norm": 0.4791051745414734,
      "learning_rate": 1.576120615259658e-05,
      "loss": 0.1023,
      "step": 12525
    },
    {
      "epoch": 0.6360728971013757,
      "grad_norm": 0.3315756916999817,
      "learning_rate": 1.575951401932416e-05,
      "loss": 0.0806,
      "step": 12530
    },
    {
      "epoch": 0.6363267170922382,
      "grad_norm": 0.3584679365158081,
      "learning_rate": 1.5757821886051748e-05,
      "loss": 0.0875,
      "step": 12535
    },
    {
      "epoch": 0.6365805370831007,
      "grad_norm": 0.40165871381759644,
      "learning_rate": 1.575612975277933e-05,
      "loss": 0.0916,
      "step": 12540
    },
    {
      "epoch": 0.6368343570739632,
      "grad_norm": 0.8896801471710205,
      "learning_rate": 1.575443761950691e-05,
      "loss": 0.0886,
      "step": 12545
    },
    {
      "epoch": 0.6370881770648257,
      "grad_norm": 0.3733901083469391,
      "learning_rate": 1.57527454862345e-05,
      "loss": 0.0791,
      "step": 12550
    },
    {
      "epoch": 0.6373419970556881,
      "grad_norm": 0.43153393268585205,
      "learning_rate": 1.575105335296208e-05,
      "loss": 0.0894,
      "step": 12555
    },
    {
      "epoch": 0.6375958170465506,
      "grad_norm": 0.6183637976646423,
      "learning_rate": 1.5749361219689666e-05,
      "loss": 0.0858,
      "step": 12560
    },
    {
      "epoch": 0.637849637037413,
      "grad_norm": 0.4100203216075897,
      "learning_rate": 1.574766908641725e-05,
      "loss": 0.1033,
      "step": 12565
    },
    {
      "epoch": 0.6381034570282755,
      "grad_norm": 0.5121060013771057,
      "learning_rate": 1.574597695314483e-05,
      "loss": 0.0868,
      "step": 12570
    },
    {
      "epoch": 0.638357277019138,
      "grad_norm": 1.0560269355773926,
      "learning_rate": 1.5744284819872416e-05,
      "loss": 0.09,
      "step": 12575
    },
    {
      "epoch": 0.6386110970100005,
      "grad_norm": 0.365310937166214,
      "learning_rate": 1.5742592686599996e-05,
      "loss": 0.0885,
      "step": 12580
    },
    {
      "epoch": 0.638864917000863,
      "grad_norm": 0.41774865984916687,
      "learning_rate": 1.574090055332758e-05,
      "loss": 0.0791,
      "step": 12585
    },
    {
      "epoch": 0.6391187369917255,
      "grad_norm": 0.4893551468849182,
      "learning_rate": 1.5739208420055167e-05,
      "loss": 0.0845,
      "step": 12590
    },
    {
      "epoch": 0.639372556982588,
      "grad_norm": 0.5731279850006104,
      "learning_rate": 1.5737516286782747e-05,
      "loss": 0.1017,
      "step": 12595
    },
    {
      "epoch": 0.6396263769734505,
      "grad_norm": 0.6109359860420227,
      "learning_rate": 1.573582415351033e-05,
      "loss": 0.0796,
      "step": 12600
    },
    {
      "epoch": 0.6398801969643129,
      "grad_norm": 0.3426520526409149,
      "learning_rate": 1.5734132020237914e-05,
      "loss": 0.0999,
      "step": 12605
    },
    {
      "epoch": 0.6401340169551754,
      "grad_norm": 0.3924323618412018,
      "learning_rate": 1.5732439886965498e-05,
      "loss": 0.0845,
      "step": 12610
    },
    {
      "epoch": 0.6403878369460378,
      "grad_norm": 1.8754016160964966,
      "learning_rate": 1.5730747753693085e-05,
      "loss": 0.1001,
      "step": 12615
    },
    {
      "epoch": 0.6406416569369003,
      "grad_norm": 0.3848492205142975,
      "learning_rate": 1.5729055620420665e-05,
      "loss": 0.0788,
      "step": 12620
    },
    {
      "epoch": 0.6408954769277628,
      "grad_norm": 0.671140730381012,
      "learning_rate": 1.572736348714825e-05,
      "loss": 0.0758,
      "step": 12625
    },
    {
      "epoch": 0.6411492969186253,
      "grad_norm": 0.38087940216064453,
      "learning_rate": 1.5725671353875832e-05,
      "loss": 0.0932,
      "step": 12630
    },
    {
      "epoch": 0.6414031169094878,
      "grad_norm": 0.5257353782653809,
      "learning_rate": 1.5723979220603415e-05,
      "loss": 0.0882,
      "step": 12635
    },
    {
      "epoch": 0.6416569369003503,
      "grad_norm": 0.35013777017593384,
      "learning_rate": 1.5722287087331e-05,
      "loss": 0.0856,
      "step": 12640
    },
    {
      "epoch": 0.6419107568912128,
      "grad_norm": 0.4160863161087036,
      "learning_rate": 1.5720594954058582e-05,
      "loss": 0.0833,
      "step": 12645
    },
    {
      "epoch": 0.6421645768820752,
      "grad_norm": 0.3413524031639099,
      "learning_rate": 1.5718902820786166e-05,
      "loss": 0.0802,
      "step": 12650
    },
    {
      "epoch": 0.6424183968729377,
      "grad_norm": 0.5095176100730896,
      "learning_rate": 1.571721068751375e-05,
      "loss": 0.0931,
      "step": 12655
    },
    {
      "epoch": 0.6426722168638002,
      "grad_norm": 0.45131492614746094,
      "learning_rate": 1.5715518554241333e-05,
      "loss": 0.0814,
      "step": 12660
    },
    {
      "epoch": 0.6429260368546627,
      "grad_norm": 0.4562227427959442,
      "learning_rate": 1.5713826420968917e-05,
      "loss": 0.0849,
      "step": 12665
    },
    {
      "epoch": 0.6431798568455251,
      "grad_norm": 0.38386526703834534,
      "learning_rate": 1.57121342876965e-05,
      "loss": 0.0848,
      "step": 12670
    },
    {
      "epoch": 0.6434336768363876,
      "grad_norm": 0.5002121329307556,
      "learning_rate": 1.5710442154424084e-05,
      "loss": 0.1066,
      "step": 12675
    },
    {
      "epoch": 0.6436874968272501,
      "grad_norm": 0.42015817761421204,
      "learning_rate": 1.5708750021151667e-05,
      "loss": 0.0995,
      "step": 12680
    },
    {
      "epoch": 0.6439413168181126,
      "grad_norm": 0.3642427921295166,
      "learning_rate": 1.570705788787925e-05,
      "loss": 0.0811,
      "step": 12685
    },
    {
      "epoch": 0.6441951368089751,
      "grad_norm": 0.40864598751068115,
      "learning_rate": 1.5705365754606834e-05,
      "loss": 0.0896,
      "step": 12690
    },
    {
      "epoch": 0.6444489567998376,
      "grad_norm": 0.3729974925518036,
      "learning_rate": 1.5703673621334418e-05,
      "loss": 0.0944,
      "step": 12695
    },
    {
      "epoch": 0.6447027767907,
      "grad_norm": 0.8188928365707397,
      "learning_rate": 1.5701981488062e-05,
      "loss": 0.0831,
      "step": 12700
    },
    {
      "epoch": 0.6449565967815625,
      "grad_norm": 0.49652352929115295,
      "learning_rate": 1.5700289354789585e-05,
      "loss": 0.0879,
      "step": 12705
    },
    {
      "epoch": 0.645210416772425,
      "grad_norm": 0.5149166584014893,
      "learning_rate": 1.569859722151717e-05,
      "loss": 0.0998,
      "step": 12710
    },
    {
      "epoch": 0.6454642367632875,
      "grad_norm": 0.3833802044391632,
      "learning_rate": 1.5696905088244752e-05,
      "loss": 0.0943,
      "step": 12715
    },
    {
      "epoch": 0.64571805675415,
      "grad_norm": 0.5231676697731018,
      "learning_rate": 1.5695212954972336e-05,
      "loss": 0.0819,
      "step": 12720
    },
    {
      "epoch": 0.6459718767450124,
      "grad_norm": 0.977109432220459,
      "learning_rate": 1.569352082169992e-05,
      "loss": 0.0956,
      "step": 12725
    },
    {
      "epoch": 0.6462256967358749,
      "grad_norm": 0.4414095878601074,
      "learning_rate": 1.5691828688427503e-05,
      "loss": 0.0873,
      "step": 12730
    },
    {
      "epoch": 0.6464795167267374,
      "grad_norm": 0.46533313393592834,
      "learning_rate": 1.5690136555155083e-05,
      "loss": 0.0878,
      "step": 12735
    },
    {
      "epoch": 0.6467333367175999,
      "grad_norm": 0.4510734975337982,
      "learning_rate": 1.568844442188267e-05,
      "loss": 0.0954,
      "step": 12740
    },
    {
      "epoch": 0.6469871567084624,
      "grad_norm": 0.43129807710647583,
      "learning_rate": 1.5686752288610253e-05,
      "loss": 0.0892,
      "step": 12745
    },
    {
      "epoch": 0.6472409766993248,
      "grad_norm": 0.5520406365394592,
      "learning_rate": 1.5685060155337837e-05,
      "loss": 0.0969,
      "step": 12750
    },
    {
      "epoch": 0.6474947966901873,
      "grad_norm": 0.40193498134613037,
      "learning_rate": 1.568336802206542e-05,
      "loss": 0.0808,
      "step": 12755
    },
    {
      "epoch": 0.6477486166810498,
      "grad_norm": 0.3910532295703888,
      "learning_rate": 1.5681675888793e-05,
      "loss": 0.0952,
      "step": 12760
    },
    {
      "epoch": 0.6480024366719123,
      "grad_norm": 0.3040367066860199,
      "learning_rate": 1.5679983755520588e-05,
      "loss": 0.0744,
      "step": 12765
    },
    {
      "epoch": 0.6482562566627748,
      "grad_norm": 0.34013864398002625,
      "learning_rate": 1.567829162224817e-05,
      "loss": 0.0849,
      "step": 12770
    },
    {
      "epoch": 0.6485100766536372,
      "grad_norm": 0.426899790763855,
      "learning_rate": 1.567659948897575e-05,
      "loss": 0.0898,
      "step": 12775
    },
    {
      "epoch": 0.6487638966444997,
      "grad_norm": 0.32523438334465027,
      "learning_rate": 1.5674907355703338e-05,
      "loss": 0.0828,
      "step": 12780
    },
    {
      "epoch": 0.6490177166353622,
      "grad_norm": 0.35754647850990295,
      "learning_rate": 1.567321522243092e-05,
      "loss": 0.0731,
      "step": 12785
    },
    {
      "epoch": 0.6492715366262247,
      "grad_norm": 0.45699095726013184,
      "learning_rate": 1.5671523089158502e-05,
      "loss": 0.0838,
      "step": 12790
    },
    {
      "epoch": 0.6495253566170872,
      "grad_norm": 0.41177016496658325,
      "learning_rate": 1.566983095588609e-05,
      "loss": 0.0801,
      "step": 12795
    },
    {
      "epoch": 0.6497791766079496,
      "grad_norm": 0.3860916793346405,
      "learning_rate": 1.566813882261367e-05,
      "loss": 0.0871,
      "step": 12800
    },
    {
      "epoch": 0.6500329965988121,
      "grad_norm": 0.6945805549621582,
      "learning_rate": 1.5666446689341256e-05,
      "loss": 0.0896,
      "step": 12805
    },
    {
      "epoch": 0.6502868165896746,
      "grad_norm": 0.3880050778388977,
      "learning_rate": 1.5664754556068836e-05,
      "loss": 0.0779,
      "step": 12810
    },
    {
      "epoch": 0.6505406365805371,
      "grad_norm": 0.689802348613739,
      "learning_rate": 1.566306242279642e-05,
      "loss": 0.0824,
      "step": 12815
    },
    {
      "epoch": 0.6507944565713996,
      "grad_norm": 0.5213664174079895,
      "learning_rate": 1.5661370289524007e-05,
      "loss": 0.0873,
      "step": 12820
    },
    {
      "epoch": 0.651048276562262,
      "grad_norm": 0.4796338975429535,
      "learning_rate": 1.5659678156251587e-05,
      "loss": 0.0889,
      "step": 12825
    },
    {
      "epoch": 0.6513020965531245,
      "grad_norm": 0.4608859419822693,
      "learning_rate": 1.565798602297917e-05,
      "loss": 0.0846,
      "step": 12830
    },
    {
      "epoch": 0.651555916543987,
      "grad_norm": 0.3637705147266388,
      "learning_rate": 1.5656293889706754e-05,
      "loss": 0.0808,
      "step": 12835
    },
    {
      "epoch": 0.6518097365348495,
      "grad_norm": 0.45812270045280457,
      "learning_rate": 1.5654601756434337e-05,
      "loss": 0.0876,
      "step": 12840
    },
    {
      "epoch": 0.652063556525712,
      "grad_norm": 0.35677555203437805,
      "learning_rate": 1.565290962316192e-05,
      "loss": 0.0921,
      "step": 12845
    },
    {
      "epoch": 0.6523173765165744,
      "grad_norm": 0.6063011884689331,
      "learning_rate": 1.5651217489889504e-05,
      "loss": 0.0882,
      "step": 12850
    },
    {
      "epoch": 0.6525711965074369,
      "grad_norm": 0.3067625164985657,
      "learning_rate": 1.5649525356617088e-05,
      "loss": 0.0763,
      "step": 12855
    },
    {
      "epoch": 0.6528250164982994,
      "grad_norm": 0.3955833315849304,
      "learning_rate": 1.564783322334467e-05,
      "loss": 0.0813,
      "step": 12860
    },
    {
      "epoch": 0.6530788364891619,
      "grad_norm": 0.4854811728000641,
      "learning_rate": 1.5646141090072255e-05,
      "loss": 0.0762,
      "step": 12865
    },
    {
      "epoch": 0.6533326564800244,
      "grad_norm": 0.3633703589439392,
      "learning_rate": 1.564444895679984e-05,
      "loss": 0.0891,
      "step": 12870
    },
    {
      "epoch": 0.6535864764708869,
      "grad_norm": 0.41939306259155273,
      "learning_rate": 1.5642756823527422e-05,
      "loss": 0.0981,
      "step": 12875
    },
    {
      "epoch": 0.6538402964617493,
      "grad_norm": 0.8128605484962463,
      "learning_rate": 1.5641064690255006e-05,
      "loss": 0.0942,
      "step": 12880
    },
    {
      "epoch": 0.6540941164526118,
      "grad_norm": 1.6264680624008179,
      "learning_rate": 1.563937255698259e-05,
      "loss": 0.0831,
      "step": 12885
    },
    {
      "epoch": 0.6543479364434743,
      "grad_norm": 0.35295000672340393,
      "learning_rate": 1.5637680423710173e-05,
      "loss": 0.0805,
      "step": 12890
    },
    {
      "epoch": 0.6546017564343368,
      "grad_norm": 0.327200323343277,
      "learning_rate": 1.5635988290437756e-05,
      "loss": 0.085,
      "step": 12895
    },
    {
      "epoch": 0.6548555764251992,
      "grad_norm": 0.37761542201042175,
      "learning_rate": 1.563429615716534e-05,
      "loss": 0.0808,
      "step": 12900
    },
    {
      "epoch": 0.6551093964160617,
      "grad_norm": 0.39181774854660034,
      "learning_rate": 1.5632604023892923e-05,
      "loss": 0.0852,
      "step": 12905
    },
    {
      "epoch": 0.6553632164069242,
      "grad_norm": 0.4610046148300171,
      "learning_rate": 1.5630911890620507e-05,
      "loss": 0.0861,
      "step": 12910
    },
    {
      "epoch": 0.6556170363977867,
      "grad_norm": 0.42793646454811096,
      "learning_rate": 1.562921975734809e-05,
      "loss": 0.0876,
      "step": 12915
    },
    {
      "epoch": 0.6558708563886492,
      "grad_norm": 0.6725565195083618,
      "learning_rate": 1.5627527624075674e-05,
      "loss": 0.1022,
      "step": 12920
    },
    {
      "epoch": 0.6561246763795117,
      "grad_norm": 0.6574685573577881,
      "learning_rate": 1.5625835490803258e-05,
      "loss": 0.1023,
      "step": 12925
    },
    {
      "epoch": 0.6563784963703742,
      "grad_norm": 0.45522236824035645,
      "learning_rate": 1.562414335753084e-05,
      "loss": 0.0794,
      "step": 12930
    },
    {
      "epoch": 0.6566323163612366,
      "grad_norm": 0.4538464844226837,
      "learning_rate": 1.5622451224258425e-05,
      "loss": 0.087,
      "step": 12935
    },
    {
      "epoch": 0.6568861363520991,
      "grad_norm": 0.6361369490623474,
      "learning_rate": 1.5620759090986005e-05,
      "loss": 0.0863,
      "step": 12940
    },
    {
      "epoch": 0.6571399563429615,
      "grad_norm": 0.47415581345558167,
      "learning_rate": 1.5619066957713592e-05,
      "loss": 0.0845,
      "step": 12945
    },
    {
      "epoch": 0.657393776333824,
      "grad_norm": 0.35891467332839966,
      "learning_rate": 1.5617374824441175e-05,
      "loss": 0.0912,
      "step": 12950
    },
    {
      "epoch": 0.6576475963246865,
      "grad_norm": 0.5842028260231018,
      "learning_rate": 1.561568269116876e-05,
      "loss": 0.0764,
      "step": 12955
    },
    {
      "epoch": 0.657901416315549,
      "grad_norm": 0.7354117035865784,
      "learning_rate": 1.5613990557896342e-05,
      "loss": 0.0835,
      "step": 12960
    },
    {
      "epoch": 0.6581552363064115,
      "grad_norm": 0.43807005882263184,
      "learning_rate": 1.5612298424623923e-05,
      "loss": 0.1034,
      "step": 12965
    },
    {
      "epoch": 0.658409056297274,
      "grad_norm": 0.5703946948051453,
      "learning_rate": 1.561060629135151e-05,
      "loss": 0.0799,
      "step": 12970
    },
    {
      "epoch": 0.6586628762881365,
      "grad_norm": 0.29095369577407837,
      "learning_rate": 1.5608914158079093e-05,
      "loss": 0.0762,
      "step": 12975
    },
    {
      "epoch": 0.658916696278999,
      "grad_norm": 0.4557504653930664,
      "learning_rate": 1.5607222024806673e-05,
      "loss": 0.0991,
      "step": 12980
    },
    {
      "epoch": 0.6591705162698615,
      "grad_norm": 0.5406267642974854,
      "learning_rate": 1.560552989153426e-05,
      "loss": 0.0837,
      "step": 12985
    },
    {
      "epoch": 0.6594243362607239,
      "grad_norm": 0.5688093304634094,
      "learning_rate": 1.560383775826184e-05,
      "loss": 0.0819,
      "step": 12990
    },
    {
      "epoch": 0.6596781562515863,
      "grad_norm": 0.4395672082901001,
      "learning_rate": 1.5602145624989427e-05,
      "loss": 0.0845,
      "step": 12995
    },
    {
      "epoch": 0.6599319762424488,
      "grad_norm": 0.3437829315662384,
      "learning_rate": 1.560045349171701e-05,
      "loss": 0.0785,
      "step": 13000
    },
    {
      "epoch": 0.6601857962333113,
      "grad_norm": 0.42482784390449524,
      "learning_rate": 1.559876135844459e-05,
      "loss": 0.0914,
      "step": 13005
    },
    {
      "epoch": 0.6604396162241738,
      "grad_norm": 0.4697912931442261,
      "learning_rate": 1.5597069225172178e-05,
      "loss": 0.0852,
      "step": 13010
    },
    {
      "epoch": 0.6606934362150363,
      "grad_norm": 0.36666378378868103,
      "learning_rate": 1.5595377091899758e-05,
      "loss": 0.0798,
      "step": 13015
    },
    {
      "epoch": 0.6609472562058988,
      "grad_norm": 0.37574443221092224,
      "learning_rate": 1.559368495862734e-05,
      "loss": 0.0932,
      "step": 13020
    },
    {
      "epoch": 0.6612010761967613,
      "grad_norm": 0.5501673221588135,
      "learning_rate": 1.559199282535493e-05,
      "loss": 0.0967,
      "step": 13025
    },
    {
      "epoch": 0.6614548961876238,
      "grad_norm": 0.48891526460647583,
      "learning_rate": 1.559030069208251e-05,
      "loss": 0.0814,
      "step": 13030
    },
    {
      "epoch": 0.6617087161784863,
      "grad_norm": 0.40868237614631653,
      "learning_rate": 1.5588608558810092e-05,
      "loss": 0.0857,
      "step": 13035
    },
    {
      "epoch": 0.6619625361693487,
      "grad_norm": 0.5038067102432251,
      "learning_rate": 1.5586916425537676e-05,
      "loss": 0.0791,
      "step": 13040
    },
    {
      "epoch": 0.6622163561602111,
      "grad_norm": 0.3892373740673065,
      "learning_rate": 1.558522429226526e-05,
      "loss": 0.0853,
      "step": 13045
    },
    {
      "epoch": 0.6624701761510736,
      "grad_norm": 0.6677381992340088,
      "learning_rate": 1.5583532158992846e-05,
      "loss": 0.083,
      "step": 13050
    },
    {
      "epoch": 0.6627239961419361,
      "grad_norm": 0.34850823879241943,
      "learning_rate": 1.5581840025720426e-05,
      "loss": 0.0779,
      "step": 13055
    },
    {
      "epoch": 0.6629778161327986,
      "grad_norm": 0.5405170321464539,
      "learning_rate": 1.558014789244801e-05,
      "loss": 0.076,
      "step": 13060
    },
    {
      "epoch": 0.6632316361236611,
      "grad_norm": 0.4197944402694702,
      "learning_rate": 1.5578455759175593e-05,
      "loss": 0.0795,
      "step": 13065
    },
    {
      "epoch": 0.6634854561145236,
      "grad_norm": 0.48564544320106506,
      "learning_rate": 1.5576763625903177e-05,
      "loss": 0.0815,
      "step": 13070
    },
    {
      "epoch": 0.6637392761053861,
      "grad_norm": 0.37770646810531616,
      "learning_rate": 1.557507149263076e-05,
      "loss": 0.0891,
      "step": 13075
    },
    {
      "epoch": 0.6639930960962486,
      "grad_norm": 0.39739593863487244,
      "learning_rate": 1.5573379359358344e-05,
      "loss": 0.0833,
      "step": 13080
    },
    {
      "epoch": 0.6642469160871111,
      "grad_norm": 0.3831949234008789,
      "learning_rate": 1.5571687226085928e-05,
      "loss": 0.0854,
      "step": 13085
    },
    {
      "epoch": 0.6645007360779736,
      "grad_norm": 0.3688269257545471,
      "learning_rate": 1.556999509281351e-05,
      "loss": 0.0814,
      "step": 13090
    },
    {
      "epoch": 0.6647545560688359,
      "grad_norm": 0.3308967649936676,
      "learning_rate": 1.5568302959541095e-05,
      "loss": 0.0668,
      "step": 13095
    },
    {
      "epoch": 0.6650083760596984,
      "grad_norm": 0.665611743927002,
      "learning_rate": 1.5566610826268678e-05,
      "loss": 0.0919,
      "step": 13100
    },
    {
      "epoch": 0.6652621960505609,
      "grad_norm": 0.3787246346473694,
      "learning_rate": 1.5564918692996262e-05,
      "loss": 0.0869,
      "step": 13105
    },
    {
      "epoch": 0.6655160160414234,
      "grad_norm": 0.5023996233940125,
      "learning_rate": 1.5563226559723845e-05,
      "loss": 0.0873,
      "step": 13110
    },
    {
      "epoch": 0.6657698360322859,
      "grad_norm": 0.36124274134635925,
      "learning_rate": 1.556153442645143e-05,
      "loss": 0.0768,
      "step": 13115
    },
    {
      "epoch": 0.6660236560231484,
      "grad_norm": 0.7764711380004883,
      "learning_rate": 1.5559842293179012e-05,
      "loss": 0.083,
      "step": 13120
    },
    {
      "epoch": 0.6662774760140109,
      "grad_norm": 0.36708641052246094,
      "learning_rate": 1.5558150159906596e-05,
      "loss": 0.0899,
      "step": 13125
    },
    {
      "epoch": 0.6665312960048734,
      "grad_norm": 0.45462819933891296,
      "learning_rate": 1.555645802663418e-05,
      "loss": 0.0968,
      "step": 13130
    },
    {
      "epoch": 0.6667851159957359,
      "grad_norm": 0.4756264388561249,
      "learning_rate": 1.5554765893361763e-05,
      "loss": 0.0873,
      "step": 13135
    },
    {
      "epoch": 0.6670389359865984,
      "grad_norm": 0.3898143470287323,
      "learning_rate": 1.5553073760089347e-05,
      "loss": 0.0862,
      "step": 13140
    },
    {
      "epoch": 0.6672927559774607,
      "grad_norm": 0.32764703035354614,
      "learning_rate": 1.555138162681693e-05,
      "loss": 0.0877,
      "step": 13145
    },
    {
      "epoch": 0.6675465759683232,
      "grad_norm": 0.5033909678459167,
      "learning_rate": 1.5549689493544514e-05,
      "loss": 0.0812,
      "step": 13150
    },
    {
      "epoch": 0.6678003959591857,
      "grad_norm": 0.42792779207229614,
      "learning_rate": 1.5547997360272097e-05,
      "loss": 0.0917,
      "step": 13155
    },
    {
      "epoch": 0.6680542159500482,
      "grad_norm": 0.37381792068481445,
      "learning_rate": 1.554630522699968e-05,
      "loss": 0.0796,
      "step": 13160
    },
    {
      "epoch": 0.6683080359409107,
      "grad_norm": 0.4677964448928833,
      "learning_rate": 1.5544613093727264e-05,
      "loss": 0.0892,
      "step": 13165
    },
    {
      "epoch": 0.6685618559317732,
      "grad_norm": 0.36071425676345825,
      "learning_rate": 1.5542920960454845e-05,
      "loss": 0.0961,
      "step": 13170
    },
    {
      "epoch": 0.6688156759226357,
      "grad_norm": 0.41257521510124207,
      "learning_rate": 1.554122882718243e-05,
      "loss": 0.0826,
      "step": 13175
    },
    {
      "epoch": 0.6690694959134982,
      "grad_norm": 0.48096606135368347,
      "learning_rate": 1.5539536693910015e-05,
      "loss": 0.0819,
      "step": 13180
    },
    {
      "epoch": 0.6693233159043607,
      "grad_norm": 0.3370281755924225,
      "learning_rate": 1.5537844560637595e-05,
      "loss": 0.0878,
      "step": 13185
    },
    {
      "epoch": 0.669577135895223,
      "grad_norm": 0.4116274416446686,
      "learning_rate": 1.5536152427365182e-05,
      "loss": 0.0878,
      "step": 13190
    },
    {
      "epoch": 0.6698309558860855,
      "grad_norm": 0.5579774975776672,
      "learning_rate": 1.5534460294092762e-05,
      "loss": 0.0886,
      "step": 13195
    },
    {
      "epoch": 0.670084775876948,
      "grad_norm": 0.5344901084899902,
      "learning_rate": 1.553276816082035e-05,
      "loss": 0.0756,
      "step": 13200
    },
    {
      "epoch": 0.6703385958678105,
      "grad_norm": 0.5055976510047913,
      "learning_rate": 1.5531076027547933e-05,
      "loss": 0.097,
      "step": 13205
    },
    {
      "epoch": 0.670592415858673,
      "grad_norm": 0.3117898106575012,
      "learning_rate": 1.5529383894275513e-05,
      "loss": 0.0839,
      "step": 13210
    },
    {
      "epoch": 0.6708462358495355,
      "grad_norm": 0.4203091561794281,
      "learning_rate": 1.55276917610031e-05,
      "loss": 0.0738,
      "step": 13215
    },
    {
      "epoch": 0.671100055840398,
      "grad_norm": 0.975365936756134,
      "learning_rate": 1.552599962773068e-05,
      "loss": 0.0877,
      "step": 13220
    },
    {
      "epoch": 0.6713538758312605,
      "grad_norm": 0.400333970785141,
      "learning_rate": 1.5524307494458263e-05,
      "loss": 0.0679,
      "step": 13225
    },
    {
      "epoch": 0.671607695822123,
      "grad_norm": 0.42872706055641174,
      "learning_rate": 1.552261536118585e-05,
      "loss": 0.0807,
      "step": 13230
    },
    {
      "epoch": 0.6718615158129855,
      "grad_norm": 0.6446133852005005,
      "learning_rate": 1.552092322791343e-05,
      "loss": 0.0725,
      "step": 13235
    },
    {
      "epoch": 0.6721153358038479,
      "grad_norm": 0.5523983836174011,
      "learning_rate": 1.5519231094641018e-05,
      "loss": 0.0851,
      "step": 13240
    },
    {
      "epoch": 0.6723691557947103,
      "grad_norm": 0.329487681388855,
      "learning_rate": 1.5517538961368598e-05,
      "loss": 0.0744,
      "step": 13245
    },
    {
      "epoch": 0.6726229757855728,
      "grad_norm": 0.38870227336883545,
      "learning_rate": 1.551584682809618e-05,
      "loss": 0.0739,
      "step": 13250
    },
    {
      "epoch": 0.6728767957764353,
      "grad_norm": 0.47514641284942627,
      "learning_rate": 1.5514154694823768e-05,
      "loss": 0.0765,
      "step": 13255
    },
    {
      "epoch": 0.6731306157672978,
      "grad_norm": 0.40743643045425415,
      "learning_rate": 1.551246256155135e-05,
      "loss": 0.0772,
      "step": 13260
    },
    {
      "epoch": 0.6733844357581603,
      "grad_norm": 0.4392092525959015,
      "learning_rate": 1.5510770428278932e-05,
      "loss": 0.094,
      "step": 13265
    },
    {
      "epoch": 0.6736382557490228,
      "grad_norm": 0.46465203166007996,
      "learning_rate": 1.5509078295006515e-05,
      "loss": 0.0869,
      "step": 13270
    },
    {
      "epoch": 0.6738920757398853,
      "grad_norm": 0.5030555725097656,
      "learning_rate": 1.55073861617341e-05,
      "loss": 0.0846,
      "step": 13275
    },
    {
      "epoch": 0.6741458957307478,
      "grad_norm": 0.36062154173851013,
      "learning_rate": 1.5505694028461682e-05,
      "loss": 0.0826,
      "step": 13280
    },
    {
      "epoch": 0.6743997157216103,
      "grad_norm": 0.29261308908462524,
      "learning_rate": 1.5504001895189266e-05,
      "loss": 0.0802,
      "step": 13285
    },
    {
      "epoch": 0.6746535357124727,
      "grad_norm": 0.40803733468055725,
      "learning_rate": 1.550230976191685e-05,
      "loss": 0.0844,
      "step": 13290
    },
    {
      "epoch": 0.6749073557033352,
      "grad_norm": 0.4190680980682373,
      "learning_rate": 1.5500617628644433e-05,
      "loss": 0.0813,
      "step": 13295
    },
    {
      "epoch": 0.6751611756941976,
      "grad_norm": 0.4443584084510803,
      "learning_rate": 1.5498925495372017e-05,
      "loss": 0.0856,
      "step": 13300
    },
    {
      "epoch": 0.6754149956850601,
      "grad_norm": 0.3262367248535156,
      "learning_rate": 1.54972333620996e-05,
      "loss": 0.087,
      "step": 13305
    },
    {
      "epoch": 0.6756688156759226,
      "grad_norm": 1.0407583713531494,
      "learning_rate": 1.5495541228827184e-05,
      "loss": 0.0806,
      "step": 13310
    },
    {
      "epoch": 0.6759226356667851,
      "grad_norm": 0.5855976939201355,
      "learning_rate": 1.5493849095554767e-05,
      "loss": 0.0895,
      "step": 13315
    },
    {
      "epoch": 0.6761764556576476,
      "grad_norm": 0.6860768795013428,
      "learning_rate": 1.549215696228235e-05,
      "loss": 0.0732,
      "step": 13320
    },
    {
      "epoch": 0.6764302756485101,
      "grad_norm": 0.5066936612129211,
      "learning_rate": 1.5490464829009934e-05,
      "loss": 0.086,
      "step": 13325
    },
    {
      "epoch": 0.6766840956393726,
      "grad_norm": 0.49966979026794434,
      "learning_rate": 1.5488772695737518e-05,
      "loss": 0.0848,
      "step": 13330
    },
    {
      "epoch": 0.6769379156302351,
      "grad_norm": 0.41109582781791687,
      "learning_rate": 1.54870805624651e-05,
      "loss": 0.0834,
      "step": 13335
    },
    {
      "epoch": 0.6771917356210975,
      "grad_norm": 0.6102433204650879,
      "learning_rate": 1.5485388429192685e-05,
      "loss": 0.1299,
      "step": 13340
    },
    {
      "epoch": 0.67744555561196,
      "grad_norm": 0.39649656414985657,
      "learning_rate": 1.548369629592027e-05,
      "loss": 0.0946,
      "step": 13345
    },
    {
      "epoch": 0.6776993756028225,
      "grad_norm": 0.375001460313797,
      "learning_rate": 1.5482004162647852e-05,
      "loss": 0.0756,
      "step": 13350
    },
    {
      "epoch": 0.6779531955936849,
      "grad_norm": 0.4396669268608093,
      "learning_rate": 1.5480312029375436e-05,
      "loss": 0.0791,
      "step": 13355
    },
    {
      "epoch": 0.6782070155845474,
      "grad_norm": 0.6256466507911682,
      "learning_rate": 1.547861989610302e-05,
      "loss": 0.0827,
      "step": 13360
    },
    {
      "epoch": 0.6784608355754099,
      "grad_norm": 0.39854422211647034,
      "learning_rate": 1.5476927762830603e-05,
      "loss": 0.0791,
      "step": 13365
    },
    {
      "epoch": 0.6787146555662724,
      "grad_norm": 0.6447271108627319,
      "learning_rate": 1.5475235629558186e-05,
      "loss": 0.0898,
      "step": 13370
    },
    {
      "epoch": 0.6789684755571349,
      "grad_norm": 0.3512633740901947,
      "learning_rate": 1.5473543496285766e-05,
      "loss": 0.0914,
      "step": 13375
    },
    {
      "epoch": 0.6792222955479974,
      "grad_norm": 0.37331703305244446,
      "learning_rate": 1.5471851363013353e-05,
      "loss": 0.0822,
      "step": 13380
    },
    {
      "epoch": 0.6794761155388599,
      "grad_norm": 0.5161044001579285,
      "learning_rate": 1.5470159229740937e-05,
      "loss": 0.0886,
      "step": 13385
    },
    {
      "epoch": 0.6797299355297223,
      "grad_norm": 0.3599933981895447,
      "learning_rate": 1.546846709646852e-05,
      "loss": 0.081,
      "step": 13390
    },
    {
      "epoch": 0.6799837555205848,
      "grad_norm": 0.4737204313278198,
      "learning_rate": 1.5466774963196104e-05,
      "loss": 0.0925,
      "step": 13395
    },
    {
      "epoch": 0.6802375755114473,
      "grad_norm": 0.46349260210990906,
      "learning_rate": 1.5465082829923684e-05,
      "loss": 0.0966,
      "step": 13400
    },
    {
      "epoch": 0.6804913955023097,
      "grad_norm": 0.4162529706954956,
      "learning_rate": 1.546339069665127e-05,
      "loss": 0.0854,
      "step": 13405
    },
    {
      "epoch": 0.6807452154931722,
      "grad_norm": 0.3948574662208557,
      "learning_rate": 1.5461698563378855e-05,
      "loss": 0.0883,
      "step": 13410
    },
    {
      "epoch": 0.6809990354840347,
      "grad_norm": 0.4505586624145508,
      "learning_rate": 1.5460006430106435e-05,
      "loss": 0.0968,
      "step": 13415
    },
    {
      "epoch": 0.6812528554748972,
      "grad_norm": 0.42507103085517883,
      "learning_rate": 1.5458314296834022e-05,
      "loss": 0.0841,
      "step": 13420
    },
    {
      "epoch": 0.6815066754657597,
      "grad_norm": 0.5045070648193359,
      "learning_rate": 1.5456622163561602e-05,
      "loss": 0.0894,
      "step": 13425
    },
    {
      "epoch": 0.6817604954566222,
      "grad_norm": 0.4296087622642517,
      "learning_rate": 1.5454930030289185e-05,
      "loss": 0.0866,
      "step": 13430
    },
    {
      "epoch": 0.6820143154474847,
      "grad_norm": 0.40591058135032654,
      "learning_rate": 1.5453237897016772e-05,
      "loss": 0.0887,
      "step": 13435
    },
    {
      "epoch": 0.6822681354383471,
      "grad_norm": 0.2938343286514282,
      "learning_rate": 1.5451545763744353e-05,
      "loss": 0.0783,
      "step": 13440
    },
    {
      "epoch": 0.6825219554292096,
      "grad_norm": 0.41390931606292725,
      "learning_rate": 1.544985363047194e-05,
      "loss": 0.0865,
      "step": 13445
    },
    {
      "epoch": 0.6827757754200721,
      "grad_norm": 1.899959683418274,
      "learning_rate": 1.544816149719952e-05,
      "loss": 0.0849,
      "step": 13450
    },
    {
      "epoch": 0.6830295954109346,
      "grad_norm": 0.6087545156478882,
      "learning_rate": 1.5446469363927103e-05,
      "loss": 0.0885,
      "step": 13455
    },
    {
      "epoch": 0.683283415401797,
      "grad_norm": 0.42178311944007874,
      "learning_rate": 1.5444777230654687e-05,
      "loss": 0.0784,
      "step": 13460
    },
    {
      "epoch": 0.6835372353926595,
      "grad_norm": 0.3378629684448242,
      "learning_rate": 1.544308509738227e-05,
      "loss": 0.0891,
      "step": 13465
    },
    {
      "epoch": 0.683791055383522,
      "grad_norm": 0.45582592487335205,
      "learning_rate": 1.5441392964109854e-05,
      "loss": 0.0785,
      "step": 13470
    },
    {
      "epoch": 0.6840448753743845,
      "grad_norm": 0.4021526277065277,
      "learning_rate": 1.5439700830837437e-05,
      "loss": 0.0861,
      "step": 13475
    },
    {
      "epoch": 0.684298695365247,
      "grad_norm": 0.42688125371932983,
      "learning_rate": 1.543800869756502e-05,
      "loss": 0.0804,
      "step": 13480
    },
    {
      "epoch": 0.6845525153561094,
      "grad_norm": 0.37077417969703674,
      "learning_rate": 1.5436316564292604e-05,
      "loss": 0.0765,
      "step": 13485
    },
    {
      "epoch": 0.6848063353469719,
      "grad_norm": 0.3427707850933075,
      "learning_rate": 1.5434624431020188e-05,
      "loss": 0.0857,
      "step": 13490
    },
    {
      "epoch": 0.6850601553378344,
      "grad_norm": 0.4085235297679901,
      "learning_rate": 1.543293229774777e-05,
      "loss": 0.088,
      "step": 13495
    },
    {
      "epoch": 0.6853139753286969,
      "grad_norm": 0.38972538709640503,
      "learning_rate": 1.5431240164475355e-05,
      "loss": 0.0877,
      "step": 13500
    },
    {
      "epoch": 0.6855677953195594,
      "grad_norm": 0.35393092036247253,
      "learning_rate": 1.542954803120294e-05,
      "loss": 0.0827,
      "step": 13505
    },
    {
      "epoch": 0.6858216153104219,
      "grad_norm": 0.4118354916572571,
      "learning_rate": 1.5427855897930522e-05,
      "loss": 0.0786,
      "step": 13510
    },
    {
      "epoch": 0.6860754353012843,
      "grad_norm": 0.5753445625305176,
      "learning_rate": 1.5426163764658106e-05,
      "loss": 0.0815,
      "step": 13515
    },
    {
      "epoch": 0.6863292552921468,
      "grad_norm": 0.48167121410369873,
      "learning_rate": 1.542447163138569e-05,
      "loss": 0.0914,
      "step": 13520
    },
    {
      "epoch": 0.6865830752830093,
      "grad_norm": 0.48669248819351196,
      "learning_rate": 1.5422779498113273e-05,
      "loss": 0.0841,
      "step": 13525
    },
    {
      "epoch": 0.6868368952738718,
      "grad_norm": 0.40434154868125916,
      "learning_rate": 1.5421087364840856e-05,
      "loss": 0.0851,
      "step": 13530
    },
    {
      "epoch": 0.6870907152647342,
      "grad_norm": 0.5444993376731873,
      "learning_rate": 1.541939523156844e-05,
      "loss": 0.0977,
      "step": 13535
    },
    {
      "epoch": 0.6873445352555967,
      "grad_norm": 0.5064772367477417,
      "learning_rate": 1.5417703098296023e-05,
      "loss": 0.0787,
      "step": 13540
    },
    {
      "epoch": 0.6875983552464592,
      "grad_norm": 0.28774887323379517,
      "learning_rate": 1.5416010965023607e-05,
      "loss": 0.0955,
      "step": 13545
    },
    {
      "epoch": 0.6878521752373217,
      "grad_norm": 0.41530025005340576,
      "learning_rate": 1.541431883175119e-05,
      "loss": 0.069,
      "step": 13550
    },
    {
      "epoch": 0.6881059952281842,
      "grad_norm": 0.769633948802948,
      "learning_rate": 1.5412626698478774e-05,
      "loss": 0.0872,
      "step": 13555
    },
    {
      "epoch": 0.6883598152190467,
      "grad_norm": 0.47131699323654175,
      "learning_rate": 1.5410934565206358e-05,
      "loss": 0.0871,
      "step": 13560
    },
    {
      "epoch": 0.6886136352099091,
      "grad_norm": 0.3732922077178955,
      "learning_rate": 1.540924243193394e-05,
      "loss": 0.0847,
      "step": 13565
    },
    {
      "epoch": 0.6888674552007716,
      "grad_norm": 0.5022915005683899,
      "learning_rate": 1.5407550298661525e-05,
      "loss": 0.0887,
      "step": 13570
    },
    {
      "epoch": 0.6891212751916341,
      "grad_norm": 0.3835129141807556,
      "learning_rate": 1.5405858165389108e-05,
      "loss": 0.0825,
      "step": 13575
    },
    {
      "epoch": 0.6893750951824966,
      "grad_norm": 0.5404291749000549,
      "learning_rate": 1.540416603211669e-05,
      "loss": 0.082,
      "step": 13580
    },
    {
      "epoch": 0.689628915173359,
      "grad_norm": 0.3274235129356384,
      "learning_rate": 1.5402473898844275e-05,
      "loss": 0.0833,
      "step": 13585
    },
    {
      "epoch": 0.6898827351642215,
      "grad_norm": 0.3606513440608978,
      "learning_rate": 1.540078176557186e-05,
      "loss": 0.079,
      "step": 13590
    },
    {
      "epoch": 0.690136555155084,
      "grad_norm": 0.6473351716995239,
      "learning_rate": 1.5399089632299442e-05,
      "loss": 0.0856,
      "step": 13595
    },
    {
      "epoch": 0.6903903751459465,
      "grad_norm": 1.008514165878296,
      "learning_rate": 1.5397397499027026e-05,
      "loss": 0.0856,
      "step": 13600
    },
    {
      "epoch": 0.690644195136809,
      "grad_norm": 0.5139790177345276,
      "learning_rate": 1.5395705365754606e-05,
      "loss": 0.087,
      "step": 13605
    },
    {
      "epoch": 0.6908980151276715,
      "grad_norm": 0.3781962990760803,
      "learning_rate": 1.5394013232482193e-05,
      "loss": 0.0808,
      "step": 13610
    },
    {
      "epoch": 0.691151835118534,
      "grad_norm": 0.3120231628417969,
      "learning_rate": 1.5392321099209777e-05,
      "loss": 0.072,
      "step": 13615
    },
    {
      "epoch": 0.6914056551093964,
      "grad_norm": 0.3772776424884796,
      "learning_rate": 1.5390628965937357e-05,
      "loss": 0.0701,
      "step": 13620
    },
    {
      "epoch": 0.6916594751002589,
      "grad_norm": 0.4236140549182892,
      "learning_rate": 1.5388936832664944e-05,
      "loss": 0.0822,
      "step": 13625
    },
    {
      "epoch": 0.6919132950911214,
      "grad_norm": 0.4006426930427551,
      "learning_rate": 1.5387244699392524e-05,
      "loss": 0.0839,
      "step": 13630
    },
    {
      "epoch": 0.6921671150819838,
      "grad_norm": 0.45694807171821594,
      "learning_rate": 1.538555256612011e-05,
      "loss": 0.0738,
      "step": 13635
    },
    {
      "epoch": 0.6924209350728463,
      "grad_norm": 0.6357932090759277,
      "learning_rate": 1.538386043284769e-05,
      "loss": 0.0842,
      "step": 13640
    },
    {
      "epoch": 0.6926747550637088,
      "grad_norm": 0.5149153470993042,
      "learning_rate": 1.5382168299575274e-05,
      "loss": 0.0776,
      "step": 13645
    },
    {
      "epoch": 0.6929285750545713,
      "grad_norm": 0.8405686616897583,
      "learning_rate": 1.538047616630286e-05,
      "loss": 0.0821,
      "step": 13650
    },
    {
      "epoch": 0.6931823950454338,
      "grad_norm": 0.35849130153656006,
      "learning_rate": 1.537878403303044e-05,
      "loss": 0.0803,
      "step": 13655
    },
    {
      "epoch": 0.6934362150362963,
      "grad_norm": 0.41094881296157837,
      "learning_rate": 1.5377091899758025e-05,
      "loss": 0.0863,
      "step": 13660
    },
    {
      "epoch": 0.6936900350271588,
      "grad_norm": 0.4219074249267578,
      "learning_rate": 1.537539976648561e-05,
      "loss": 0.0951,
      "step": 13665
    },
    {
      "epoch": 0.6939438550180212,
      "grad_norm": 0.4253228008747101,
      "learning_rate": 1.5373707633213192e-05,
      "loss": 0.0918,
      "step": 13670
    },
    {
      "epoch": 0.6941976750088837,
      "grad_norm": 0.5601229071617126,
      "learning_rate": 1.5372015499940776e-05,
      "loss": 0.0704,
      "step": 13675
    },
    {
      "epoch": 0.6944514949997462,
      "grad_norm": 0.40687859058380127,
      "learning_rate": 1.537032336666836e-05,
      "loss": 0.0689,
      "step": 13680
    },
    {
      "epoch": 0.6947053149906086,
      "grad_norm": 0.4325862526893616,
      "learning_rate": 1.5368631233395943e-05,
      "loss": 0.0917,
      "step": 13685
    },
    {
      "epoch": 0.6949591349814711,
      "grad_norm": 0.44314250349998474,
      "learning_rate": 1.5366939100123526e-05,
      "loss": 0.0812,
      "step": 13690
    },
    {
      "epoch": 0.6952129549723336,
      "grad_norm": 0.35597068071365356,
      "learning_rate": 1.536524696685111e-05,
      "loss": 0.0767,
      "step": 13695
    },
    {
      "epoch": 0.6954667749631961,
      "grad_norm": 0.7931150794029236,
      "learning_rate": 1.5363554833578693e-05,
      "loss": 0.0794,
      "step": 13700
    },
    {
      "epoch": 0.6957205949540586,
      "grad_norm": 0.511711835861206,
      "learning_rate": 1.5361862700306277e-05,
      "loss": 0.074,
      "step": 13705
    },
    {
      "epoch": 0.6959744149449211,
      "grad_norm": 0.8357803225517273,
      "learning_rate": 1.536017056703386e-05,
      "loss": 0.0688,
      "step": 13710
    },
    {
      "epoch": 0.6962282349357836,
      "grad_norm": 0.3736538887023926,
      "learning_rate": 1.5358478433761444e-05,
      "loss": 0.0676,
      "step": 13715
    },
    {
      "epoch": 0.696482054926646,
      "grad_norm": 0.38975706696510315,
      "learning_rate": 1.5356786300489028e-05,
      "loss": 0.0779,
      "step": 13720
    },
    {
      "epoch": 0.6967358749175085,
      "grad_norm": 0.524750828742981,
      "learning_rate": 1.535509416721661e-05,
      "loss": 0.0827,
      "step": 13725
    },
    {
      "epoch": 0.696989694908371,
      "grad_norm": 0.44146502017974854,
      "learning_rate": 1.5353402033944195e-05,
      "loss": 0.08,
      "step": 13730
    },
    {
      "epoch": 0.6972435148992334,
      "grad_norm": 0.49009469151496887,
      "learning_rate": 1.5351709900671778e-05,
      "loss": 0.0832,
      "step": 13735
    },
    {
      "epoch": 0.6974973348900959,
      "grad_norm": 0.5824193358421326,
      "learning_rate": 1.5350017767399362e-05,
      "loss": 0.0844,
      "step": 13740
    },
    {
      "epoch": 0.6977511548809584,
      "grad_norm": 0.3237861394882202,
      "learning_rate": 1.5348325634126945e-05,
      "loss": 0.0803,
      "step": 13745
    },
    {
      "epoch": 0.6980049748718209,
      "grad_norm": 0.42894473671913147,
      "learning_rate": 1.534663350085453e-05,
      "loss": 0.0822,
      "step": 13750
    },
    {
      "epoch": 0.6982587948626834,
      "grad_norm": 0.43603384494781494,
      "learning_rate": 1.5344941367582112e-05,
      "loss": 0.0749,
      "step": 13755
    },
    {
      "epoch": 0.6985126148535459,
      "grad_norm": 0.3695114850997925,
      "learning_rate": 1.5343249234309696e-05,
      "loss": 0.0844,
      "step": 13760
    },
    {
      "epoch": 0.6987664348444084,
      "grad_norm": 0.43473580479621887,
      "learning_rate": 1.534155710103728e-05,
      "loss": 0.0835,
      "step": 13765
    },
    {
      "epoch": 0.6990202548352709,
      "grad_norm": 1.5757832527160645,
      "learning_rate": 1.5339864967764863e-05,
      "loss": 0.0821,
      "step": 13770
    },
    {
      "epoch": 0.6992740748261334,
      "grad_norm": 0.6827208399772644,
      "learning_rate": 1.5338172834492447e-05,
      "loss": 0.0784,
      "step": 13775
    },
    {
      "epoch": 0.6995278948169957,
      "grad_norm": 0.30773502588272095,
      "learning_rate": 1.533648070122003e-05,
      "loss": 0.1152,
      "step": 13780
    },
    {
      "epoch": 0.6997817148078582,
      "grad_norm": 0.4392739534378052,
      "learning_rate": 1.5334788567947614e-05,
      "loss": 0.0784,
      "step": 13785
    },
    {
      "epoch": 0.7000355347987207,
      "grad_norm": 0.7191630601882935,
      "learning_rate": 1.5333096434675197e-05,
      "loss": 0.0789,
      "step": 13790
    },
    {
      "epoch": 0.7002893547895832,
      "grad_norm": 0.2967609763145447,
      "learning_rate": 1.533140430140278e-05,
      "loss": 0.0778,
      "step": 13795
    },
    {
      "epoch": 0.7005431747804457,
      "grad_norm": 0.9909408688545227,
      "learning_rate": 1.5329712168130364e-05,
      "loss": 0.0884,
      "step": 13800
    },
    {
      "epoch": 0.7007969947713082,
      "grad_norm": 0.35766205191612244,
      "learning_rate": 1.5328020034857948e-05,
      "loss": 0.0733,
      "step": 13805
    },
    {
      "epoch": 0.7010508147621707,
      "grad_norm": 0.44659724831581116,
      "learning_rate": 1.5326327901585528e-05,
      "loss": 0.0882,
      "step": 13810
    },
    {
      "epoch": 0.7013046347530332,
      "grad_norm": 0.34097808599472046,
      "learning_rate": 1.5324635768313115e-05,
      "loss": 0.0785,
      "step": 13815
    },
    {
      "epoch": 0.7015584547438957,
      "grad_norm": 0.34435734152793884,
      "learning_rate": 1.5322943635040695e-05,
      "loss": 0.0818,
      "step": 13820
    },
    {
      "epoch": 0.7018122747347582,
      "grad_norm": 0.4698044955730438,
      "learning_rate": 1.532125150176828e-05,
      "loss": 0.0848,
      "step": 13825
    },
    {
      "epoch": 0.7020660947256205,
      "grad_norm": 0.4470899999141693,
      "learning_rate": 1.5319559368495866e-05,
      "loss": 0.0815,
      "step": 13830
    },
    {
      "epoch": 0.702319914716483,
      "grad_norm": 0.3169173300266266,
      "learning_rate": 1.5317867235223446e-05,
      "loss": 0.0738,
      "step": 13835
    },
    {
      "epoch": 0.7025737347073455,
      "grad_norm": 0.8150441646575928,
      "learning_rate": 1.5316175101951033e-05,
      "loss": 0.0763,
      "step": 13840
    },
    {
      "epoch": 0.702827554698208,
      "grad_norm": 0.35342103242874146,
      "learning_rate": 1.5314482968678613e-05,
      "loss": 0.0833,
      "step": 13845
    },
    {
      "epoch": 0.7030813746890705,
      "grad_norm": 0.771515429019928,
      "learning_rate": 1.5312790835406196e-05,
      "loss": 0.0825,
      "step": 13850
    },
    {
      "epoch": 0.703335194679933,
      "grad_norm": 0.3441005051136017,
      "learning_rate": 1.5311098702133783e-05,
      "loss": 0.0842,
      "step": 13855
    },
    {
      "epoch": 0.7035890146707955,
      "grad_norm": 0.45794785022735596,
      "learning_rate": 1.5309406568861363e-05,
      "loss": 0.0833,
      "step": 13860
    },
    {
      "epoch": 0.703842834661658,
      "grad_norm": 0.41484132409095764,
      "learning_rate": 1.5307714435588947e-05,
      "loss": 0.0749,
      "step": 13865
    },
    {
      "epoch": 0.7040966546525205,
      "grad_norm": 0.38982802629470825,
      "learning_rate": 1.530602230231653e-05,
      "loss": 0.0762,
      "step": 13870
    },
    {
      "epoch": 0.704350474643383,
      "grad_norm": 0.3871098458766937,
      "learning_rate": 1.5304330169044114e-05,
      "loss": 0.0876,
      "step": 13875
    },
    {
      "epoch": 0.7046042946342453,
      "grad_norm": 0.3723105192184448,
      "learning_rate": 1.53026380357717e-05,
      "loss": 0.0687,
      "step": 13880
    },
    {
      "epoch": 0.7048581146251078,
      "grad_norm": 0.4953676462173462,
      "learning_rate": 1.530094590249928e-05,
      "loss": 0.0743,
      "step": 13885
    },
    {
      "epoch": 0.7051119346159703,
      "grad_norm": 0.43342849612236023,
      "learning_rate": 1.5299253769226865e-05,
      "loss": 0.0738,
      "step": 13890
    },
    {
      "epoch": 0.7053657546068328,
      "grad_norm": 0.6391082406044006,
      "learning_rate": 1.5297561635954448e-05,
      "loss": 0.0903,
      "step": 13895
    },
    {
      "epoch": 0.7056195745976953,
      "grad_norm": 0.46714121103286743,
      "learning_rate": 1.5295869502682032e-05,
      "loss": 0.0842,
      "step": 13900
    },
    {
      "epoch": 0.7058733945885578,
      "grad_norm": 0.5677165985107422,
      "learning_rate": 1.5294177369409615e-05,
      "loss": 0.0879,
      "step": 13905
    },
    {
      "epoch": 0.7061272145794203,
      "grad_norm": 0.4069088101387024,
      "learning_rate": 1.52924852361372e-05,
      "loss": 0.0744,
      "step": 13910
    },
    {
      "epoch": 0.7063810345702828,
      "grad_norm": 0.44778573513031006,
      "learning_rate": 1.5290793102864782e-05,
      "loss": 0.073,
      "step": 13915
    },
    {
      "epoch": 0.7066348545611453,
      "grad_norm": 0.40604692697525024,
      "learning_rate": 1.5289100969592366e-05,
      "loss": 0.089,
      "step": 13920
    },
    {
      "epoch": 0.7068886745520078,
      "grad_norm": 0.3721408545970917,
      "learning_rate": 1.528740883631995e-05,
      "loss": 0.0789,
      "step": 13925
    },
    {
      "epoch": 0.7071424945428701,
      "grad_norm": 0.31769973039627075,
      "learning_rate": 1.5285716703047533e-05,
      "loss": 0.0816,
      "step": 13930
    },
    {
      "epoch": 0.7073963145337326,
      "grad_norm": 0.4818114638328552,
      "learning_rate": 1.5284024569775117e-05,
      "loss": 0.0868,
      "step": 13935
    },
    {
      "epoch": 0.7076501345245951,
      "grad_norm": 0.3946104943752289,
      "learning_rate": 1.52823324365027e-05,
      "loss": 0.0707,
      "step": 13940
    },
    {
      "epoch": 0.7079039545154576,
      "grad_norm": 0.6479266285896301,
      "learning_rate": 1.5280640303230284e-05,
      "loss": 0.0821,
      "step": 13945
    },
    {
      "epoch": 0.7081577745063201,
      "grad_norm": 0.32928308844566345,
      "learning_rate": 1.5278948169957867e-05,
      "loss": 0.0737,
      "step": 13950
    },
    {
      "epoch": 0.7084115944971826,
      "grad_norm": 0.39813652634620667,
      "learning_rate": 1.527725603668545e-05,
      "loss": 0.0796,
      "step": 13955
    },
    {
      "epoch": 0.7086654144880451,
      "grad_norm": 0.4560108780860901,
      "learning_rate": 1.5275563903413034e-05,
      "loss": 0.0915,
      "step": 13960
    },
    {
      "epoch": 0.7089192344789076,
      "grad_norm": 1.1247360706329346,
      "learning_rate": 1.5273871770140618e-05,
      "loss": 0.0962,
      "step": 13965
    },
    {
      "epoch": 0.7091730544697701,
      "grad_norm": 0.43578168749809265,
      "learning_rate": 1.52721796368682e-05,
      "loss": 0.0834,
      "step": 13970
    },
    {
      "epoch": 0.7094268744606326,
      "grad_norm": 0.45874667167663574,
      "learning_rate": 1.5270487503595785e-05,
      "loss": 0.072,
      "step": 13975
    },
    {
      "epoch": 0.709680694451495,
      "grad_norm": 0.5207342505455017,
      "learning_rate": 1.526879537032337e-05,
      "loss": 0.0886,
      "step": 13980
    },
    {
      "epoch": 0.7099345144423574,
      "grad_norm": 0.39302659034729004,
      "learning_rate": 1.5267103237050952e-05,
      "loss": 0.0836,
      "step": 13985
    },
    {
      "epoch": 0.7101883344332199,
      "grad_norm": 0.44471731781959534,
      "learning_rate": 1.5265411103778536e-05,
      "loss": 0.1,
      "step": 13990
    },
    {
      "epoch": 0.7104421544240824,
      "grad_norm": 0.386536180973053,
      "learning_rate": 1.526371897050612e-05,
      "loss": 0.0816,
      "step": 13995
    },
    {
      "epoch": 0.7106959744149449,
      "grad_norm": 0.4404861330986023,
      "learning_rate": 1.52620268372337e-05,
      "loss": 0.0936,
      "step": 14000
    },
    {
      "epoch": 0.7109497944058074,
      "grad_norm": 0.7851703763008118,
      "learning_rate": 1.5260334703961286e-05,
      "loss": 0.0798,
      "step": 14005
    },
    {
      "epoch": 0.7112036143966699,
      "grad_norm": 0.6960633993148804,
      "learning_rate": 1.5258642570688868e-05,
      "loss": 0.0838,
      "step": 14010
    },
    {
      "epoch": 0.7114574343875324,
      "grad_norm": 0.47619956731796265,
      "learning_rate": 1.5256950437416452e-05,
      "loss": 0.0904,
      "step": 14015
    },
    {
      "epoch": 0.7117112543783949,
      "grad_norm": 0.6438000202178955,
      "learning_rate": 1.5255258304144037e-05,
      "loss": 0.0736,
      "step": 14020
    },
    {
      "epoch": 0.7119650743692574,
      "grad_norm": 0.4882700741291046,
      "learning_rate": 1.5253566170871619e-05,
      "loss": 0.0878,
      "step": 14025
    },
    {
      "epoch": 0.7122188943601198,
      "grad_norm": 0.36032137274742126,
      "learning_rate": 1.5251874037599204e-05,
      "loss": 0.0772,
      "step": 14030
    },
    {
      "epoch": 0.7124727143509823,
      "grad_norm": 0.5064921379089355,
      "learning_rate": 1.5250181904326786e-05,
      "loss": 0.0799,
      "step": 14035
    },
    {
      "epoch": 0.7127265343418447,
      "grad_norm": 0.586258590221405,
      "learning_rate": 1.524848977105437e-05,
      "loss": 0.0856,
      "step": 14040
    },
    {
      "epoch": 0.7129803543327072,
      "grad_norm": 0.4364579916000366,
      "learning_rate": 1.5246797637781953e-05,
      "loss": 0.0797,
      "step": 14045
    },
    {
      "epoch": 0.7132341743235697,
      "grad_norm": 0.30518513917922974,
      "learning_rate": 1.5245105504509536e-05,
      "loss": 0.0821,
      "step": 14050
    },
    {
      "epoch": 0.7134879943144322,
      "grad_norm": 0.3665226399898529,
      "learning_rate": 1.5243413371237118e-05,
      "loss": 0.0899,
      "step": 14055
    },
    {
      "epoch": 0.7137418143052947,
      "grad_norm": 0.5251713395118713,
      "learning_rate": 1.5241721237964704e-05,
      "loss": 0.0837,
      "step": 14060
    },
    {
      "epoch": 0.7139956342961572,
      "grad_norm": 0.37838253378868103,
      "learning_rate": 1.5240029104692287e-05,
      "loss": 0.0744,
      "step": 14065
    },
    {
      "epoch": 0.7142494542870197,
      "grad_norm": 0.4053058922290802,
      "learning_rate": 1.5238336971419869e-05,
      "loss": 0.0696,
      "step": 14070
    },
    {
      "epoch": 0.7145032742778821,
      "grad_norm": 0.4535035192966461,
      "learning_rate": 1.5236644838147454e-05,
      "loss": 0.0718,
      "step": 14075
    },
    {
      "epoch": 0.7147570942687446,
      "grad_norm": 0.2996232807636261,
      "learning_rate": 1.5234952704875036e-05,
      "loss": 0.0823,
      "step": 14080
    },
    {
      "epoch": 0.715010914259607,
      "grad_norm": 0.40874767303466797,
      "learning_rate": 1.5233260571602621e-05,
      "loss": 0.0793,
      "step": 14085
    },
    {
      "epoch": 0.7152647342504695,
      "grad_norm": 0.34170955419540405,
      "learning_rate": 1.5231568438330205e-05,
      "loss": 0.0722,
      "step": 14090
    },
    {
      "epoch": 0.715518554241332,
      "grad_norm": 0.42445680499076843,
      "learning_rate": 1.5229876305057787e-05,
      "loss": 0.0707,
      "step": 14095
    },
    {
      "epoch": 0.7157723742321945,
      "grad_norm": 0.5553193688392639,
      "learning_rate": 1.5228184171785372e-05,
      "loss": 0.0758,
      "step": 14100
    },
    {
      "epoch": 0.716026194223057,
      "grad_norm": 0.3143168091773987,
      "learning_rate": 1.5226492038512954e-05,
      "loss": 0.0619,
      "step": 14105
    },
    {
      "epoch": 0.7162800142139195,
      "grad_norm": 0.633912980556488,
      "learning_rate": 1.5224799905240537e-05,
      "loss": 0.0746,
      "step": 14110
    },
    {
      "epoch": 0.716533834204782,
      "grad_norm": 0.3377459943294525,
      "learning_rate": 1.5223107771968123e-05,
      "loss": 0.0761,
      "step": 14115
    },
    {
      "epoch": 0.7167876541956445,
      "grad_norm": 0.3944687247276306,
      "learning_rate": 1.5221415638695704e-05,
      "loss": 0.0766,
      "step": 14120
    },
    {
      "epoch": 0.7170414741865069,
      "grad_norm": 0.4088902771472931,
      "learning_rate": 1.521972350542329e-05,
      "loss": 0.077,
      "step": 14125
    },
    {
      "epoch": 0.7172952941773694,
      "grad_norm": 0.4192088842391968,
      "learning_rate": 1.5218031372150872e-05,
      "loss": 0.0778,
      "step": 14130
    },
    {
      "epoch": 0.7175491141682319,
      "grad_norm": 0.36348608136177063,
      "learning_rate": 1.5216339238878455e-05,
      "loss": 0.0854,
      "step": 14135
    },
    {
      "epoch": 0.7178029341590944,
      "grad_norm": 0.3958783447742462,
      "learning_rate": 1.521464710560604e-05,
      "loss": 0.0742,
      "step": 14140
    },
    {
      "epoch": 0.7180567541499568,
      "grad_norm": 0.44732987880706787,
      "learning_rate": 1.5212954972333622e-05,
      "loss": 0.076,
      "step": 14145
    },
    {
      "epoch": 0.7183105741408193,
      "grad_norm": 1.2303112745285034,
      "learning_rate": 1.5211262839061206e-05,
      "loss": 0.0784,
      "step": 14150
    },
    {
      "epoch": 0.7185643941316818,
      "grad_norm": 0.4170047342777252,
      "learning_rate": 1.520957070578879e-05,
      "loss": 0.0863,
      "step": 14155
    },
    {
      "epoch": 0.7188182141225443,
      "grad_norm": 0.3879324793815613,
      "learning_rate": 1.5207878572516373e-05,
      "loss": 0.0756,
      "step": 14160
    },
    {
      "epoch": 0.7190720341134068,
      "grad_norm": 0.516608476638794,
      "learning_rate": 1.5206186439243955e-05,
      "loss": 0.0821,
      "step": 14165
    },
    {
      "epoch": 0.7193258541042693,
      "grad_norm": 0.4164332151412964,
      "learning_rate": 1.520449430597154e-05,
      "loss": 0.0678,
      "step": 14170
    },
    {
      "epoch": 0.7195796740951317,
      "grad_norm": 0.4437199532985687,
      "learning_rate": 1.5202802172699123e-05,
      "loss": 0.0825,
      "step": 14175
    },
    {
      "epoch": 0.7198334940859942,
      "grad_norm": 0.49337777495384216,
      "learning_rate": 1.5201110039426707e-05,
      "loss": 0.106,
      "step": 14180
    },
    {
      "epoch": 0.7200873140768567,
      "grad_norm": 0.43861648440361023,
      "learning_rate": 1.519941790615429e-05,
      "loss": 0.0871,
      "step": 14185
    },
    {
      "epoch": 0.7203411340677192,
      "grad_norm": 194.89373779296875,
      "learning_rate": 1.5197725772881872e-05,
      "loss": 0.0967,
      "step": 14190
    },
    {
      "epoch": 0.7205949540585816,
      "grad_norm": 0.3648269474506378,
      "learning_rate": 1.5196033639609458e-05,
      "loss": 0.0855,
      "step": 14195
    },
    {
      "epoch": 0.7208487740494441,
      "grad_norm": 0.4006633758544922,
      "learning_rate": 1.5194341506337041e-05,
      "loss": 0.0798,
      "step": 14200
    },
    {
      "epoch": 0.7211025940403066,
      "grad_norm": 0.4306533932685852,
      "learning_rate": 1.5192649373064623e-05,
      "loss": 0.0915,
      "step": 14205
    },
    {
      "epoch": 0.7213564140311691,
      "grad_norm": 0.8204346895217896,
      "learning_rate": 1.5190957239792208e-05,
      "loss": 0.0843,
      "step": 14210
    },
    {
      "epoch": 0.7216102340220316,
      "grad_norm": 0.40292853116989136,
      "learning_rate": 1.518926510651979e-05,
      "loss": 0.0846,
      "step": 14215
    },
    {
      "epoch": 0.7218640540128941,
      "grad_norm": 0.4678119122982025,
      "learning_rate": 1.5187572973247374e-05,
      "loss": 0.0818,
      "step": 14220
    },
    {
      "epoch": 0.7221178740037565,
      "grad_norm": 0.5010687708854675,
      "learning_rate": 1.5185880839974957e-05,
      "loss": 0.081,
      "step": 14225
    },
    {
      "epoch": 0.722371693994619,
      "grad_norm": 1.1079010963439941,
      "learning_rate": 1.518418870670254e-05,
      "loss": 0.0954,
      "step": 14230
    },
    {
      "epoch": 0.7226255139854815,
      "grad_norm": 0.32439419627189636,
      "learning_rate": 1.5182496573430126e-05,
      "loss": 0.0801,
      "step": 14235
    },
    {
      "epoch": 0.722879333976344,
      "grad_norm": 0.4611896276473999,
      "learning_rate": 1.5180804440157708e-05,
      "loss": 0.07,
      "step": 14240
    },
    {
      "epoch": 0.7231331539672065,
      "grad_norm": 0.45286303758621216,
      "learning_rate": 1.5179112306885291e-05,
      "loss": 0.0791,
      "step": 14245
    },
    {
      "epoch": 0.723386973958069,
      "grad_norm": 0.6157929301261902,
      "learning_rate": 1.5177420173612875e-05,
      "loss": 0.0787,
      "step": 14250
    },
    {
      "epoch": 0.7236407939489314,
      "grad_norm": 0.4677484333515167,
      "learning_rate": 1.5175728040340458e-05,
      "loss": 0.078,
      "step": 14255
    },
    {
      "epoch": 0.7238946139397939,
      "grad_norm": 0.4395294487476349,
      "learning_rate": 1.517403590706804e-05,
      "loss": 0.0879,
      "step": 14260
    },
    {
      "epoch": 0.7241484339306564,
      "grad_norm": 0.48061785101890564,
      "learning_rate": 1.5172343773795626e-05,
      "loss": 0.0812,
      "step": 14265
    },
    {
      "epoch": 0.7244022539215189,
      "grad_norm": 0.378859281539917,
      "learning_rate": 1.5170651640523209e-05,
      "loss": 0.0837,
      "step": 14270
    },
    {
      "epoch": 0.7246560739123813,
      "grad_norm": 0.40391427278518677,
      "learning_rate": 1.5168959507250793e-05,
      "loss": 0.0909,
      "step": 14275
    },
    {
      "epoch": 0.7249098939032438,
      "grad_norm": 0.3788898289203644,
      "learning_rate": 1.5167267373978376e-05,
      "loss": 0.0832,
      "step": 14280
    },
    {
      "epoch": 0.7251637138941063,
      "grad_norm": 0.2744353413581848,
      "learning_rate": 1.5165575240705958e-05,
      "loss": 0.0712,
      "step": 14285
    },
    {
      "epoch": 0.7254175338849688,
      "grad_norm": 0.49795615673065186,
      "learning_rate": 1.5163883107433543e-05,
      "loss": 0.0686,
      "step": 14290
    },
    {
      "epoch": 0.7256713538758313,
      "grad_norm": 1.3748142719268799,
      "learning_rate": 1.5162190974161127e-05,
      "loss": 0.0906,
      "step": 14295
    },
    {
      "epoch": 0.7259251738666938,
      "grad_norm": 0.409864604473114,
      "learning_rate": 1.5160498840888709e-05,
      "loss": 0.082,
      "step": 14300
    },
    {
      "epoch": 0.7261789938575562,
      "grad_norm": 0.4577494263648987,
      "learning_rate": 1.5158806707616294e-05,
      "loss": 0.0847,
      "step": 14305
    },
    {
      "epoch": 0.7264328138484187,
      "grad_norm": 0.3561350703239441,
      "learning_rate": 1.5157114574343876e-05,
      "loss": 0.0878,
      "step": 14310
    },
    {
      "epoch": 0.7266866338392812,
      "grad_norm": 0.3978186547756195,
      "learning_rate": 1.515542244107146e-05,
      "loss": 0.0881,
      "step": 14315
    },
    {
      "epoch": 0.7269404538301436,
      "grad_norm": 0.2696934938430786,
      "learning_rate": 1.5153730307799045e-05,
      "loss": 0.0994,
      "step": 14320
    },
    {
      "epoch": 0.7271942738210061,
      "grad_norm": 0.47918376326560974,
      "learning_rate": 1.5152038174526626e-05,
      "loss": 0.0808,
      "step": 14325
    },
    {
      "epoch": 0.7274480938118686,
      "grad_norm": 0.39144304394721985,
      "learning_rate": 1.5150346041254212e-05,
      "loss": 0.0863,
      "step": 14330
    },
    {
      "epoch": 0.7277019138027311,
      "grad_norm": 0.39072200655937195,
      "learning_rate": 1.5148653907981793e-05,
      "loss": 0.0732,
      "step": 14335
    },
    {
      "epoch": 0.7279557337935936,
      "grad_norm": 0.3815653324127197,
      "learning_rate": 1.5146961774709377e-05,
      "loss": 0.0839,
      "step": 14340
    },
    {
      "epoch": 0.7282095537844561,
      "grad_norm": 0.4559372067451477,
      "learning_rate": 1.5145269641436962e-05,
      "loss": 0.0841,
      "step": 14345
    },
    {
      "epoch": 0.7284633737753186,
      "grad_norm": 0.42653796076774597,
      "learning_rate": 1.5143577508164544e-05,
      "loss": 0.0842,
      "step": 14350
    },
    {
      "epoch": 0.728717193766181,
      "grad_norm": 0.4985381066799164,
      "learning_rate": 1.5141885374892128e-05,
      "loss": 0.0883,
      "step": 14355
    },
    {
      "epoch": 0.7289710137570435,
      "grad_norm": 0.5497139692306519,
      "learning_rate": 1.5140193241619711e-05,
      "loss": 0.0819,
      "step": 14360
    },
    {
      "epoch": 0.729224833747906,
      "grad_norm": 0.595309853553772,
      "learning_rate": 1.5138501108347295e-05,
      "loss": 0.0848,
      "step": 14365
    },
    {
      "epoch": 0.7294786537387684,
      "grad_norm": 0.47207218408584595,
      "learning_rate": 1.513680897507488e-05,
      "loss": 0.0805,
      "step": 14370
    },
    {
      "epoch": 0.7297324737296309,
      "grad_norm": 0.38911616802215576,
      "learning_rate": 1.5135116841802462e-05,
      "loss": 0.0855,
      "step": 14375
    },
    {
      "epoch": 0.7299862937204934,
      "grad_norm": 0.6578527092933655,
      "learning_rate": 1.5133424708530045e-05,
      "loss": 0.0821,
      "step": 14380
    },
    {
      "epoch": 0.7302401137113559,
      "grad_norm": 0.3873518705368042,
      "learning_rate": 1.5131732575257629e-05,
      "loss": 0.0787,
      "step": 14385
    },
    {
      "epoch": 0.7304939337022184,
      "grad_norm": 0.3727337718009949,
      "learning_rate": 1.5130040441985212e-05,
      "loss": 0.0711,
      "step": 14390
    },
    {
      "epoch": 0.7307477536930809,
      "grad_norm": 0.3558390140533447,
      "learning_rate": 1.5128348308712794e-05,
      "loss": 0.0633,
      "step": 14395
    },
    {
      "epoch": 0.7310015736839434,
      "grad_norm": 0.44894295930862427,
      "learning_rate": 1.512665617544038e-05,
      "loss": 0.0775,
      "step": 14400
    },
    {
      "epoch": 0.7312553936748059,
      "grad_norm": 0.44867026805877686,
      "learning_rate": 1.5124964042167961e-05,
      "loss": 0.0862,
      "step": 14405
    },
    {
      "epoch": 0.7315092136656683,
      "grad_norm": 0.3655999004840851,
      "learning_rate": 1.5123271908895545e-05,
      "loss": 0.0851,
      "step": 14410
    },
    {
      "epoch": 0.7317630336565308,
      "grad_norm": 0.4013903737068176,
      "learning_rate": 1.512157977562313e-05,
      "loss": 0.0861,
      "step": 14415
    },
    {
      "epoch": 0.7320168536473932,
      "grad_norm": 0.3827629089355469,
      "learning_rate": 1.5119887642350712e-05,
      "loss": 0.0884,
      "step": 14420
    },
    {
      "epoch": 0.7322706736382557,
      "grad_norm": 0.48554348945617676,
      "learning_rate": 1.5118195509078297e-05,
      "loss": 0.0761,
      "step": 14425
    },
    {
      "epoch": 0.7325244936291182,
      "grad_norm": 0.3680451214313507,
      "learning_rate": 1.5116503375805879e-05,
      "loss": 0.0731,
      "step": 14430
    },
    {
      "epoch": 0.7327783136199807,
      "grad_norm": 0.47662273049354553,
      "learning_rate": 1.5114811242533463e-05,
      "loss": 0.0814,
      "step": 14435
    },
    {
      "epoch": 0.7330321336108432,
      "grad_norm": 0.4592245817184448,
      "learning_rate": 1.5113119109261048e-05,
      "loss": 0.0804,
      "step": 14440
    },
    {
      "epoch": 0.7332859536017057,
      "grad_norm": 0.4146338701248169,
      "learning_rate": 1.511142697598863e-05,
      "loss": 0.0902,
      "step": 14445
    },
    {
      "epoch": 0.7335397735925682,
      "grad_norm": 0.4722953736782074,
      "learning_rate": 1.5109734842716213e-05,
      "loss": 0.0863,
      "step": 14450
    },
    {
      "epoch": 0.7337935935834307,
      "grad_norm": 0.7869719862937927,
      "learning_rate": 1.5108042709443797e-05,
      "loss": 0.0794,
      "step": 14455
    },
    {
      "epoch": 0.7340474135742932,
      "grad_norm": 0.4236357808113098,
      "learning_rate": 1.510635057617138e-05,
      "loss": 0.0723,
      "step": 14460
    },
    {
      "epoch": 0.7343012335651556,
      "grad_norm": 0.38017693161964417,
      "learning_rate": 1.5104658442898962e-05,
      "loss": 0.0717,
      "step": 14465
    },
    {
      "epoch": 0.734555053556018,
      "grad_norm": 0.3802855908870697,
      "learning_rate": 1.5102966309626547e-05,
      "loss": 0.0795,
      "step": 14470
    },
    {
      "epoch": 0.7348088735468805,
      "grad_norm": 0.506759762763977,
      "learning_rate": 1.5101274176354131e-05,
      "loss": 0.0783,
      "step": 14475
    },
    {
      "epoch": 0.735062693537743,
      "grad_norm": 0.6665279269218445,
      "learning_rate": 1.5099582043081715e-05,
      "loss": 0.0778,
      "step": 14480
    },
    {
      "epoch": 0.7353165135286055,
      "grad_norm": 0.5124176144599915,
      "learning_rate": 1.5097889909809298e-05,
      "loss": 0.0895,
      "step": 14485
    },
    {
      "epoch": 0.735570333519468,
      "grad_norm": 0.3736747205257416,
      "learning_rate": 1.509619777653688e-05,
      "loss": 0.0754,
      "step": 14490
    },
    {
      "epoch": 0.7358241535103305,
      "grad_norm": 0.5663830637931824,
      "learning_rate": 1.5094505643264465e-05,
      "loss": 0.0744,
      "step": 14495
    },
    {
      "epoch": 0.736077973501193,
      "grad_norm": 0.45994728803634644,
      "learning_rate": 1.5092813509992049e-05,
      "loss": 0.0805,
      "step": 14500
    },
    {
      "epoch": 0.7363317934920555,
      "grad_norm": 0.4209723472595215,
      "learning_rate": 1.509112137671963e-05,
      "loss": 0.07,
      "step": 14505
    },
    {
      "epoch": 0.736585613482918,
      "grad_norm": 0.3272269666194916,
      "learning_rate": 1.5089429243447216e-05,
      "loss": 0.0657,
      "step": 14510
    },
    {
      "epoch": 0.7368394334737804,
      "grad_norm": 0.4161807894706726,
      "learning_rate": 1.5087737110174798e-05,
      "loss": 0.0726,
      "step": 14515
    },
    {
      "epoch": 0.7370932534646428,
      "grad_norm": 0.43482705950737,
      "learning_rate": 1.5086044976902383e-05,
      "loss": 0.0788,
      "step": 14520
    },
    {
      "epoch": 0.7373470734555053,
      "grad_norm": 0.5974509716033936,
      "learning_rate": 1.5084352843629966e-05,
      "loss": 0.0749,
      "step": 14525
    },
    {
      "epoch": 0.7376008934463678,
      "grad_norm": 0.5275264978408813,
      "learning_rate": 1.5082660710357548e-05,
      "loss": 0.0762,
      "step": 14530
    },
    {
      "epoch": 0.7378547134372303,
      "grad_norm": 0.4393943250179291,
      "learning_rate": 1.5080968577085134e-05,
      "loss": 0.0831,
      "step": 14535
    },
    {
      "epoch": 0.7381085334280928,
      "grad_norm": 0.3507572114467621,
      "learning_rate": 1.5079276443812715e-05,
      "loss": 0.0859,
      "step": 14540
    },
    {
      "epoch": 0.7383623534189553,
      "grad_norm": 0.3876013457775116,
      "learning_rate": 1.5077584310540299e-05,
      "loss": 0.0779,
      "step": 14545
    },
    {
      "epoch": 0.7386161734098178,
      "grad_norm": 0.37205737829208374,
      "learning_rate": 1.5075892177267884e-05,
      "loss": 0.0737,
      "step": 14550
    },
    {
      "epoch": 0.7388699934006803,
      "grad_norm": 0.4882403314113617,
      "learning_rate": 1.5074200043995466e-05,
      "loss": 0.0878,
      "step": 14555
    },
    {
      "epoch": 0.7391238133915428,
      "grad_norm": 0.5520374774932861,
      "learning_rate": 1.507250791072305e-05,
      "loss": 0.0966,
      "step": 14560
    },
    {
      "epoch": 0.7393776333824053,
      "grad_norm": 0.367116779088974,
      "learning_rate": 1.5070815777450633e-05,
      "loss": 0.0819,
      "step": 14565
    },
    {
      "epoch": 0.7396314533732676,
      "grad_norm": 0.4586057960987091,
      "learning_rate": 1.5069123644178217e-05,
      "loss": 0.0776,
      "step": 14570
    },
    {
      "epoch": 0.7398852733641301,
      "grad_norm": 0.4211220145225525,
      "learning_rate": 1.5067431510905802e-05,
      "loss": 0.0726,
      "step": 14575
    },
    {
      "epoch": 0.7401390933549926,
      "grad_norm": 0.4111827611923218,
      "learning_rate": 1.5065739377633384e-05,
      "loss": 0.0803,
      "step": 14580
    },
    {
      "epoch": 0.7403929133458551,
      "grad_norm": 0.49343472719192505,
      "learning_rate": 1.5064047244360966e-05,
      "loss": 0.0835,
      "step": 14585
    },
    {
      "epoch": 0.7406467333367176,
      "grad_norm": 0.3709668815135956,
      "learning_rate": 1.506235511108855e-05,
      "loss": 0.0845,
      "step": 14590
    },
    {
      "epoch": 0.7409005533275801,
      "grad_norm": 0.45862144231796265,
      "learning_rate": 1.5060662977816134e-05,
      "loss": 0.0886,
      "step": 14595
    },
    {
      "epoch": 0.7411543733184426,
      "grad_norm": 0.4751611351966858,
      "learning_rate": 1.5058970844543716e-05,
      "loss": 0.0785,
      "step": 14600
    },
    {
      "epoch": 0.7414081933093051,
      "grad_norm": 0.302118718624115,
      "learning_rate": 1.5057278711271301e-05,
      "loss": 0.0766,
      "step": 14605
    },
    {
      "epoch": 0.7416620133001676,
      "grad_norm": 0.5119585990905762,
      "learning_rate": 1.5055586577998883e-05,
      "loss": 0.0839,
      "step": 14610
    },
    {
      "epoch": 0.74191583329103,
      "grad_norm": 0.3752945363521576,
      "learning_rate": 1.5053894444726469e-05,
      "loss": 0.076,
      "step": 14615
    },
    {
      "epoch": 0.7421696532818924,
      "grad_norm": 0.5176377296447754,
      "learning_rate": 1.5052202311454052e-05,
      "loss": 0.0817,
      "step": 14620
    },
    {
      "epoch": 0.7424234732727549,
      "grad_norm": 0.44988998770713806,
      "learning_rate": 1.5050510178181634e-05,
      "loss": 0.0824,
      "step": 14625
    },
    {
      "epoch": 0.7426772932636174,
      "grad_norm": 0.3214319348335266,
      "learning_rate": 1.504881804490922e-05,
      "loss": 0.0744,
      "step": 14630
    },
    {
      "epoch": 0.7429311132544799,
      "grad_norm": 0.5503107905387878,
      "learning_rate": 1.5047125911636801e-05,
      "loss": 0.0731,
      "step": 14635
    },
    {
      "epoch": 0.7431849332453424,
      "grad_norm": 1.3438857793807983,
      "learning_rate": 1.5045433778364385e-05,
      "loss": 0.0832,
      "step": 14640
    },
    {
      "epoch": 0.7434387532362049,
      "grad_norm": 0.447172075510025,
      "learning_rate": 1.504374164509197e-05,
      "loss": 0.072,
      "step": 14645
    },
    {
      "epoch": 0.7436925732270674,
      "grad_norm": 0.39440739154815674,
      "learning_rate": 1.5042049511819552e-05,
      "loss": 0.0921,
      "step": 14650
    },
    {
      "epoch": 0.7439463932179299,
      "grad_norm": 0.39479300379753113,
      "learning_rate": 1.5040357378547135e-05,
      "loss": 0.0791,
      "step": 14655
    },
    {
      "epoch": 0.7442002132087924,
      "grad_norm": 0.5535449385643005,
      "learning_rate": 1.5038665245274719e-05,
      "loss": 0.0776,
      "step": 14660
    },
    {
      "epoch": 0.7444540331996548,
      "grad_norm": 0.3662354648113251,
      "learning_rate": 1.5036973112002302e-05,
      "loss": 0.0746,
      "step": 14665
    },
    {
      "epoch": 0.7447078531905172,
      "grad_norm": 0.34598585963249207,
      "learning_rate": 1.5035280978729888e-05,
      "loss": 0.0781,
      "step": 14670
    },
    {
      "epoch": 0.7449616731813797,
      "grad_norm": 0.43272921442985535,
      "learning_rate": 1.503358884545747e-05,
      "loss": 0.0801,
      "step": 14675
    },
    {
      "epoch": 0.7452154931722422,
      "grad_norm": 0.3881741464138031,
      "learning_rate": 1.5031896712185053e-05,
      "loss": 0.0823,
      "step": 14680
    },
    {
      "epoch": 0.7454693131631047,
      "grad_norm": 0.40420830249786377,
      "learning_rate": 1.5030204578912636e-05,
      "loss": 0.0881,
      "step": 14685
    },
    {
      "epoch": 0.7457231331539672,
      "grad_norm": 0.6293859481811523,
      "learning_rate": 1.502851244564022e-05,
      "loss": 0.0792,
      "step": 14690
    },
    {
      "epoch": 0.7459769531448297,
      "grad_norm": 0.6415058970451355,
      "learning_rate": 1.5026820312367802e-05,
      "loss": 0.0787,
      "step": 14695
    },
    {
      "epoch": 0.7462307731356922,
      "grad_norm": 0.9900127053260803,
      "learning_rate": 1.5025128179095387e-05,
      "loss": 0.08,
      "step": 14700
    },
    {
      "epoch": 0.7464845931265547,
      "grad_norm": 0.39871150255203247,
      "learning_rate": 1.502343604582297e-05,
      "loss": 0.0727,
      "step": 14705
    },
    {
      "epoch": 0.7467384131174172,
      "grad_norm": 0.46820712089538574,
      "learning_rate": 1.5021743912550553e-05,
      "loss": 0.0884,
      "step": 14710
    },
    {
      "epoch": 0.7469922331082796,
      "grad_norm": 0.8658115267753601,
      "learning_rate": 1.5020051779278138e-05,
      "loss": 0.0752,
      "step": 14715
    },
    {
      "epoch": 0.747246053099142,
      "grad_norm": 0.5432624220848083,
      "learning_rate": 1.501835964600572e-05,
      "loss": 0.072,
      "step": 14720
    },
    {
      "epoch": 0.7474998730900045,
      "grad_norm": 0.3773350119590759,
      "learning_rate": 1.5016667512733305e-05,
      "loss": 0.07,
      "step": 14725
    },
    {
      "epoch": 0.747753693080867,
      "grad_norm": 0.4341411292552948,
      "learning_rate": 1.5014975379460888e-05,
      "loss": 0.0742,
      "step": 14730
    },
    {
      "epoch": 0.7480075130717295,
      "grad_norm": 0.5234106183052063,
      "learning_rate": 1.501328324618847e-05,
      "loss": 0.0716,
      "step": 14735
    },
    {
      "epoch": 0.748261333062592,
      "grad_norm": 0.7145912051200867,
      "learning_rate": 1.5011591112916055e-05,
      "loss": 0.0841,
      "step": 14740
    },
    {
      "epoch": 0.7485151530534545,
      "grad_norm": 0.42830368876457214,
      "learning_rate": 1.5009898979643637e-05,
      "loss": 0.075,
      "step": 14745
    },
    {
      "epoch": 0.748768973044317,
      "grad_norm": 0.4660949409008026,
      "learning_rate": 1.5008206846371221e-05,
      "loss": 0.0733,
      "step": 14750
    },
    {
      "epoch": 0.7490227930351795,
      "grad_norm": 0.34171634912490845,
      "learning_rate": 1.5006514713098806e-05,
      "loss": 0.0887,
      "step": 14755
    },
    {
      "epoch": 0.749276613026042,
      "grad_norm": 0.8460459113121033,
      "learning_rate": 1.5004822579826388e-05,
      "loss": 0.0861,
      "step": 14760
    },
    {
      "epoch": 0.7495304330169044,
      "grad_norm": 0.5516984462738037,
      "learning_rate": 1.5003130446553973e-05,
      "loss": 0.081,
      "step": 14765
    },
    {
      "epoch": 0.7497842530077669,
      "grad_norm": 0.4025214910507202,
      "learning_rate": 1.5001438313281555e-05,
      "loss": 0.0787,
      "step": 14770
    },
    {
      "epoch": 0.7500380729986293,
      "grad_norm": 0.44463062286376953,
      "learning_rate": 1.4999746180009139e-05,
      "loss": 0.0833,
      "step": 14775
    },
    {
      "epoch": 0.7502918929894918,
      "grad_norm": 0.5166744589805603,
      "learning_rate": 1.4998054046736724e-05,
      "loss": 0.084,
      "step": 14780
    },
    {
      "epoch": 0.7505457129803543,
      "grad_norm": 0.5265631079673767,
      "learning_rate": 1.4996361913464306e-05,
      "loss": 0.0787,
      "step": 14785
    },
    {
      "epoch": 0.7507995329712168,
      "grad_norm": 0.45194345712661743,
      "learning_rate": 1.4994669780191888e-05,
      "loss": 0.0765,
      "step": 14790
    },
    {
      "epoch": 0.7510533529620793,
      "grad_norm": 0.4983544945716858,
      "learning_rate": 1.4992977646919473e-05,
      "loss": 0.0803,
      "step": 14795
    },
    {
      "epoch": 0.7513071729529418,
      "grad_norm": 0.4568130075931549,
      "learning_rate": 1.4991285513647056e-05,
      "loss": 0.0702,
      "step": 14800
    },
    {
      "epoch": 0.7515609929438043,
      "grad_norm": 0.2652099132537842,
      "learning_rate": 1.4989593380374638e-05,
      "loss": 0.0802,
      "step": 14805
    },
    {
      "epoch": 0.7518148129346668,
      "grad_norm": 0.40893828868865967,
      "learning_rate": 1.4987901247102223e-05,
      "loss": 0.0747,
      "step": 14810
    },
    {
      "epoch": 0.7520686329255292,
      "grad_norm": 0.8311263918876648,
      "learning_rate": 1.4986209113829805e-05,
      "loss": 0.0787,
      "step": 14815
    },
    {
      "epoch": 0.7523224529163917,
      "grad_norm": 0.3804740905761719,
      "learning_rate": 1.498451698055739e-05,
      "loss": 0.0737,
      "step": 14820
    },
    {
      "epoch": 0.7525762729072542,
      "grad_norm": 0.3137752115726471,
      "learning_rate": 1.4982824847284974e-05,
      "loss": 0.0749,
      "step": 14825
    },
    {
      "epoch": 0.7528300928981166,
      "grad_norm": 0.3906417191028595,
      "learning_rate": 1.4981132714012556e-05,
      "loss": 0.0833,
      "step": 14830
    },
    {
      "epoch": 0.7530839128889791,
      "grad_norm": 0.6462694406509399,
      "learning_rate": 1.4979440580740141e-05,
      "loss": 0.0784,
      "step": 14835
    },
    {
      "epoch": 0.7533377328798416,
      "grad_norm": 0.5136402249336243,
      "learning_rate": 1.4977748447467723e-05,
      "loss": 0.0754,
      "step": 14840
    },
    {
      "epoch": 0.7535915528707041,
      "grad_norm": 0.5540376305580139,
      "learning_rate": 1.4976056314195307e-05,
      "loss": 0.0796,
      "step": 14845
    },
    {
      "epoch": 0.7538453728615666,
      "grad_norm": 0.4849870800971985,
      "learning_rate": 1.4974364180922892e-05,
      "loss": 0.0757,
      "step": 14850
    },
    {
      "epoch": 0.7540991928524291,
      "grad_norm": 0.739065408706665,
      "learning_rate": 1.4972672047650474e-05,
      "loss": 0.0847,
      "step": 14855
    },
    {
      "epoch": 0.7543530128432916,
      "grad_norm": 0.5149842500686646,
      "learning_rate": 1.4970979914378057e-05,
      "loss": 0.0752,
      "step": 14860
    },
    {
      "epoch": 0.754606832834154,
      "grad_norm": 0.42165669798851013,
      "learning_rate": 1.496928778110564e-05,
      "loss": 0.0831,
      "step": 14865
    },
    {
      "epoch": 0.7548606528250165,
      "grad_norm": 0.49979931116104126,
      "learning_rate": 1.4967595647833224e-05,
      "loss": 0.0733,
      "step": 14870
    },
    {
      "epoch": 0.755114472815879,
      "grad_norm": 0.49712690711021423,
      "learning_rate": 1.496590351456081e-05,
      "loss": 0.0803,
      "step": 14875
    },
    {
      "epoch": 0.7553682928067414,
      "grad_norm": 0.4191552400588989,
      "learning_rate": 1.4964211381288391e-05,
      "loss": 0.0834,
      "step": 14880
    },
    {
      "epoch": 0.7556221127976039,
      "grad_norm": 0.3019554316997528,
      "learning_rate": 1.4962519248015975e-05,
      "loss": 0.0778,
      "step": 14885
    },
    {
      "epoch": 0.7558759327884664,
      "grad_norm": 0.5018486380577087,
      "learning_rate": 1.4960827114743558e-05,
      "loss": 0.0766,
      "step": 14890
    },
    {
      "epoch": 0.7561297527793289,
      "grad_norm": 0.33077868819236755,
      "learning_rate": 1.4959134981471142e-05,
      "loss": 0.0752,
      "step": 14895
    },
    {
      "epoch": 0.7563835727701914,
      "grad_norm": 0.54078209400177,
      "learning_rate": 1.4957442848198724e-05,
      "loss": 0.0796,
      "step": 14900
    },
    {
      "epoch": 0.7566373927610539,
      "grad_norm": 0.4296625554561615,
      "learning_rate": 1.4955750714926309e-05,
      "loss": 0.0742,
      "step": 14905
    },
    {
      "epoch": 0.7568912127519163,
      "grad_norm": 0.33885428309440613,
      "learning_rate": 1.4954058581653893e-05,
      "loss": 0.0722,
      "step": 14910
    },
    {
      "epoch": 0.7571450327427788,
      "grad_norm": 0.4666841924190521,
      "learning_rate": 1.4952366448381476e-05,
      "loss": 0.0831,
      "step": 14915
    },
    {
      "epoch": 0.7573988527336413,
      "grad_norm": 0.45907166600227356,
      "learning_rate": 1.495067431510906e-05,
      "loss": 0.0714,
      "step": 14920
    },
    {
      "epoch": 0.7576526727245038,
      "grad_norm": 0.32929760217666626,
      "learning_rate": 1.4948982181836642e-05,
      "loss": 0.0708,
      "step": 14925
    },
    {
      "epoch": 0.7579064927153663,
      "grad_norm": 0.4375610649585724,
      "learning_rate": 1.4947290048564227e-05,
      "loss": 0.0753,
      "step": 14930
    },
    {
      "epoch": 0.7581603127062287,
      "grad_norm": 0.32175201177597046,
      "learning_rate": 1.494559791529181e-05,
      "loss": 0.0816,
      "step": 14935
    },
    {
      "epoch": 0.7584141326970912,
      "grad_norm": 0.31863266229629517,
      "learning_rate": 1.4943905782019392e-05,
      "loss": 0.0768,
      "step": 14940
    },
    {
      "epoch": 0.7586679526879537,
      "grad_norm": 0.5156002640724182,
      "learning_rate": 1.4942213648746977e-05,
      "loss": 0.0784,
      "step": 14945
    },
    {
      "epoch": 0.7589217726788162,
      "grad_norm": 0.3342384696006775,
      "learning_rate": 1.494052151547456e-05,
      "loss": 0.0766,
      "step": 14950
    },
    {
      "epoch": 0.7591755926696787,
      "grad_norm": 0.3848695456981659,
      "learning_rate": 1.4938829382202143e-05,
      "loss": 0.0882,
      "step": 14955
    },
    {
      "epoch": 0.7594294126605411,
      "grad_norm": 0.32741138339042664,
      "learning_rate": 1.4937137248929728e-05,
      "loss": 0.0804,
      "step": 14960
    },
    {
      "epoch": 0.7596832326514036,
      "grad_norm": 0.4225481152534485,
      "learning_rate": 1.493544511565731e-05,
      "loss": 0.0775,
      "step": 14965
    },
    {
      "epoch": 0.7599370526422661,
      "grad_norm": 0.5659987926483154,
      "learning_rate": 1.4933752982384895e-05,
      "loss": 0.09,
      "step": 14970
    },
    {
      "epoch": 0.7601908726331286,
      "grad_norm": 0.46714866161346436,
      "learning_rate": 1.4932060849112477e-05,
      "loss": 0.076,
      "step": 14975
    },
    {
      "epoch": 0.7604446926239911,
      "grad_norm": 0.3290693759918213,
      "learning_rate": 1.493036871584006e-05,
      "loss": 0.0818,
      "step": 14980
    },
    {
      "epoch": 0.7606985126148536,
      "grad_norm": 0.43824639916419983,
      "learning_rate": 1.4928676582567646e-05,
      "loss": 0.0692,
      "step": 14985
    },
    {
      "epoch": 0.760952332605716,
      "grad_norm": 0.34384244680404663,
      "learning_rate": 1.4926984449295228e-05,
      "loss": 0.0865,
      "step": 14990
    },
    {
      "epoch": 0.7612061525965785,
      "grad_norm": 0.45859983563423157,
      "learning_rate": 1.492529231602281e-05,
      "loss": 0.0699,
      "step": 14995
    },
    {
      "epoch": 0.761459972587441,
      "grad_norm": 0.4020313024520874,
      "learning_rate": 1.4923600182750395e-05,
      "loss": 0.079,
      "step": 15000
    },
    {
      "epoch": 0.7617137925783035,
      "grad_norm": 0.41197675466537476,
      "learning_rate": 1.4921908049477978e-05,
      "loss": 0.0748,
      "step": 15005
    },
    {
      "epoch": 0.7619676125691659,
      "grad_norm": 0.5557354092597961,
      "learning_rate": 1.4920215916205563e-05,
      "loss": 0.084,
      "step": 15010
    },
    {
      "epoch": 0.7622214325600284,
      "grad_norm": 0.39988163113594055,
      "learning_rate": 1.4918523782933145e-05,
      "loss": 0.0835,
      "step": 15015
    },
    {
      "epoch": 0.7624752525508909,
      "grad_norm": 0.3414739966392517,
      "learning_rate": 1.4916831649660727e-05,
      "loss": 0.0888,
      "step": 15020
    },
    {
      "epoch": 0.7627290725417534,
      "grad_norm": 0.37848934531211853,
      "learning_rate": 1.4915139516388312e-05,
      "loss": 0.0784,
      "step": 15025
    },
    {
      "epoch": 0.7629828925326159,
      "grad_norm": 0.4667239189147949,
      "learning_rate": 1.4913447383115896e-05,
      "loss": 0.08,
      "step": 15030
    },
    {
      "epoch": 0.7632367125234784,
      "grad_norm": 0.42531412839889526,
      "learning_rate": 1.4911755249843478e-05,
      "loss": 0.0815,
      "step": 15035
    },
    {
      "epoch": 0.7634905325143408,
      "grad_norm": 0.4677846133708954,
      "learning_rate": 1.4910063116571063e-05,
      "loss": 0.0779,
      "step": 15040
    },
    {
      "epoch": 0.7637443525052033,
      "grad_norm": 0.3771125078201294,
      "learning_rate": 1.4908370983298645e-05,
      "loss": 0.0764,
      "step": 15045
    },
    {
      "epoch": 0.7639981724960658,
      "grad_norm": 0.4332607388496399,
      "learning_rate": 1.4906678850026228e-05,
      "loss": 0.0751,
      "step": 15050
    },
    {
      "epoch": 0.7642519924869283,
      "grad_norm": 0.47496718168258667,
      "learning_rate": 1.4904986716753814e-05,
      "loss": 0.0791,
      "step": 15055
    },
    {
      "epoch": 0.7645058124777907,
      "grad_norm": 0.3843788504600525,
      "learning_rate": 1.4903294583481396e-05,
      "loss": 0.0725,
      "step": 15060
    },
    {
      "epoch": 0.7647596324686532,
      "grad_norm": 0.44778764247894287,
      "learning_rate": 1.490160245020898e-05,
      "loss": 0.0842,
      "step": 15065
    },
    {
      "epoch": 0.7650134524595157,
      "grad_norm": 0.4595566391944885,
      "learning_rate": 1.4899910316936563e-05,
      "loss": 0.086,
      "step": 15070
    },
    {
      "epoch": 0.7652672724503782,
      "grad_norm": 0.4201864004135132,
      "learning_rate": 1.4898218183664146e-05,
      "loss": 0.0818,
      "step": 15075
    },
    {
      "epoch": 0.7655210924412407,
      "grad_norm": 0.35480204224586487,
      "learning_rate": 1.4896526050391731e-05,
      "loss": 0.0768,
      "step": 15080
    },
    {
      "epoch": 0.7657749124321032,
      "grad_norm": 0.34363895654678345,
      "learning_rate": 1.4894833917119313e-05,
      "loss": 0.0735,
      "step": 15085
    },
    {
      "epoch": 0.7660287324229657,
      "grad_norm": 0.39018622040748596,
      "learning_rate": 1.4893141783846897e-05,
      "loss": 0.0808,
      "step": 15090
    },
    {
      "epoch": 0.7662825524138281,
      "grad_norm": 0.40973344445228577,
      "learning_rate": 1.489144965057448e-05,
      "loss": 0.0845,
      "step": 15095
    },
    {
      "epoch": 0.7665363724046906,
      "grad_norm": 1.6443856954574585,
      "learning_rate": 1.4889757517302064e-05,
      "loss": 0.0759,
      "step": 15100
    },
    {
      "epoch": 0.7667901923955531,
      "grad_norm": 0.5674831867218018,
      "learning_rate": 1.4888065384029646e-05,
      "loss": 0.0916,
      "step": 15105
    },
    {
      "epoch": 0.7670440123864155,
      "grad_norm": 0.6491461992263794,
      "learning_rate": 1.4886373250757231e-05,
      "loss": 0.0727,
      "step": 15110
    },
    {
      "epoch": 0.767297832377278,
      "grad_norm": 0.43828409910202026,
      "learning_rate": 1.4884681117484815e-05,
      "loss": 0.069,
      "step": 15115
    },
    {
      "epoch": 0.7675516523681405,
      "grad_norm": 0.3990132212638855,
      "learning_rate": 1.4882988984212398e-05,
      "loss": 0.0752,
      "step": 15120
    },
    {
      "epoch": 0.767805472359003,
      "grad_norm": 0.6415324211120605,
      "learning_rate": 1.4881296850939982e-05,
      "loss": 0.0887,
      "step": 15125
    },
    {
      "epoch": 0.7680592923498655,
      "grad_norm": 0.415190190076828,
      "learning_rate": 1.4879604717667563e-05,
      "loss": 0.0854,
      "step": 15130
    },
    {
      "epoch": 0.768313112340728,
      "grad_norm": 0.43115556240081787,
      "learning_rate": 1.4877912584395149e-05,
      "loss": 0.0841,
      "step": 15135
    },
    {
      "epoch": 0.7685669323315905,
      "grad_norm": 0.4042651653289795,
      "learning_rate": 1.4876220451122732e-05,
      "loss": 0.0774,
      "step": 15140
    },
    {
      "epoch": 0.768820752322453,
      "grad_norm": 0.4800015985965729,
      "learning_rate": 1.4874528317850314e-05,
      "loss": 0.0781,
      "step": 15145
    },
    {
      "epoch": 0.7690745723133154,
      "grad_norm": 0.40148481726646423,
      "learning_rate": 1.48728361845779e-05,
      "loss": 0.0808,
      "step": 15150
    },
    {
      "epoch": 0.7693283923041778,
      "grad_norm": 0.44380658864974976,
      "learning_rate": 1.4871144051305481e-05,
      "loss": 0.0862,
      "step": 15155
    },
    {
      "epoch": 0.7695822122950403,
      "grad_norm": 0.41752004623413086,
      "learning_rate": 1.4869451918033066e-05,
      "loss": 0.0728,
      "step": 15160
    },
    {
      "epoch": 0.7698360322859028,
      "grad_norm": 0.3876765966415405,
      "learning_rate": 1.486775978476065e-05,
      "loss": 0.0759,
      "step": 15165
    },
    {
      "epoch": 0.7700898522767653,
      "grad_norm": 0.3961372971534729,
      "learning_rate": 1.4866067651488232e-05,
      "loss": 0.0676,
      "step": 15170
    },
    {
      "epoch": 0.7703436722676278,
      "grad_norm": 0.5213141441345215,
      "learning_rate": 1.4864375518215817e-05,
      "loss": 0.085,
      "step": 15175
    },
    {
      "epoch": 0.7705974922584903,
      "grad_norm": 0.4002414643764496,
      "learning_rate": 1.4862683384943399e-05,
      "loss": 0.0855,
      "step": 15180
    },
    {
      "epoch": 0.7708513122493528,
      "grad_norm": 0.26624107360839844,
      "learning_rate": 1.4860991251670982e-05,
      "loss": 0.0856,
      "step": 15185
    },
    {
      "epoch": 0.7711051322402153,
      "grad_norm": 0.4855886697769165,
      "learning_rate": 1.4859299118398568e-05,
      "loss": 0.0772,
      "step": 15190
    },
    {
      "epoch": 0.7713589522310778,
      "grad_norm": 0.5074960589408875,
      "learning_rate": 1.485760698512615e-05,
      "loss": 0.0737,
      "step": 15195
    },
    {
      "epoch": 0.7716127722219402,
      "grad_norm": 0.5205410122871399,
      "learning_rate": 1.4855914851853731e-05,
      "loss": 0.077,
      "step": 15200
    },
    {
      "epoch": 0.7718665922128026,
      "grad_norm": 0.4280433654785156,
      "learning_rate": 1.4854222718581317e-05,
      "loss": 0.0697,
      "step": 15205
    },
    {
      "epoch": 0.7721204122036651,
      "grad_norm": 0.3168952465057373,
      "learning_rate": 1.48525305853089e-05,
      "loss": 0.0795,
      "step": 15210
    },
    {
      "epoch": 0.7723742321945276,
      "grad_norm": 0.39712512493133545,
      "learning_rate": 1.4850838452036484e-05,
      "loss": 0.0829,
      "step": 15215
    },
    {
      "epoch": 0.7726280521853901,
      "grad_norm": 0.5094771385192871,
      "learning_rate": 1.4849146318764067e-05,
      "loss": 0.0805,
      "step": 15220
    },
    {
      "epoch": 0.7728818721762526,
      "grad_norm": 0.2554795444011688,
      "learning_rate": 1.4847454185491649e-05,
      "loss": 0.0655,
      "step": 15225
    },
    {
      "epoch": 0.7731356921671151,
      "grad_norm": 0.46590444445610046,
      "learning_rate": 1.4845762052219234e-05,
      "loss": 0.0816,
      "step": 15230
    },
    {
      "epoch": 0.7733895121579776,
      "grad_norm": 0.5981921553611755,
      "learning_rate": 1.4844069918946818e-05,
      "loss": 0.0721,
      "step": 15235
    },
    {
      "epoch": 0.7736433321488401,
      "grad_norm": 0.35320785641670227,
      "learning_rate": 1.48423777856744e-05,
      "loss": 0.0768,
      "step": 15240
    },
    {
      "epoch": 0.7738971521397026,
      "grad_norm": 0.38590070605278015,
      "learning_rate": 1.4840685652401985e-05,
      "loss": 0.0823,
      "step": 15245
    },
    {
      "epoch": 0.774150972130565,
      "grad_norm": 0.4039193093776703,
      "learning_rate": 1.4838993519129567e-05,
      "loss": 0.0674,
      "step": 15250
    },
    {
      "epoch": 0.7744047921214274,
      "grad_norm": 0.423473060131073,
      "learning_rate": 1.4837301385857152e-05,
      "loss": 0.0734,
      "step": 15255
    },
    {
      "epoch": 0.7746586121122899,
      "grad_norm": 0.4255438446998596,
      "learning_rate": 1.4835609252584736e-05,
      "loss": 0.0739,
      "step": 15260
    },
    {
      "epoch": 0.7749124321031524,
      "grad_norm": 0.3805790841579437,
      "learning_rate": 1.4833917119312317e-05,
      "loss": 0.0755,
      "step": 15265
    },
    {
      "epoch": 0.7751662520940149,
      "grad_norm": 0.4251479506492615,
      "learning_rate": 1.4832224986039903e-05,
      "loss": 0.0817,
      "step": 15270
    },
    {
      "epoch": 0.7754200720848774,
      "grad_norm": 0.6027906537055969,
      "learning_rate": 1.4830532852767485e-05,
      "loss": 0.0794,
      "step": 15275
    },
    {
      "epoch": 0.7756738920757399,
      "grad_norm": 0.5641241669654846,
      "learning_rate": 1.4828840719495068e-05,
      "loss": 0.0707,
      "step": 15280
    },
    {
      "epoch": 0.7759277120666024,
      "grad_norm": 0.3357883393764496,
      "learning_rate": 1.4827148586222653e-05,
      "loss": 0.0798,
      "step": 15285
    },
    {
      "epoch": 0.7761815320574649,
      "grad_norm": 0.6147197484970093,
      "learning_rate": 1.4825456452950235e-05,
      "loss": 0.0761,
      "step": 15290
    },
    {
      "epoch": 0.7764353520483274,
      "grad_norm": 0.30087563395500183,
      "learning_rate": 1.4823764319677819e-05,
      "loss": 0.0663,
      "step": 15295
    },
    {
      "epoch": 0.7766891720391899,
      "grad_norm": 0.3757729232311249,
      "learning_rate": 1.4822072186405402e-05,
      "loss": 0.0746,
      "step": 15300
    },
    {
      "epoch": 0.7769429920300522,
      "grad_norm": 0.38213640451431274,
      "learning_rate": 1.4820380053132986e-05,
      "loss": 0.0839,
      "step": 15305
    },
    {
      "epoch": 0.7771968120209147,
      "grad_norm": 0.6010876297950745,
      "learning_rate": 1.4818687919860571e-05,
      "loss": 0.0756,
      "step": 15310
    },
    {
      "epoch": 0.7774506320117772,
      "grad_norm": 0.39273327589035034,
      "learning_rate": 1.4816995786588153e-05,
      "loss": 0.0731,
      "step": 15315
    },
    {
      "epoch": 0.7777044520026397,
      "grad_norm": 0.5562280416488647,
      "learning_rate": 1.4815303653315736e-05,
      "loss": 0.0676,
      "step": 15320
    },
    {
      "epoch": 0.7779582719935022,
      "grad_norm": 0.37354519963264465,
      "learning_rate": 1.481361152004332e-05,
      "loss": 0.0792,
      "step": 15325
    },
    {
      "epoch": 0.7782120919843647,
      "grad_norm": 0.4271200895309448,
      "learning_rate": 1.4811919386770904e-05,
      "loss": 0.0678,
      "step": 15330
    },
    {
      "epoch": 0.7784659119752272,
      "grad_norm": 0.4866403043270111,
      "learning_rate": 1.4810227253498485e-05,
      "loss": 0.0778,
      "step": 15335
    },
    {
      "epoch": 0.7787197319660897,
      "grad_norm": 0.561621367931366,
      "learning_rate": 1.480853512022607e-05,
      "loss": 0.0753,
      "step": 15340
    },
    {
      "epoch": 0.7789735519569522,
      "grad_norm": 0.38731345534324646,
      "learning_rate": 1.4806842986953654e-05,
      "loss": 0.0828,
      "step": 15345
    },
    {
      "epoch": 0.7792273719478147,
      "grad_norm": 0.3867878019809723,
      "learning_rate": 1.4805150853681236e-05,
      "loss": 0.0858,
      "step": 15350
    },
    {
      "epoch": 0.779481191938677,
      "grad_norm": 0.43254461884498596,
      "learning_rate": 1.4803458720408821e-05,
      "loss": 0.0646,
      "step": 15355
    },
    {
      "epoch": 0.7797350119295395,
      "grad_norm": 0.35966265201568604,
      "learning_rate": 1.4801766587136403e-05,
      "loss": 0.0726,
      "step": 15360
    },
    {
      "epoch": 0.779988831920402,
      "grad_norm": 0.36605480313301086,
      "learning_rate": 1.4800074453863988e-05,
      "loss": 0.0721,
      "step": 15365
    },
    {
      "epoch": 0.7802426519112645,
      "grad_norm": 0.37214845418930054,
      "learning_rate": 1.4798382320591572e-05,
      "loss": 0.0836,
      "step": 15370
    },
    {
      "epoch": 0.780496471902127,
      "grad_norm": 0.43033382296562195,
      "learning_rate": 1.4796690187319154e-05,
      "loss": 0.0707,
      "step": 15375
    },
    {
      "epoch": 0.7807502918929895,
      "grad_norm": 0.41933897137641907,
      "learning_rate": 1.4794998054046739e-05,
      "loss": 0.074,
      "step": 15380
    },
    {
      "epoch": 0.781004111883852,
      "grad_norm": 0.4193064570426941,
      "learning_rate": 1.4793305920774321e-05,
      "loss": 0.0732,
      "step": 15385
    },
    {
      "epoch": 0.7812579318747145,
      "grad_norm": 0.35709092020988464,
      "learning_rate": 1.4791613787501904e-05,
      "loss": 0.0831,
      "step": 15390
    },
    {
      "epoch": 0.781511751865577,
      "grad_norm": 0.38249507546424866,
      "learning_rate": 1.4789921654229488e-05,
      "loss": 0.0753,
      "step": 15395
    },
    {
      "epoch": 0.7817655718564395,
      "grad_norm": 0.6960234045982361,
      "learning_rate": 1.4788229520957071e-05,
      "loss": 0.0752,
      "step": 15400
    },
    {
      "epoch": 0.7820193918473018,
      "grad_norm": 0.4672459065914154,
      "learning_rate": 1.4786537387684657e-05,
      "loss": 0.0851,
      "step": 15405
    },
    {
      "epoch": 0.7822732118381643,
      "grad_norm": 0.3816010057926178,
      "learning_rate": 1.4784845254412239e-05,
      "loss": 0.0764,
      "step": 15410
    },
    {
      "epoch": 0.7825270318290268,
      "grad_norm": 0.7014495730400085,
      "learning_rate": 1.4783153121139822e-05,
      "loss": 0.0694,
      "step": 15415
    },
    {
      "epoch": 0.7827808518198893,
      "grad_norm": 0.327251136302948,
      "learning_rate": 1.4781460987867406e-05,
      "loss": 0.0716,
      "step": 15420
    },
    {
      "epoch": 0.7830346718107518,
      "grad_norm": 0.5149769186973572,
      "learning_rate": 1.477976885459499e-05,
      "loss": 0.0734,
      "step": 15425
    },
    {
      "epoch": 0.7832884918016143,
      "grad_norm": 0.40763530135154724,
      "learning_rate": 1.4778076721322571e-05,
      "loss": 0.0812,
      "step": 15430
    },
    {
      "epoch": 0.7835423117924768,
      "grad_norm": 0.32969698309898376,
      "learning_rate": 1.4776384588050156e-05,
      "loss": 0.0799,
      "step": 15435
    },
    {
      "epoch": 0.7837961317833393,
      "grad_norm": 0.3335864245891571,
      "learning_rate": 1.477469245477774e-05,
      "loss": 0.0785,
      "step": 15440
    },
    {
      "epoch": 0.7840499517742018,
      "grad_norm": 0.48730531334877014,
      "learning_rate": 1.4773000321505322e-05,
      "loss": 0.0849,
      "step": 15445
    },
    {
      "epoch": 0.7843037717650642,
      "grad_norm": 0.7397474646568298,
      "learning_rate": 1.4771308188232907e-05,
      "loss": 0.0778,
      "step": 15450
    },
    {
      "epoch": 0.7845575917559267,
      "grad_norm": 0.39907628297805786,
      "learning_rate": 1.4769616054960489e-05,
      "loss": 0.0729,
      "step": 15455
    },
    {
      "epoch": 0.7848114117467891,
      "grad_norm": 0.47134730219841003,
      "learning_rate": 1.4767923921688074e-05,
      "loss": 0.0771,
      "step": 15460
    },
    {
      "epoch": 0.7850652317376516,
      "grad_norm": 0.41393592953681946,
      "learning_rate": 1.4766231788415658e-05,
      "loss": 0.0772,
      "step": 15465
    },
    {
      "epoch": 0.7853190517285141,
      "grad_norm": 0.3614552319049835,
      "learning_rate": 1.476453965514324e-05,
      "loss": 0.0718,
      "step": 15470
    },
    {
      "epoch": 0.7855728717193766,
      "grad_norm": 0.4395851194858551,
      "learning_rate": 1.4762847521870825e-05,
      "loss": 0.0825,
      "step": 15475
    },
    {
      "epoch": 0.7858266917102391,
      "grad_norm": 0.3465043902397156,
      "learning_rate": 1.4761155388598407e-05,
      "loss": 0.0746,
      "step": 15480
    },
    {
      "epoch": 0.7860805117011016,
      "grad_norm": 0.3366214632987976,
      "learning_rate": 1.475946325532599e-05,
      "loss": 0.071,
      "step": 15485
    },
    {
      "epoch": 0.7863343316919641,
      "grad_norm": 0.6985540986061096,
      "learning_rate": 1.4757771122053575e-05,
      "loss": 0.0701,
      "step": 15490
    },
    {
      "epoch": 0.7865881516828266,
      "grad_norm": 0.6810505390167236,
      "learning_rate": 1.4756078988781157e-05,
      "loss": 0.0881,
      "step": 15495
    },
    {
      "epoch": 0.786841971673689,
      "grad_norm": 0.8365420699119568,
      "learning_rate": 1.4754386855508742e-05,
      "loss": 0.0842,
      "step": 15500
    },
    {
      "epoch": 0.7870957916645515,
      "grad_norm": 0.4822964370250702,
      "learning_rate": 1.4752694722236324e-05,
      "loss": 0.07,
      "step": 15505
    },
    {
      "epoch": 0.787349611655414,
      "grad_norm": 0.3959048390388489,
      "learning_rate": 1.4751002588963908e-05,
      "loss": 0.078,
      "step": 15510
    },
    {
      "epoch": 0.7876034316462764,
      "grad_norm": 0.32304689288139343,
      "learning_rate": 1.4749310455691493e-05,
      "loss": 0.0814,
      "step": 15515
    },
    {
      "epoch": 0.7878572516371389,
      "grad_norm": 0.39487403631210327,
      "learning_rate": 1.4747618322419075e-05,
      "loss": 0.0779,
      "step": 15520
    },
    {
      "epoch": 0.7881110716280014,
      "grad_norm": 0.48532378673553467,
      "learning_rate": 1.4745926189146658e-05,
      "loss": 0.077,
      "step": 15525
    },
    {
      "epoch": 0.7883648916188639,
      "grad_norm": 0.38463422656059265,
      "learning_rate": 1.4744234055874242e-05,
      "loss": 0.0725,
      "step": 15530
    },
    {
      "epoch": 0.7886187116097264,
      "grad_norm": 0.5803540945053101,
      "learning_rate": 1.4742541922601826e-05,
      "loss": 0.0746,
      "step": 15535
    },
    {
      "epoch": 0.7888725316005889,
      "grad_norm": 0.37139931321144104,
      "learning_rate": 1.4740849789329407e-05,
      "loss": 0.0752,
      "step": 15540
    },
    {
      "epoch": 0.7891263515914514,
      "grad_norm": 0.39508089423179626,
      "learning_rate": 1.4739157656056993e-05,
      "loss": 0.0804,
      "step": 15545
    },
    {
      "epoch": 0.7893801715823138,
      "grad_norm": 0.4758867919445038,
      "learning_rate": 1.4737465522784576e-05,
      "loss": 0.0706,
      "step": 15550
    },
    {
      "epoch": 0.7896339915731763,
      "grad_norm": 0.3260742425918579,
      "learning_rate": 1.473577338951216e-05,
      "loss": 0.0762,
      "step": 15555
    },
    {
      "epoch": 0.7898878115640388,
      "grad_norm": 0.37266799807548523,
      "learning_rate": 1.4734081256239743e-05,
      "loss": 0.0875,
      "step": 15560
    },
    {
      "epoch": 0.7901416315549012,
      "grad_norm": 0.28276917338371277,
      "learning_rate": 1.4732389122967325e-05,
      "loss": 0.0633,
      "step": 15565
    },
    {
      "epoch": 0.7903954515457637,
      "grad_norm": 0.4956200122833252,
      "learning_rate": 1.473069698969491e-05,
      "loss": 0.0704,
      "step": 15570
    },
    {
      "epoch": 0.7906492715366262,
      "grad_norm": 0.5198813676834106,
      "learning_rate": 1.4729004856422492e-05,
      "loss": 0.0722,
      "step": 15575
    },
    {
      "epoch": 0.7909030915274887,
      "grad_norm": 0.40935298800468445,
      "learning_rate": 1.4727312723150076e-05,
      "loss": 0.0756,
      "step": 15580
    },
    {
      "epoch": 0.7911569115183512,
      "grad_norm": 0.3992593586444855,
      "learning_rate": 1.4725620589877661e-05,
      "loss": 0.0722,
      "step": 15585
    },
    {
      "epoch": 0.7914107315092137,
      "grad_norm": 0.3877473473548889,
      "learning_rate": 1.4723928456605243e-05,
      "loss": 0.0882,
      "step": 15590
    },
    {
      "epoch": 0.7916645515000762,
      "grad_norm": 0.3663567900657654,
      "learning_rate": 1.4722236323332826e-05,
      "loss": 0.0718,
      "step": 15595
    },
    {
      "epoch": 0.7919183714909386,
      "grad_norm": 0.37373846769332886,
      "learning_rate": 1.472054419006041e-05,
      "loss": 0.0821,
      "step": 15600
    },
    {
      "epoch": 0.7921721914818011,
      "grad_norm": 0.45595207810401917,
      "learning_rate": 1.4718852056787993e-05,
      "loss": 0.0716,
      "step": 15605
    },
    {
      "epoch": 0.7924260114726636,
      "grad_norm": 0.5352175831794739,
      "learning_rate": 1.4717159923515579e-05,
      "loss": 0.0715,
      "step": 15610
    },
    {
      "epoch": 0.792679831463526,
      "grad_norm": 0.3262074589729309,
      "learning_rate": 1.471546779024316e-05,
      "loss": 0.0737,
      "step": 15615
    },
    {
      "epoch": 0.7929336514543885,
      "grad_norm": 0.36793214082717896,
      "learning_rate": 1.4713775656970744e-05,
      "loss": 0.0679,
      "step": 15620
    },
    {
      "epoch": 0.793187471445251,
      "grad_norm": 0.41612109541893005,
      "learning_rate": 1.4712083523698328e-05,
      "loss": 0.0823,
      "step": 15625
    },
    {
      "epoch": 0.7934412914361135,
      "grad_norm": 0.40798622369766235,
      "learning_rate": 1.4710391390425911e-05,
      "loss": 0.0725,
      "step": 15630
    },
    {
      "epoch": 0.793695111426976,
      "grad_norm": 0.5767354369163513,
      "learning_rate": 1.4708699257153493e-05,
      "loss": 0.0813,
      "step": 15635
    },
    {
      "epoch": 0.7939489314178385,
      "grad_norm": 0.615403413772583,
      "learning_rate": 1.4707007123881078e-05,
      "loss": 0.0857,
      "step": 15640
    },
    {
      "epoch": 0.794202751408701,
      "grad_norm": 0.34347468614578247,
      "learning_rate": 1.4705314990608662e-05,
      "loss": 0.0702,
      "step": 15645
    },
    {
      "epoch": 0.7944565713995634,
      "grad_norm": 0.3589203655719757,
      "learning_rate": 1.4703622857336245e-05,
      "loss": 0.0724,
      "step": 15650
    },
    {
      "epoch": 0.7947103913904259,
      "grad_norm": 0.7738829851150513,
      "learning_rate": 1.4701930724063829e-05,
      "loss": 0.0768,
      "step": 15655
    },
    {
      "epoch": 0.7949642113812884,
      "grad_norm": 0.3893224000930786,
      "learning_rate": 1.470023859079141e-05,
      "loss": 0.0675,
      "step": 15660
    },
    {
      "epoch": 0.7952180313721509,
      "grad_norm": 0.43465477228164673,
      "learning_rate": 1.4698546457518996e-05,
      "loss": 0.0744,
      "step": 15665
    },
    {
      "epoch": 0.7954718513630133,
      "grad_norm": 0.3958378732204437,
      "learning_rate": 1.469685432424658e-05,
      "loss": 0.0753,
      "step": 15670
    },
    {
      "epoch": 0.7957256713538758,
      "grad_norm": 0.5245358347892761,
      "learning_rate": 1.4695162190974161e-05,
      "loss": 0.0748,
      "step": 15675
    },
    {
      "epoch": 0.7959794913447383,
      "grad_norm": 0.3974064290523529,
      "learning_rate": 1.4693470057701747e-05,
      "loss": 0.0786,
      "step": 15680
    },
    {
      "epoch": 0.7962333113356008,
      "grad_norm": 0.4498586356639862,
      "learning_rate": 1.4691777924429328e-05,
      "loss": 0.0844,
      "step": 15685
    },
    {
      "epoch": 0.7964871313264633,
      "grad_norm": 0.4059452712535858,
      "learning_rate": 1.4690085791156912e-05,
      "loss": 0.0637,
      "step": 15690
    },
    {
      "epoch": 0.7967409513173258,
      "grad_norm": 0.5134612321853638,
      "learning_rate": 1.4688393657884497e-05,
      "loss": 0.0767,
      "step": 15695
    },
    {
      "epoch": 0.7969947713081882,
      "grad_norm": 0.4138130843639374,
      "learning_rate": 1.4686701524612079e-05,
      "loss": 0.0845,
      "step": 15700
    },
    {
      "epoch": 0.7972485912990507,
      "grad_norm": 0.37701699137687683,
      "learning_rate": 1.4685009391339664e-05,
      "loss": 0.0743,
      "step": 15705
    },
    {
      "epoch": 0.7975024112899132,
      "grad_norm": 0.3598290681838989,
      "learning_rate": 1.4683317258067246e-05,
      "loss": 0.072,
      "step": 15710
    },
    {
      "epoch": 0.7977562312807757,
      "grad_norm": 0.548689067363739,
      "learning_rate": 1.468162512479483e-05,
      "loss": 0.0662,
      "step": 15715
    },
    {
      "epoch": 0.7980100512716382,
      "grad_norm": 0.3460603654384613,
      "learning_rate": 1.4679932991522415e-05,
      "loss": 0.0595,
      "step": 15720
    },
    {
      "epoch": 0.7982638712625006,
      "grad_norm": 0.5318562388420105,
      "learning_rate": 1.4678240858249997e-05,
      "loss": 0.0726,
      "step": 15725
    },
    {
      "epoch": 0.7985176912533631,
      "grad_norm": 1.7499229907989502,
      "learning_rate": 1.467654872497758e-05,
      "loss": 0.0727,
      "step": 15730
    },
    {
      "epoch": 0.7987715112442256,
      "grad_norm": 0.6805680990219116,
      "learning_rate": 1.4674856591705164e-05,
      "loss": 0.0822,
      "step": 15735
    },
    {
      "epoch": 0.7990253312350881,
      "grad_norm": 0.6451191306114197,
      "learning_rate": 1.4673164458432747e-05,
      "loss": 0.0772,
      "step": 15740
    },
    {
      "epoch": 0.7992791512259505,
      "grad_norm": 0.5430915951728821,
      "learning_rate": 1.4671472325160333e-05,
      "loss": 0.0869,
      "step": 15745
    },
    {
      "epoch": 0.799532971216813,
      "grad_norm": 0.4242735803127289,
      "learning_rate": 1.4669780191887915e-05,
      "loss": 0.0781,
      "step": 15750
    },
    {
      "epoch": 0.7997867912076755,
      "grad_norm": 0.4273703098297119,
      "learning_rate": 1.4668088058615498e-05,
      "loss": 0.0732,
      "step": 15755
    },
    {
      "epoch": 0.800040611198538,
      "grad_norm": 0.4131914973258972,
      "learning_rate": 1.4666395925343082e-05,
      "loss": 0.0792,
      "step": 15760
    },
    {
      "epoch": 0.8002944311894005,
      "grad_norm": 0.7269368171691895,
      "learning_rate": 1.4664703792070665e-05,
      "loss": 0.0706,
      "step": 15765
    },
    {
      "epoch": 0.800548251180263,
      "grad_norm": 0.730088472366333,
      "learning_rate": 1.4663011658798247e-05,
      "loss": 0.0803,
      "step": 15770
    },
    {
      "epoch": 0.8008020711711255,
      "grad_norm": 0.4482845962047577,
      "learning_rate": 1.4661319525525832e-05,
      "loss": 0.0718,
      "step": 15775
    },
    {
      "epoch": 0.8010558911619879,
      "grad_norm": 0.4157123863697052,
      "learning_rate": 1.4659627392253414e-05,
      "loss": 0.0782,
      "step": 15780
    },
    {
      "epoch": 0.8013097111528504,
      "grad_norm": 0.3921131491661072,
      "learning_rate": 1.4657935258980998e-05,
      "loss": 0.066,
      "step": 15785
    },
    {
      "epoch": 0.8015635311437129,
      "grad_norm": 0.33425718545913696,
      "learning_rate": 1.4656243125708583e-05,
      "loss": 0.0675,
      "step": 15790
    },
    {
      "epoch": 0.8018173511345753,
      "grad_norm": 0.5760083794593811,
      "learning_rate": 1.4654550992436165e-05,
      "loss": 0.0846,
      "step": 15795
    },
    {
      "epoch": 0.8020711711254378,
      "grad_norm": 0.5779892206192017,
      "learning_rate": 1.465285885916375e-05,
      "loss": 0.0821,
      "step": 15800
    },
    {
      "epoch": 0.8023249911163003,
      "grad_norm": 0.3755846917629242,
      "learning_rate": 1.4651166725891332e-05,
      "loss": 0.0715,
      "step": 15805
    },
    {
      "epoch": 0.8025788111071628,
      "grad_norm": 0.3848758637905121,
      "learning_rate": 1.4649474592618915e-05,
      "loss": 0.0857,
      "step": 15810
    },
    {
      "epoch": 0.8028326310980253,
      "grad_norm": 0.3276999890804291,
      "learning_rate": 1.46477824593465e-05,
      "loss": 0.0747,
      "step": 15815
    },
    {
      "epoch": 0.8030864510888878,
      "grad_norm": 0.33379969000816345,
      "learning_rate": 1.4646090326074082e-05,
      "loss": 0.0722,
      "step": 15820
    },
    {
      "epoch": 0.8033402710797503,
      "grad_norm": 0.37067776918411255,
      "learning_rate": 1.4644398192801666e-05,
      "loss": 0.0737,
      "step": 15825
    },
    {
      "epoch": 0.8035940910706127,
      "grad_norm": 0.29267624020576477,
      "learning_rate": 1.464270605952925e-05,
      "loss": 0.063,
      "step": 15830
    },
    {
      "epoch": 0.8038479110614752,
      "grad_norm": 0.3490564227104187,
      "learning_rate": 1.4641013926256833e-05,
      "loss": 0.0909,
      "step": 15835
    },
    {
      "epoch": 0.8041017310523377,
      "grad_norm": 0.4481734335422516,
      "learning_rate": 1.4639321792984415e-05,
      "loss": 0.0741,
      "step": 15840
    },
    {
      "epoch": 0.8043555510432001,
      "grad_norm": 0.3517626225948334,
      "learning_rate": 1.4637629659712e-05,
      "loss": 0.0731,
      "step": 15845
    },
    {
      "epoch": 0.8046093710340626,
      "grad_norm": 0.4463517963886261,
      "learning_rate": 1.4635937526439584e-05,
      "loss": 0.0801,
      "step": 15850
    },
    {
      "epoch": 0.8048631910249251,
      "grad_norm": 0.3373356759548187,
      "learning_rate": 1.4634245393167167e-05,
      "loss": 0.0672,
      "step": 15855
    },
    {
      "epoch": 0.8051170110157876,
      "grad_norm": 0.4135652482509613,
      "learning_rate": 1.463255325989475e-05,
      "loss": 0.0605,
      "step": 15860
    },
    {
      "epoch": 0.8053708310066501,
      "grad_norm": 0.4091756343841553,
      "learning_rate": 1.4630861126622333e-05,
      "loss": 0.0711,
      "step": 15865
    },
    {
      "epoch": 0.8056246509975126,
      "grad_norm": 0.4166070520877838,
      "learning_rate": 1.4629168993349918e-05,
      "loss": 0.0641,
      "step": 15870
    },
    {
      "epoch": 0.8058784709883751,
      "grad_norm": 0.37992623448371887,
      "learning_rate": 1.4627476860077501e-05,
      "loss": 0.0744,
      "step": 15875
    },
    {
      "epoch": 0.8061322909792376,
      "grad_norm": 0.39718085527420044,
      "learning_rate": 1.4625784726805083e-05,
      "loss": 0.0715,
      "step": 15880
    },
    {
      "epoch": 0.8063861109701,
      "grad_norm": 0.3552018404006958,
      "learning_rate": 1.4624092593532669e-05,
      "loss": 0.0718,
      "step": 15885
    },
    {
      "epoch": 0.8066399309609625,
      "grad_norm": 0.39325806498527527,
      "learning_rate": 1.462240046026025e-05,
      "loss": 0.0743,
      "step": 15890
    },
    {
      "epoch": 0.8068937509518249,
      "grad_norm": 0.3813045620918274,
      "learning_rate": 1.4620708326987836e-05,
      "loss": 0.0717,
      "step": 15895
    },
    {
      "epoch": 0.8071475709426874,
      "grad_norm": 0.5478102564811707,
      "learning_rate": 1.461901619371542e-05,
      "loss": 0.081,
      "step": 15900
    },
    {
      "epoch": 0.8074013909335499,
      "grad_norm": 0.7054280638694763,
      "learning_rate": 1.4617324060443001e-05,
      "loss": 0.0769,
      "step": 15905
    },
    {
      "epoch": 0.8076552109244124,
      "grad_norm": 0.5671230554580688,
      "learning_rate": 1.4615631927170586e-05,
      "loss": 0.068,
      "step": 15910
    },
    {
      "epoch": 0.8079090309152749,
      "grad_norm": 0.6359015703201294,
      "learning_rate": 1.4613939793898168e-05,
      "loss": 0.06,
      "step": 15915
    },
    {
      "epoch": 0.8081628509061374,
      "grad_norm": 0.4305456876754761,
      "learning_rate": 1.4612247660625752e-05,
      "loss": 0.065,
      "step": 15920
    },
    {
      "epoch": 0.8084166708969999,
      "grad_norm": 0.5645986199378967,
      "learning_rate": 1.4610555527353337e-05,
      "loss": 0.0746,
      "step": 15925
    },
    {
      "epoch": 0.8086704908878624,
      "grad_norm": 0.4272991716861725,
      "learning_rate": 1.4608863394080919e-05,
      "loss": 0.082,
      "step": 15930
    },
    {
      "epoch": 0.8089243108787249,
      "grad_norm": 0.4712943732738495,
      "learning_rate": 1.4607171260808502e-05,
      "loss": 0.0835,
      "step": 15935
    },
    {
      "epoch": 0.8091781308695873,
      "grad_norm": 0.36924442648887634,
      "learning_rate": 1.4605479127536086e-05,
      "loss": 0.0757,
      "step": 15940
    },
    {
      "epoch": 0.8094319508604497,
      "grad_norm": 0.9191230535507202,
      "learning_rate": 1.460378699426367e-05,
      "loss": 0.0721,
      "step": 15945
    },
    {
      "epoch": 0.8096857708513122,
      "grad_norm": 0.466630220413208,
      "learning_rate": 1.4602094860991255e-05,
      "loss": 0.0777,
      "step": 15950
    },
    {
      "epoch": 0.8099395908421747,
      "grad_norm": 0.6259939670562744,
      "learning_rate": 1.4600402727718836e-05,
      "loss": 0.0699,
      "step": 15955
    },
    {
      "epoch": 0.8101934108330372,
      "grad_norm": 0.31263235211372375,
      "learning_rate": 1.4598710594446418e-05,
      "loss": 0.0688,
      "step": 15960
    },
    {
      "epoch": 0.8104472308238997,
      "grad_norm": 0.32614171504974365,
      "learning_rate": 1.4597018461174004e-05,
      "loss": 0.0667,
      "step": 15965
    },
    {
      "epoch": 0.8107010508147622,
      "grad_norm": 0.551091730594635,
      "learning_rate": 1.4595326327901587e-05,
      "loss": 0.0614,
      "step": 15970
    },
    {
      "epoch": 0.8109548708056247,
      "grad_norm": 0.44654572010040283,
      "learning_rate": 1.4593634194629169e-05,
      "loss": 0.0857,
      "step": 15975
    },
    {
      "epoch": 0.8112086907964872,
      "grad_norm": 0.8355388641357422,
      "learning_rate": 1.4591942061356754e-05,
      "loss": 0.0589,
      "step": 15980
    },
    {
      "epoch": 0.8114625107873497,
      "grad_norm": 0.3554515242576599,
      "learning_rate": 1.4590249928084336e-05,
      "loss": 0.0672,
      "step": 15985
    },
    {
      "epoch": 0.8117163307782121,
      "grad_norm": 0.8215906023979187,
      "learning_rate": 1.458855779481192e-05,
      "loss": 0.0651,
      "step": 15990
    },
    {
      "epoch": 0.8119701507690745,
      "grad_norm": 0.5176529884338379,
      "learning_rate": 1.4586865661539505e-05,
      "loss": 0.0811,
      "step": 15995
    },
    {
      "epoch": 0.812223970759937,
      "grad_norm": 0.37184882164001465,
      "learning_rate": 1.4585173528267087e-05,
      "loss": 0.0811,
      "step": 16000
    },
    {
      "epoch": 0.8124777907507995,
      "grad_norm": 0.36488497257232666,
      "learning_rate": 1.4583481394994672e-05,
      "loss": 0.0694,
      "step": 16005
    },
    {
      "epoch": 0.812731610741662,
      "grad_norm": 0.29020920395851135,
      "learning_rate": 1.4581789261722254e-05,
      "loss": 0.0725,
      "step": 16010
    },
    {
      "epoch": 0.8129854307325245,
      "grad_norm": 0.41967329382896423,
      "learning_rate": 1.4580097128449837e-05,
      "loss": 0.0749,
      "step": 16015
    },
    {
      "epoch": 0.813239250723387,
      "grad_norm": 0.35643911361694336,
      "learning_rate": 1.4578404995177423e-05,
      "loss": 0.0728,
      "step": 16020
    },
    {
      "epoch": 0.8134930707142495,
      "grad_norm": 0.5488142967224121,
      "learning_rate": 1.4576712861905004e-05,
      "loss": 0.076,
      "step": 16025
    },
    {
      "epoch": 0.813746890705112,
      "grad_norm": 0.6067050695419312,
      "learning_rate": 1.4575020728632588e-05,
      "loss": 0.0723,
      "step": 16030
    },
    {
      "epoch": 0.8140007106959745,
      "grad_norm": 0.3196711838245392,
      "learning_rate": 1.4573328595360171e-05,
      "loss": 0.0752,
      "step": 16035
    },
    {
      "epoch": 0.8142545306868368,
      "grad_norm": 0.3165784478187561,
      "learning_rate": 1.4571636462087755e-05,
      "loss": 0.0608,
      "step": 16040
    },
    {
      "epoch": 0.8145083506776993,
      "grad_norm": 0.37109699845314026,
      "learning_rate": 1.456994432881534e-05,
      "loss": 0.077,
      "step": 16045
    },
    {
      "epoch": 0.8147621706685618,
      "grad_norm": 0.5756739377975464,
      "learning_rate": 1.4568252195542922e-05,
      "loss": 0.0746,
      "step": 16050
    },
    {
      "epoch": 0.8150159906594243,
      "grad_norm": 0.5761229395866394,
      "learning_rate": 1.4566560062270506e-05,
      "loss": 0.0769,
      "step": 16055
    },
    {
      "epoch": 0.8152698106502868,
      "grad_norm": 0.6240622401237488,
      "learning_rate": 1.456486792899809e-05,
      "loss": 0.0838,
      "step": 16060
    },
    {
      "epoch": 0.8155236306411493,
      "grad_norm": 0.40298014879226685,
      "learning_rate": 1.4563175795725673e-05,
      "loss": 0.0701,
      "step": 16065
    },
    {
      "epoch": 0.8157774506320118,
      "grad_norm": 0.4362527132034302,
      "learning_rate": 1.4561483662453255e-05,
      "loss": 0.0739,
      "step": 16070
    },
    {
      "epoch": 0.8160312706228743,
      "grad_norm": 0.37834423780441284,
      "learning_rate": 1.455979152918084e-05,
      "loss": 0.079,
      "step": 16075
    },
    {
      "epoch": 0.8162850906137368,
      "grad_norm": 0.3305414319038391,
      "learning_rate": 1.4558099395908423e-05,
      "loss": 0.0662,
      "step": 16080
    },
    {
      "epoch": 0.8165389106045993,
      "grad_norm": 0.45846548676490784,
      "learning_rate": 1.4556407262636005e-05,
      "loss": 0.0704,
      "step": 16085
    },
    {
      "epoch": 0.8167927305954616,
      "grad_norm": 0.42119476199150085,
      "learning_rate": 1.455471512936359e-05,
      "loss": 0.0734,
      "step": 16090
    },
    {
      "epoch": 0.8170465505863241,
      "grad_norm": 0.4292179346084595,
      "learning_rate": 1.4553022996091172e-05,
      "loss": 0.0665,
      "step": 16095
    },
    {
      "epoch": 0.8173003705771866,
      "grad_norm": 0.3152535855770111,
      "learning_rate": 1.4551330862818758e-05,
      "loss": 0.0719,
      "step": 16100
    },
    {
      "epoch": 0.8175541905680491,
      "grad_norm": 0.3996585011482239,
      "learning_rate": 1.4549638729546341e-05,
      "loss": 0.0616,
      "step": 16105
    },
    {
      "epoch": 0.8178080105589116,
      "grad_norm": 0.3468000292778015,
      "learning_rate": 1.4547946596273923e-05,
      "loss": 0.074,
      "step": 16110
    },
    {
      "epoch": 0.8180618305497741,
      "grad_norm": 0.4962630569934845,
      "learning_rate": 1.4546254463001508e-05,
      "loss": 0.0705,
      "step": 16115
    },
    {
      "epoch": 0.8183156505406366,
      "grad_norm": 0.48265331983566284,
      "learning_rate": 1.454456232972909e-05,
      "loss": 0.0718,
      "step": 16120
    },
    {
      "epoch": 0.8185694705314991,
      "grad_norm": 0.4469800293445587,
      "learning_rate": 1.4542870196456674e-05,
      "loss": 0.0932,
      "step": 16125
    },
    {
      "epoch": 0.8188232905223616,
      "grad_norm": 0.6275296211242676,
      "learning_rate": 1.4541178063184259e-05,
      "loss": 0.0755,
      "step": 16130
    },
    {
      "epoch": 0.8190771105132241,
      "grad_norm": 0.5437877178192139,
      "learning_rate": 1.453948592991184e-05,
      "loss": 0.0762,
      "step": 16135
    },
    {
      "epoch": 0.8193309305040865,
      "grad_norm": 0.3738441467285156,
      "learning_rate": 1.4537793796639426e-05,
      "loss": 0.0734,
      "step": 16140
    },
    {
      "epoch": 0.8195847504949489,
      "grad_norm": 0.41383683681488037,
      "learning_rate": 1.4536101663367008e-05,
      "loss": 0.0834,
      "step": 16145
    },
    {
      "epoch": 0.8198385704858114,
      "grad_norm": 0.49464190006256104,
      "learning_rate": 1.4534409530094591e-05,
      "loss": 0.0773,
      "step": 16150
    },
    {
      "epoch": 0.8200923904766739,
      "grad_norm": 0.5497142672538757,
      "learning_rate": 1.4532717396822177e-05,
      "loss": 0.0715,
      "step": 16155
    },
    {
      "epoch": 0.8203462104675364,
      "grad_norm": 1.2365076541900635,
      "learning_rate": 1.4531025263549758e-05,
      "loss": 0.0737,
      "step": 16160
    },
    {
      "epoch": 0.8206000304583989,
      "grad_norm": 0.47389277815818787,
      "learning_rate": 1.452933313027734e-05,
      "loss": 0.0762,
      "step": 16165
    },
    {
      "epoch": 0.8208538504492614,
      "grad_norm": 0.33445248007774353,
      "learning_rate": 1.4527640997004925e-05,
      "loss": 0.0774,
      "step": 16170
    },
    {
      "epoch": 0.8211076704401239,
      "grad_norm": 0.4635807275772095,
      "learning_rate": 1.4525948863732509e-05,
      "loss": 0.0739,
      "step": 16175
    },
    {
      "epoch": 0.8213614904309864,
      "grad_norm": 0.7265046834945679,
      "learning_rate": 1.4524256730460091e-05,
      "loss": 0.0809,
      "step": 16180
    },
    {
      "epoch": 0.8216153104218489,
      "grad_norm": 0.3342432379722595,
      "learning_rate": 1.4522564597187676e-05,
      "loss": 0.0651,
      "step": 16185
    },
    {
      "epoch": 0.8218691304127113,
      "grad_norm": 0.38529881834983826,
      "learning_rate": 1.4520872463915258e-05,
      "loss": 0.082,
      "step": 16190
    },
    {
      "epoch": 0.8221229504035737,
      "grad_norm": 0.45302921533584595,
      "learning_rate": 1.4519180330642843e-05,
      "loss": 0.0688,
      "step": 16195
    },
    {
      "epoch": 0.8223767703944362,
      "grad_norm": 0.33290159702301025,
      "learning_rate": 1.4517488197370427e-05,
      "loss": 0.075,
      "step": 16200
    },
    {
      "epoch": 0.8226305903852987,
      "grad_norm": 0.34083133935928345,
      "learning_rate": 1.4515796064098009e-05,
      "loss": 0.0755,
      "step": 16205
    },
    {
      "epoch": 0.8228844103761612,
      "grad_norm": 0.3705480992794037,
      "learning_rate": 1.4514103930825594e-05,
      "loss": 0.0721,
      "step": 16210
    },
    {
      "epoch": 0.8231382303670237,
      "grad_norm": 0.39600300788879395,
      "learning_rate": 1.4512411797553176e-05,
      "loss": 0.0708,
      "step": 16215
    },
    {
      "epoch": 0.8233920503578862,
      "grad_norm": 0.32571184635162354,
      "learning_rate": 1.451071966428076e-05,
      "loss": 0.0758,
      "step": 16220
    },
    {
      "epoch": 0.8236458703487487,
      "grad_norm": 0.791690468788147,
      "learning_rate": 1.4509027531008344e-05,
      "loss": 0.084,
      "step": 16225
    },
    {
      "epoch": 0.8238996903396112,
      "grad_norm": 0.3258461058139801,
      "learning_rate": 1.4507335397735926e-05,
      "loss": 0.0741,
      "step": 16230
    },
    {
      "epoch": 0.8241535103304737,
      "grad_norm": 0.3211754262447357,
      "learning_rate": 1.450564326446351e-05,
      "loss": 0.079,
      "step": 16235
    },
    {
      "epoch": 0.8244073303213361,
      "grad_norm": 0.3197133541107178,
      "learning_rate": 1.4503951131191093e-05,
      "loss": 0.0674,
      "step": 16240
    },
    {
      "epoch": 0.8246611503121986,
      "grad_norm": 0.39093726873397827,
      "learning_rate": 1.4502258997918677e-05,
      "loss": 0.0686,
      "step": 16245
    },
    {
      "epoch": 0.824914970303061,
      "grad_norm": 0.4963763356208801,
      "learning_rate": 1.4500566864646262e-05,
      "loss": 0.0775,
      "step": 16250
    },
    {
      "epoch": 0.8251687902939235,
      "grad_norm": 0.40643295645713806,
      "learning_rate": 1.4498874731373844e-05,
      "loss": 0.0698,
      "step": 16255
    },
    {
      "epoch": 0.825422610284786,
      "grad_norm": 0.39619845151901245,
      "learning_rate": 1.4497182598101428e-05,
      "loss": 0.0713,
      "step": 16260
    },
    {
      "epoch": 0.8256764302756485,
      "grad_norm": 0.4349762201309204,
      "learning_rate": 1.4495490464829011e-05,
      "loss": 0.0682,
      "step": 16265
    },
    {
      "epoch": 0.825930250266511,
      "grad_norm": 0.5058432817459106,
      "learning_rate": 1.4493798331556595e-05,
      "loss": 0.0729,
      "step": 16270
    },
    {
      "epoch": 0.8261840702573735,
      "grad_norm": 0.3339829444885254,
      "learning_rate": 1.4492106198284177e-05,
      "loss": 0.0715,
      "step": 16275
    },
    {
      "epoch": 0.826437890248236,
      "grad_norm": 0.33098337054252625,
      "learning_rate": 1.4490414065011762e-05,
      "loss": 0.0671,
      "step": 16280
    },
    {
      "epoch": 0.8266917102390984,
      "grad_norm": 0.42175155878067017,
      "learning_rate": 1.4488721931739345e-05,
      "loss": 0.071,
      "step": 16285
    },
    {
      "epoch": 0.8269455302299609,
      "grad_norm": 0.8576765656471252,
      "learning_rate": 1.4487029798466929e-05,
      "loss": 0.0698,
      "step": 16290
    },
    {
      "epoch": 0.8271993502208234,
      "grad_norm": 1.0109632015228271,
      "learning_rate": 1.4485337665194512e-05,
      "loss": 0.0794,
      "step": 16295
    },
    {
      "epoch": 0.8274531702116859,
      "grad_norm": 0.4387573301792145,
      "learning_rate": 1.4483645531922094e-05,
      "loss": 0.0758,
      "step": 16300
    },
    {
      "epoch": 0.8277069902025483,
      "grad_norm": 0.3630094826221466,
      "learning_rate": 1.448195339864968e-05,
      "loss": 0.0605,
      "step": 16305
    },
    {
      "epoch": 0.8279608101934108,
      "grad_norm": 0.4075300991535187,
      "learning_rate": 1.4480261265377263e-05,
      "loss": 0.0648,
      "step": 16310
    },
    {
      "epoch": 0.8282146301842733,
      "grad_norm": 0.37269020080566406,
      "learning_rate": 1.4478569132104845e-05,
      "loss": 0.0723,
      "step": 16315
    },
    {
      "epoch": 0.8284684501751358,
      "grad_norm": 0.6695398092269897,
      "learning_rate": 1.447687699883243e-05,
      "loss": 0.0678,
      "step": 16320
    },
    {
      "epoch": 0.8287222701659983,
      "grad_norm": 0.36002203822135925,
      "learning_rate": 1.4475184865560012e-05,
      "loss": 0.0634,
      "step": 16325
    },
    {
      "epoch": 0.8289760901568608,
      "grad_norm": 0.6074070334434509,
      "learning_rate": 1.4473492732287596e-05,
      "loss": 0.0842,
      "step": 16330
    },
    {
      "epoch": 0.8292299101477232,
      "grad_norm": 0.37209317088127136,
      "learning_rate": 1.447180059901518e-05,
      "loss": 0.0701,
      "step": 16335
    },
    {
      "epoch": 0.8294837301385857,
      "grad_norm": 0.3380017578601837,
      "learning_rate": 1.4470108465742763e-05,
      "loss": 0.0701,
      "step": 16340
    },
    {
      "epoch": 0.8297375501294482,
      "grad_norm": 0.6713643074035645,
      "learning_rate": 1.4468416332470348e-05,
      "loss": 0.0703,
      "step": 16345
    },
    {
      "epoch": 0.8299913701203107,
      "grad_norm": 0.9763821959495544,
      "learning_rate": 1.446672419919793e-05,
      "loss": 0.0714,
      "step": 16350
    },
    {
      "epoch": 0.8302451901111731,
      "grad_norm": 0.3959788680076599,
      "learning_rate": 1.4465032065925513e-05,
      "loss": 0.0737,
      "step": 16355
    },
    {
      "epoch": 0.8304990101020356,
      "grad_norm": 1.2258071899414062,
      "learning_rate": 1.4463339932653098e-05,
      "loss": 0.0774,
      "step": 16360
    },
    {
      "epoch": 0.8307528300928981,
      "grad_norm": 0.34479087591171265,
      "learning_rate": 1.446164779938068e-05,
      "loss": 0.0748,
      "step": 16365
    },
    {
      "epoch": 0.8310066500837606,
      "grad_norm": 0.39507466554641724,
      "learning_rate": 1.4459955666108262e-05,
      "loss": 0.0747,
      "step": 16370
    },
    {
      "epoch": 0.8312604700746231,
      "grad_norm": 0.5069799423217773,
      "learning_rate": 1.4458263532835847e-05,
      "loss": 0.0771,
      "step": 16375
    },
    {
      "epoch": 0.8315142900654856,
      "grad_norm": 0.5372797846794128,
      "learning_rate": 1.4456571399563431e-05,
      "loss": 0.0886,
      "step": 16380
    },
    {
      "epoch": 0.831768110056348,
      "grad_norm": 0.5318045616149902,
      "learning_rate": 1.4454879266291015e-05,
      "loss": 0.0835,
      "step": 16385
    },
    {
      "epoch": 0.8320219300472105,
      "grad_norm": 0.3830041289329529,
      "learning_rate": 1.4453187133018598e-05,
      "loss": 0.0646,
      "step": 16390
    },
    {
      "epoch": 0.832275750038073,
      "grad_norm": 0.40192267298698425,
      "learning_rate": 1.445149499974618e-05,
      "loss": 0.074,
      "step": 16395
    },
    {
      "epoch": 0.8325295700289355,
      "grad_norm": 0.6122369170188904,
      "learning_rate": 1.4449802866473765e-05,
      "loss": 0.0659,
      "step": 16400
    },
    {
      "epoch": 0.832783390019798,
      "grad_norm": 0.455731064081192,
      "learning_rate": 1.4448110733201349e-05,
      "loss": 0.08,
      "step": 16405
    },
    {
      "epoch": 0.8330372100106604,
      "grad_norm": 0.3991694450378418,
      "learning_rate": 1.444641859992893e-05,
      "loss": 0.0677,
      "step": 16410
    },
    {
      "epoch": 0.8332910300015229,
      "grad_norm": 0.49834415316581726,
      "learning_rate": 1.4444726466656516e-05,
      "loss": 0.0779,
      "step": 16415
    },
    {
      "epoch": 0.8335448499923854,
      "grad_norm": 0.3147116005420685,
      "learning_rate": 1.4443034333384098e-05,
      "loss": 0.0675,
      "step": 16420
    },
    {
      "epoch": 0.8337986699832479,
      "grad_norm": 0.4364480674266815,
      "learning_rate": 1.4441342200111681e-05,
      "loss": 0.0724,
      "step": 16425
    },
    {
      "epoch": 0.8340524899741104,
      "grad_norm": 0.3125317096710205,
      "learning_rate": 1.4439650066839266e-05,
      "loss": 0.0704,
      "step": 16430
    },
    {
      "epoch": 0.8343063099649728,
      "grad_norm": 0.3873157799243927,
      "learning_rate": 1.4437957933566848e-05,
      "loss": 0.0661,
      "step": 16435
    },
    {
      "epoch": 0.8345601299558353,
      "grad_norm": 0.24474631249904633,
      "learning_rate": 1.4436265800294434e-05,
      "loss": 0.0687,
      "step": 16440
    },
    {
      "epoch": 0.8348139499466978,
      "grad_norm": 0.37103623151779175,
      "learning_rate": 1.4434573667022015e-05,
      "loss": 0.0598,
      "step": 16445
    },
    {
      "epoch": 0.8350677699375603,
      "grad_norm": 0.2716129422187805,
      "learning_rate": 1.4432881533749599e-05,
      "loss": 0.0674,
      "step": 16450
    },
    {
      "epoch": 0.8353215899284228,
      "grad_norm": 0.39446020126342773,
      "learning_rate": 1.4431189400477184e-05,
      "loss": 0.0708,
      "step": 16455
    },
    {
      "epoch": 0.8355754099192853,
      "grad_norm": 0.39217862486839294,
      "learning_rate": 1.4429497267204766e-05,
      "loss": 0.0718,
      "step": 16460
    },
    {
      "epoch": 0.8358292299101477,
      "grad_norm": 0.5310700535774231,
      "learning_rate": 1.442780513393235e-05,
      "loss": 0.078,
      "step": 16465
    },
    {
      "epoch": 0.8360830499010102,
      "grad_norm": 0.337480753660202,
      "learning_rate": 1.4426113000659933e-05,
      "loss": 0.0693,
      "step": 16470
    },
    {
      "epoch": 0.8363368698918727,
      "grad_norm": 0.45497629046440125,
      "learning_rate": 1.4424420867387517e-05,
      "loss": 0.0831,
      "step": 16475
    },
    {
      "epoch": 0.8365906898827352,
      "grad_norm": 0.3556462526321411,
      "learning_rate": 1.4422728734115098e-05,
      "loss": 0.0653,
      "step": 16480
    },
    {
      "epoch": 0.8368445098735976,
      "grad_norm": 0.43825340270996094,
      "learning_rate": 1.4421036600842684e-05,
      "loss": 0.0688,
      "step": 16485
    },
    {
      "epoch": 0.8370983298644601,
      "grad_norm": 0.4429861605167389,
      "learning_rate": 1.4419344467570267e-05,
      "loss": 0.0736,
      "step": 16490
    },
    {
      "epoch": 0.8373521498553226,
      "grad_norm": 0.35255303978919983,
      "learning_rate": 1.441765233429785e-05,
      "loss": 0.066,
      "step": 16495
    },
    {
      "epoch": 0.8376059698461851,
      "grad_norm": 0.532727062702179,
      "learning_rate": 1.4415960201025434e-05,
      "loss": 0.0857,
      "step": 16500
    },
    {
      "epoch": 0.8378597898370476,
      "grad_norm": 0.38991039991378784,
      "learning_rate": 1.4414268067753016e-05,
      "loss": 0.0721,
      "step": 16505
    },
    {
      "epoch": 0.83811360982791,
      "grad_norm": 0.38206273317337036,
      "learning_rate": 1.4412575934480601e-05,
      "loss": 0.0816,
      "step": 16510
    },
    {
      "epoch": 0.8383674298187725,
      "grad_norm": 0.271957665681839,
      "learning_rate": 1.4410883801208185e-05,
      "loss": 0.0779,
      "step": 16515
    },
    {
      "epoch": 0.838621249809635,
      "grad_norm": 0.3314172327518463,
      "learning_rate": 1.4409191667935767e-05,
      "loss": 0.0634,
      "step": 16520
    },
    {
      "epoch": 0.8388750698004975,
      "grad_norm": 0.5224104523658752,
      "learning_rate": 1.4407499534663352e-05,
      "loss": 0.077,
      "step": 16525
    },
    {
      "epoch": 0.83912888979136,
      "grad_norm": 0.7262583374977112,
      "learning_rate": 1.4405807401390934e-05,
      "loss": 0.0909,
      "step": 16530
    },
    {
      "epoch": 0.8393827097822224,
      "grad_norm": 0.45320960879325867,
      "learning_rate": 1.440411526811852e-05,
      "loss": 0.0732,
      "step": 16535
    },
    {
      "epoch": 0.8396365297730849,
      "grad_norm": 0.3000989854335785,
      "learning_rate": 1.4402423134846103e-05,
      "loss": 0.0695,
      "step": 16540
    },
    {
      "epoch": 0.8398903497639474,
      "grad_norm": 0.6859510540962219,
      "learning_rate": 1.4400731001573685e-05,
      "loss": 0.0705,
      "step": 16545
    },
    {
      "epoch": 0.8401441697548099,
      "grad_norm": 0.4406552314758301,
      "learning_rate": 1.439903886830127e-05,
      "loss": 0.0774,
      "step": 16550
    },
    {
      "epoch": 0.8403979897456724,
      "grad_norm": 0.3408234715461731,
      "learning_rate": 1.4397346735028852e-05,
      "loss": 0.0699,
      "step": 16555
    },
    {
      "epoch": 0.8406518097365349,
      "grad_norm": 1.4002503156661987,
      "learning_rate": 1.4395654601756435e-05,
      "loss": 0.0739,
      "step": 16560
    },
    {
      "epoch": 0.8409056297273974,
      "grad_norm": 0.5265543460845947,
      "learning_rate": 1.4393962468484019e-05,
      "loss": 0.0817,
      "step": 16565
    },
    {
      "epoch": 0.8411594497182598,
      "grad_norm": 0.45000794529914856,
      "learning_rate": 1.4392270335211602e-05,
      "loss": 0.0729,
      "step": 16570
    },
    {
      "epoch": 0.8414132697091223,
      "grad_norm": 0.2590148150920868,
      "learning_rate": 1.4390578201939184e-05,
      "loss": 0.0773,
      "step": 16575
    },
    {
      "epoch": 0.8416670896999847,
      "grad_norm": 0.28776997327804565,
      "learning_rate": 1.438888606866677e-05,
      "loss": 0.0686,
      "step": 16580
    },
    {
      "epoch": 0.8419209096908472,
      "grad_norm": 0.6979045271873474,
      "learning_rate": 1.4387193935394353e-05,
      "loss": 0.0793,
      "step": 16585
    },
    {
      "epoch": 0.8421747296817097,
      "grad_norm": 0.33594322204589844,
      "learning_rate": 1.4385501802121936e-05,
      "loss": 0.0708,
      "step": 16590
    },
    {
      "epoch": 0.8424285496725722,
      "grad_norm": 0.4978824555873871,
      "learning_rate": 1.438380966884952e-05,
      "loss": 0.0698,
      "step": 16595
    },
    {
      "epoch": 0.8426823696634347,
      "grad_norm": 0.41426607966423035,
      "learning_rate": 1.4382117535577102e-05,
      "loss": 0.0734,
      "step": 16600
    },
    {
      "epoch": 0.8429361896542972,
      "grad_norm": 0.44715553522109985,
      "learning_rate": 1.4380425402304687e-05,
      "loss": 0.0693,
      "step": 16605
    },
    {
      "epoch": 0.8431900096451597,
      "grad_norm": 0.33712249994277954,
      "learning_rate": 1.437873326903227e-05,
      "loss": 0.0695,
      "step": 16610
    },
    {
      "epoch": 0.8434438296360222,
      "grad_norm": 0.4558604061603546,
      "learning_rate": 1.4377041135759852e-05,
      "loss": 0.0802,
      "step": 16615
    },
    {
      "epoch": 0.8436976496268846,
      "grad_norm": 0.4930158853530884,
      "learning_rate": 1.4375349002487438e-05,
      "loss": 0.066,
      "step": 16620
    },
    {
      "epoch": 0.8439514696177471,
      "grad_norm": 0.42849859595298767,
      "learning_rate": 1.437365686921502e-05,
      "loss": 0.0672,
      "step": 16625
    },
    {
      "epoch": 0.8442052896086095,
      "grad_norm": 0.4327075183391571,
      "learning_rate": 1.4371964735942605e-05,
      "loss": 0.0713,
      "step": 16630
    },
    {
      "epoch": 0.844459109599472,
      "grad_norm": 0.4134119749069214,
      "learning_rate": 1.4370272602670188e-05,
      "loss": 0.0785,
      "step": 16635
    },
    {
      "epoch": 0.8447129295903345,
      "grad_norm": 0.4089963436126709,
      "learning_rate": 1.436858046939777e-05,
      "loss": 0.0609,
      "step": 16640
    },
    {
      "epoch": 0.844966749581197,
      "grad_norm": 0.3104897737503052,
      "learning_rate": 1.4366888336125355e-05,
      "loss": 0.0628,
      "step": 16645
    },
    {
      "epoch": 0.8452205695720595,
      "grad_norm": 0.40896421670913696,
      "learning_rate": 1.4365196202852937e-05,
      "loss": 0.0745,
      "step": 16650
    },
    {
      "epoch": 0.845474389562922,
      "grad_norm": 0.5574126839637756,
      "learning_rate": 1.436350406958052e-05,
      "loss": 0.0812,
      "step": 16655
    },
    {
      "epoch": 0.8457282095537845,
      "grad_norm": 0.3526981770992279,
      "learning_rate": 1.4361811936308106e-05,
      "loss": 0.074,
      "step": 16660
    },
    {
      "epoch": 0.845982029544647,
      "grad_norm": 0.37888669967651367,
      "learning_rate": 1.4360119803035688e-05,
      "loss": 0.0769,
      "step": 16665
    },
    {
      "epoch": 0.8462358495355095,
      "grad_norm": 0.41363152861595154,
      "learning_rate": 1.4358427669763271e-05,
      "loss": 0.0788,
      "step": 16670
    },
    {
      "epoch": 0.846489669526372,
      "grad_norm": 1.3909887075424194,
      "learning_rate": 1.4356735536490855e-05,
      "loss": 0.0736,
      "step": 16675
    },
    {
      "epoch": 0.8467434895172343,
      "grad_norm": 0.40531304478645325,
      "learning_rate": 1.4355043403218439e-05,
      "loss": 0.0637,
      "step": 16680
    },
    {
      "epoch": 0.8469973095080968,
      "grad_norm": 0.291532963514328,
      "learning_rate": 1.4353351269946024e-05,
      "loss": 0.0682,
      "step": 16685
    },
    {
      "epoch": 0.8472511294989593,
      "grad_norm": 0.3950155973434448,
      "learning_rate": 1.4351659136673606e-05,
      "loss": 0.0655,
      "step": 16690
    },
    {
      "epoch": 0.8475049494898218,
      "grad_norm": 0.44813594222068787,
      "learning_rate": 1.434996700340119e-05,
      "loss": 0.0853,
      "step": 16695
    },
    {
      "epoch": 0.8477587694806843,
      "grad_norm": 0.38060590624809265,
      "learning_rate": 1.4348274870128773e-05,
      "loss": 0.0691,
      "step": 16700
    },
    {
      "epoch": 0.8480125894715468,
      "grad_norm": 0.546687126159668,
      "learning_rate": 1.4346582736856356e-05,
      "loss": 0.0684,
      "step": 16705
    },
    {
      "epoch": 0.8482664094624093,
      "grad_norm": 0.22715900838375092,
      "learning_rate": 1.4344890603583938e-05,
      "loss": 0.059,
      "step": 16710
    },
    {
      "epoch": 0.8485202294532718,
      "grad_norm": 0.290096253156662,
      "learning_rate": 1.4343198470311523e-05,
      "loss": 0.0594,
      "step": 16715
    },
    {
      "epoch": 0.8487740494441343,
      "grad_norm": 0.384205162525177,
      "learning_rate": 1.4341506337039107e-05,
      "loss": 0.0711,
      "step": 16720
    },
    {
      "epoch": 0.8490278694349968,
      "grad_norm": 0.2885967195034027,
      "learning_rate": 1.4339814203766689e-05,
      "loss": 0.0674,
      "step": 16725
    },
    {
      "epoch": 0.8492816894258591,
      "grad_norm": 0.5165429711341858,
      "learning_rate": 1.4338122070494274e-05,
      "loss": 0.0674,
      "step": 16730
    },
    {
      "epoch": 0.8495355094167216,
      "grad_norm": 0.36283811926841736,
      "learning_rate": 1.4336429937221856e-05,
      "loss": 0.0731,
      "step": 16735
    },
    {
      "epoch": 0.8497893294075841,
      "grad_norm": 0.4545622766017914,
      "learning_rate": 1.4334737803949441e-05,
      "loss": 0.0768,
      "step": 16740
    },
    {
      "epoch": 0.8500431493984466,
      "grad_norm": 0.30138736963272095,
      "learning_rate": 1.4333045670677023e-05,
      "loss": 0.0595,
      "step": 16745
    },
    {
      "epoch": 0.8502969693893091,
      "grad_norm": 0.38298335671424866,
      "learning_rate": 1.4331353537404606e-05,
      "loss": 0.0751,
      "step": 16750
    },
    {
      "epoch": 0.8505507893801716,
      "grad_norm": 0.6622296571731567,
      "learning_rate": 1.4329661404132192e-05,
      "loss": 0.0842,
      "step": 16755
    },
    {
      "epoch": 0.8508046093710341,
      "grad_norm": 0.5098762512207031,
      "learning_rate": 1.4327969270859774e-05,
      "loss": 0.0754,
      "step": 16760
    },
    {
      "epoch": 0.8510584293618966,
      "grad_norm": 0.4182951748371124,
      "learning_rate": 1.4326277137587357e-05,
      "loss": 0.055,
      "step": 16765
    },
    {
      "epoch": 0.8513122493527591,
      "grad_norm": 0.34056827425956726,
      "learning_rate": 1.432458500431494e-05,
      "loss": 0.0762,
      "step": 16770
    },
    {
      "epoch": 0.8515660693436216,
      "grad_norm": 0.4602746367454529,
      "learning_rate": 1.4322892871042524e-05,
      "loss": 0.0731,
      "step": 16775
    },
    {
      "epoch": 0.8518198893344839,
      "grad_norm": 0.46367478370666504,
      "learning_rate": 1.432120073777011e-05,
      "loss": 0.0662,
      "step": 16780
    },
    {
      "epoch": 0.8520737093253464,
      "grad_norm": 0.3692100942134857,
      "learning_rate": 1.4319508604497691e-05,
      "loss": 0.0732,
      "step": 16785
    },
    {
      "epoch": 0.8523275293162089,
      "grad_norm": 0.32194337248802185,
      "learning_rate": 1.4317816471225275e-05,
      "loss": 0.0694,
      "step": 16790
    },
    {
      "epoch": 0.8525813493070714,
      "grad_norm": 0.4015413820743561,
      "learning_rate": 1.4316124337952858e-05,
      "loss": 0.066,
      "step": 16795
    },
    {
      "epoch": 0.8528351692979339,
      "grad_norm": 0.4497564435005188,
      "learning_rate": 1.4314432204680442e-05,
      "loss": 0.0801,
      "step": 16800
    },
    {
      "epoch": 0.8530889892887964,
      "grad_norm": 0.5693073272705078,
      "learning_rate": 1.4312740071408024e-05,
      "loss": 0.0772,
      "step": 16805
    },
    {
      "epoch": 0.8533428092796589,
      "grad_norm": 0.7299516201019287,
      "learning_rate": 1.4311047938135609e-05,
      "loss": 0.0772,
      "step": 16810
    },
    {
      "epoch": 0.8535966292705214,
      "grad_norm": 0.370604544878006,
      "learning_rate": 1.4309355804863193e-05,
      "loss": 0.0696,
      "step": 16815
    },
    {
      "epoch": 0.8538504492613839,
      "grad_norm": 0.33534955978393555,
      "learning_rate": 1.4307663671590774e-05,
      "loss": 0.0736,
      "step": 16820
    },
    {
      "epoch": 0.8541042692522464,
      "grad_norm": 0.5034054517745972,
      "learning_rate": 1.430597153831836e-05,
      "loss": 0.0769,
      "step": 16825
    },
    {
      "epoch": 0.8543580892431087,
      "grad_norm": 0.2479061633348465,
      "learning_rate": 1.4304279405045942e-05,
      "loss": 0.0654,
      "step": 16830
    },
    {
      "epoch": 0.8546119092339712,
      "grad_norm": 0.5529633164405823,
      "learning_rate": 1.4302587271773527e-05,
      "loss": 0.069,
      "step": 16835
    },
    {
      "epoch": 0.8548657292248337,
      "grad_norm": 0.4025406837463379,
      "learning_rate": 1.430089513850111e-05,
      "loss": 0.0684,
      "step": 16840
    },
    {
      "epoch": 0.8551195492156962,
      "grad_norm": 0.3485677242279053,
      "learning_rate": 1.4299203005228692e-05,
      "loss": 0.0658,
      "step": 16845
    },
    {
      "epoch": 0.8553733692065587,
      "grad_norm": 0.46029365062713623,
      "learning_rate": 1.4297510871956277e-05,
      "loss": 0.0692,
      "step": 16850
    },
    {
      "epoch": 0.8556271891974212,
      "grad_norm": 0.47386330366134644,
      "learning_rate": 1.429581873868386e-05,
      "loss": 0.0738,
      "step": 16855
    },
    {
      "epoch": 0.8558810091882837,
      "grad_norm": 0.3624202311038971,
      "learning_rate": 1.4294126605411443e-05,
      "loss": 0.0723,
      "step": 16860
    },
    {
      "epoch": 0.8561348291791462,
      "grad_norm": 1.0836365222930908,
      "learning_rate": 1.4292434472139028e-05,
      "loss": 0.0803,
      "step": 16865
    },
    {
      "epoch": 0.8563886491700087,
      "grad_norm": 0.45594584941864014,
      "learning_rate": 1.429074233886661e-05,
      "loss": 0.0678,
      "step": 16870
    },
    {
      "epoch": 0.8566424691608711,
      "grad_norm": 0.34718087315559387,
      "learning_rate": 1.4289050205594195e-05,
      "loss": 0.0613,
      "step": 16875
    },
    {
      "epoch": 0.8568962891517335,
      "grad_norm": 0.30333346128463745,
      "learning_rate": 1.4287358072321777e-05,
      "loss": 0.0635,
      "step": 16880
    },
    {
      "epoch": 0.857150109142596,
      "grad_norm": 0.44795432686805725,
      "learning_rate": 1.428566593904936e-05,
      "loss": 0.0678,
      "step": 16885
    },
    {
      "epoch": 0.8574039291334585,
      "grad_norm": 0.33551299571990967,
      "learning_rate": 1.4283973805776946e-05,
      "loss": 0.0641,
      "step": 16890
    },
    {
      "epoch": 0.857657749124321,
      "grad_norm": 0.42668506503105164,
      "learning_rate": 1.4282281672504528e-05,
      "loss": 0.0738,
      "step": 16895
    },
    {
      "epoch": 0.8579115691151835,
      "grad_norm": 0.5033948421478271,
      "learning_rate": 1.4280589539232111e-05,
      "loss": 0.0791,
      "step": 16900
    },
    {
      "epoch": 0.858165389106046,
      "grad_norm": 0.5407020449638367,
      "learning_rate": 1.4278897405959695e-05,
      "loss": 0.0817,
      "step": 16905
    },
    {
      "epoch": 0.8584192090969085,
      "grad_norm": 1.9471144676208496,
      "learning_rate": 1.4277205272687278e-05,
      "loss": 0.0785,
      "step": 16910
    },
    {
      "epoch": 0.858673029087771,
      "grad_norm": 0.38863787055015564,
      "learning_rate": 1.427551313941486e-05,
      "loss": 0.067,
      "step": 16915
    },
    {
      "epoch": 0.8589268490786335,
      "grad_norm": 0.5478694438934326,
      "learning_rate": 1.4273821006142445e-05,
      "loss": 0.0718,
      "step": 16920
    },
    {
      "epoch": 0.8591806690694959,
      "grad_norm": 0.5812584757804871,
      "learning_rate": 1.4272128872870029e-05,
      "loss": 0.0845,
      "step": 16925
    },
    {
      "epoch": 0.8594344890603584,
      "grad_norm": 0.4321596324443817,
      "learning_rate": 1.4270436739597612e-05,
      "loss": 0.0734,
      "step": 16930
    },
    {
      "epoch": 0.8596883090512208,
      "grad_norm": 0.512316107749939,
      "learning_rate": 1.4268744606325196e-05,
      "loss": 0.065,
      "step": 16935
    },
    {
      "epoch": 0.8599421290420833,
      "grad_norm": 0.4328327775001526,
      "learning_rate": 1.4267052473052778e-05,
      "loss": 0.0685,
      "step": 16940
    },
    {
      "epoch": 0.8601959490329458,
      "grad_norm": 0.5791327357292175,
      "learning_rate": 1.4265360339780363e-05,
      "loss": 0.0775,
      "step": 16945
    },
    {
      "epoch": 0.8604497690238083,
      "grad_norm": 0.387787401676178,
      "learning_rate": 1.4263668206507945e-05,
      "loss": 0.069,
      "step": 16950
    },
    {
      "epoch": 0.8607035890146708,
      "grad_norm": 0.38102301955223083,
      "learning_rate": 1.4261976073235528e-05,
      "loss": 0.067,
      "step": 16955
    },
    {
      "epoch": 0.8609574090055333,
      "grad_norm": 0.3059747815132141,
      "learning_rate": 1.4260283939963114e-05,
      "loss": 0.0582,
      "step": 16960
    },
    {
      "epoch": 0.8612112289963958,
      "grad_norm": 0.38691335916519165,
      "learning_rate": 1.4258591806690696e-05,
      "loss": 0.0761,
      "step": 16965
    },
    {
      "epoch": 0.8614650489872583,
      "grad_norm": 0.3976000249385834,
      "learning_rate": 1.4256899673418279e-05,
      "loss": 0.0695,
      "step": 16970
    },
    {
      "epoch": 0.8617188689781207,
      "grad_norm": 0.4070408344268799,
      "learning_rate": 1.4255207540145863e-05,
      "loss": 0.0702,
      "step": 16975
    },
    {
      "epoch": 0.8619726889689832,
      "grad_norm": 0.6455148458480835,
      "learning_rate": 1.4253515406873446e-05,
      "loss": 0.0797,
      "step": 16980
    },
    {
      "epoch": 0.8622265089598457,
      "grad_norm": 0.36329150199890137,
      "learning_rate": 1.4251823273601031e-05,
      "loss": 0.0669,
      "step": 16985
    },
    {
      "epoch": 0.8624803289507081,
      "grad_norm": 0.36581847071647644,
      "learning_rate": 1.4250131140328613e-05,
      "loss": 0.0639,
      "step": 16990
    },
    {
      "epoch": 0.8627341489415706,
      "grad_norm": 0.2543756067752838,
      "learning_rate": 1.4248439007056197e-05,
      "loss": 0.064,
      "step": 16995
    },
    {
      "epoch": 0.8629879689324331,
      "grad_norm": 0.7665139436721802,
      "learning_rate": 1.424674687378378e-05,
      "loss": 0.072,
      "step": 17000
    },
    {
      "epoch": 0.8632417889232956,
      "grad_norm": 0.2957867383956909,
      "learning_rate": 1.4245054740511364e-05,
      "loss": 0.0718,
      "step": 17005
    },
    {
      "epoch": 0.8634956089141581,
      "grad_norm": 0.5713722705841064,
      "learning_rate": 1.4243362607238946e-05,
      "loss": 0.0748,
      "step": 17010
    },
    {
      "epoch": 0.8637494289050206,
      "grad_norm": 0.43129876255989075,
      "learning_rate": 1.4241670473966531e-05,
      "loss": 0.085,
      "step": 17015
    },
    {
      "epoch": 0.8640032488958831,
      "grad_norm": 0.6972838640213013,
      "learning_rate": 1.4239978340694115e-05,
      "loss": 0.0695,
      "step": 17020
    },
    {
      "epoch": 0.8642570688867455,
      "grad_norm": 0.8373203873634338,
      "learning_rate": 1.4238286207421698e-05,
      "loss": 0.0758,
      "step": 17025
    },
    {
      "epoch": 0.864510888877608,
      "grad_norm": 0.2909460961818695,
      "learning_rate": 1.4236594074149282e-05,
      "loss": 0.0682,
      "step": 17030
    },
    {
      "epoch": 0.8647647088684705,
      "grad_norm": 0.30259981751441956,
      "learning_rate": 1.4234901940876863e-05,
      "loss": 0.0677,
      "step": 17035
    },
    {
      "epoch": 0.865018528859333,
      "grad_norm": 0.36737337708473206,
      "learning_rate": 1.4233209807604449e-05,
      "loss": 0.0763,
      "step": 17040
    },
    {
      "epoch": 0.8652723488501954,
      "grad_norm": 0.32847535610198975,
      "learning_rate": 1.4231517674332032e-05,
      "loss": 0.0706,
      "step": 17045
    },
    {
      "epoch": 0.8655261688410579,
      "grad_norm": 0.4564855992794037,
      "learning_rate": 1.4229825541059614e-05,
      "loss": 0.0808,
      "step": 17050
    },
    {
      "epoch": 0.8657799888319204,
      "grad_norm": 0.4900711178779602,
      "learning_rate": 1.42281334077872e-05,
      "loss": 0.0595,
      "step": 17055
    },
    {
      "epoch": 0.8660338088227829,
      "grad_norm": 0.6517011523246765,
      "learning_rate": 1.4226441274514781e-05,
      "loss": 0.0665,
      "step": 17060
    },
    {
      "epoch": 0.8662876288136454,
      "grad_norm": 0.3914799094200134,
      "learning_rate": 1.4224749141242365e-05,
      "loss": 0.0696,
      "step": 17065
    },
    {
      "epoch": 0.8665414488045079,
      "grad_norm": 0.42431968450546265,
      "learning_rate": 1.422305700796995e-05,
      "loss": 0.0654,
      "step": 17070
    },
    {
      "epoch": 0.8667952687953703,
      "grad_norm": 0.34742167592048645,
      "learning_rate": 1.4221364874697532e-05,
      "loss": 0.0769,
      "step": 17075
    },
    {
      "epoch": 0.8670490887862328,
      "grad_norm": 0.8822436928749084,
      "learning_rate": 1.4219672741425117e-05,
      "loss": 0.0787,
      "step": 17080
    },
    {
      "epoch": 0.8673029087770953,
      "grad_norm": 0.6568184494972229,
      "learning_rate": 1.4217980608152699e-05,
      "loss": 0.0593,
      "step": 17085
    },
    {
      "epoch": 0.8675567287679578,
      "grad_norm": 0.53938227891922,
      "learning_rate": 1.4216288474880282e-05,
      "loss": 0.083,
      "step": 17090
    },
    {
      "epoch": 0.8678105487588202,
      "grad_norm": 0.4471435248851776,
      "learning_rate": 1.4214596341607868e-05,
      "loss": 0.0744,
      "step": 17095
    },
    {
      "epoch": 0.8680643687496827,
      "grad_norm": 0.3798215389251709,
      "learning_rate": 1.421290420833545e-05,
      "loss": 0.0754,
      "step": 17100
    },
    {
      "epoch": 0.8683181887405452,
      "grad_norm": 0.5007672309875488,
      "learning_rate": 1.4211212075063033e-05,
      "loss": 0.071,
      "step": 17105
    },
    {
      "epoch": 0.8685720087314077,
      "grad_norm": 0.4671100676059723,
      "learning_rate": 1.4209519941790617e-05,
      "loss": 0.0858,
      "step": 17110
    },
    {
      "epoch": 0.8688258287222702,
      "grad_norm": 0.4374212920665741,
      "learning_rate": 1.42078278085182e-05,
      "loss": 0.0684,
      "step": 17115
    },
    {
      "epoch": 0.8690796487131326,
      "grad_norm": 0.46207401156425476,
      "learning_rate": 1.4206135675245785e-05,
      "loss": 0.0733,
      "step": 17120
    },
    {
      "epoch": 0.8693334687039951,
      "grad_norm": 0.4559246301651001,
      "learning_rate": 1.4204443541973367e-05,
      "loss": 0.0798,
      "step": 17125
    },
    {
      "epoch": 0.8695872886948576,
      "grad_norm": 0.47378304600715637,
      "learning_rate": 1.4202751408700949e-05,
      "loss": 0.0763,
      "step": 17130
    },
    {
      "epoch": 0.8698411086857201,
      "grad_norm": 0.36055704951286316,
      "learning_rate": 1.4201059275428534e-05,
      "loss": 0.0663,
      "step": 17135
    },
    {
      "epoch": 0.8700949286765826,
      "grad_norm": 0.4687144160270691,
      "learning_rate": 1.4199367142156118e-05,
      "loss": 0.072,
      "step": 17140
    },
    {
      "epoch": 0.870348748667445,
      "grad_norm": 0.565639078617096,
      "learning_rate": 1.41976750088837e-05,
      "loss": 0.0769,
      "step": 17145
    },
    {
      "epoch": 0.8706025686583075,
      "grad_norm": 0.3410448431968689,
      "learning_rate": 1.4195982875611285e-05,
      "loss": 0.0585,
      "step": 17150
    },
    {
      "epoch": 0.87085638864917,
      "grad_norm": 0.42370325326919556,
      "learning_rate": 1.4194290742338867e-05,
      "loss": 0.0721,
      "step": 17155
    },
    {
      "epoch": 0.8711102086400325,
      "grad_norm": 0.6147048473358154,
      "learning_rate": 1.419259860906645e-05,
      "loss": 0.0712,
      "step": 17160
    },
    {
      "epoch": 0.871364028630895,
      "grad_norm": 0.3107832074165344,
      "learning_rate": 1.4190906475794036e-05,
      "loss": 0.0717,
      "step": 17165
    },
    {
      "epoch": 0.8716178486217574,
      "grad_norm": 0.4038301706314087,
      "learning_rate": 1.4189214342521617e-05,
      "loss": 0.0792,
      "step": 17170
    },
    {
      "epoch": 0.8718716686126199,
      "grad_norm": 0.3050367832183838,
      "learning_rate": 1.4187522209249203e-05,
      "loss": 0.0693,
      "step": 17175
    },
    {
      "epoch": 0.8721254886034824,
      "grad_norm": 0.3587893843650818,
      "learning_rate": 1.4185830075976785e-05,
      "loss": 0.0657,
      "step": 17180
    },
    {
      "epoch": 0.8723793085943449,
      "grad_norm": 0.3330988883972168,
      "learning_rate": 1.4184137942704368e-05,
      "loss": 0.0652,
      "step": 17185
    },
    {
      "epoch": 0.8726331285852074,
      "grad_norm": 0.3522055745124817,
      "learning_rate": 1.4182445809431953e-05,
      "loss": 0.0708,
      "step": 17190
    },
    {
      "epoch": 0.8728869485760699,
      "grad_norm": 0.35526609420776367,
      "learning_rate": 1.4180753676159535e-05,
      "loss": 0.0776,
      "step": 17195
    },
    {
      "epoch": 0.8731407685669323,
      "grad_norm": 0.41325056552886963,
      "learning_rate": 1.4179061542887119e-05,
      "loss": 0.0582,
      "step": 17200
    },
    {
      "epoch": 0.8733945885577948,
      "grad_norm": 0.3183586001396179,
      "learning_rate": 1.4177369409614702e-05,
      "loss": 0.0649,
      "step": 17205
    },
    {
      "epoch": 0.8736484085486573,
      "grad_norm": 0.3800111711025238,
      "learning_rate": 1.4175677276342286e-05,
      "loss": 0.0656,
      "step": 17210
    },
    {
      "epoch": 0.8739022285395198,
      "grad_norm": 0.31832781434059143,
      "learning_rate": 1.4173985143069868e-05,
      "loss": 0.075,
      "step": 17215
    },
    {
      "epoch": 0.8741560485303822,
      "grad_norm": 0.3157210350036621,
      "learning_rate": 1.4172293009797453e-05,
      "loss": 0.075,
      "step": 17220
    },
    {
      "epoch": 0.8744098685212447,
      "grad_norm": 0.36594507098197937,
      "learning_rate": 1.4170600876525036e-05,
      "loss": 0.0729,
      "step": 17225
    },
    {
      "epoch": 0.8746636885121072,
      "grad_norm": 0.4369713366031647,
      "learning_rate": 1.416890874325262e-05,
      "loss": 0.0576,
      "step": 17230
    },
    {
      "epoch": 0.8749175085029697,
      "grad_norm": 0.3689056932926178,
      "learning_rate": 1.4167216609980204e-05,
      "loss": 0.072,
      "step": 17235
    },
    {
      "epoch": 0.8751713284938322,
      "grad_norm": 0.2857867181301117,
      "learning_rate": 1.4165524476707785e-05,
      "loss": 0.0661,
      "step": 17240
    },
    {
      "epoch": 0.8754251484846947,
      "grad_norm": 0.4573242962360382,
      "learning_rate": 1.416383234343537e-05,
      "loss": 0.0656,
      "step": 17245
    },
    {
      "epoch": 0.8756789684755572,
      "grad_norm": 0.48289331793785095,
      "learning_rate": 1.4162140210162954e-05,
      "loss": 0.0797,
      "step": 17250
    },
    {
      "epoch": 0.8759327884664196,
      "grad_norm": 0.4658796191215515,
      "learning_rate": 1.4160448076890536e-05,
      "loss": 0.0687,
      "step": 17255
    },
    {
      "epoch": 0.8761866084572821,
      "grad_norm": 0.8445097804069519,
      "learning_rate": 1.4158755943618121e-05,
      "loss": 0.0683,
      "step": 17260
    },
    {
      "epoch": 0.8764404284481446,
      "grad_norm": 0.49470093846321106,
      "learning_rate": 1.4157063810345703e-05,
      "loss": 0.0727,
      "step": 17265
    },
    {
      "epoch": 0.876694248439007,
      "grad_norm": 0.5909218788146973,
      "learning_rate": 1.4155371677073288e-05,
      "loss": 0.0631,
      "step": 17270
    },
    {
      "epoch": 0.8769480684298695,
      "grad_norm": 0.33578768372535706,
      "learning_rate": 1.4153679543800872e-05,
      "loss": 0.0719,
      "step": 17275
    },
    {
      "epoch": 0.877201888420732,
      "grad_norm": 0.3844059109687805,
      "learning_rate": 1.4151987410528454e-05,
      "loss": 0.0691,
      "step": 17280
    },
    {
      "epoch": 0.8774557084115945,
      "grad_norm": 0.7239956259727478,
      "learning_rate": 1.4150295277256039e-05,
      "loss": 0.0643,
      "step": 17285
    },
    {
      "epoch": 0.877709528402457,
      "grad_norm": 0.4169676601886749,
      "learning_rate": 1.414860314398362e-05,
      "loss": 0.0677,
      "step": 17290
    },
    {
      "epoch": 0.8779633483933195,
      "grad_norm": 0.5026949644088745,
      "learning_rate": 1.4146911010711204e-05,
      "loss": 0.0826,
      "step": 17295
    },
    {
      "epoch": 0.878217168384182,
      "grad_norm": 0.353205144405365,
      "learning_rate": 1.414521887743879e-05,
      "loss": 0.0703,
      "step": 17300
    },
    {
      "epoch": 0.8784709883750444,
      "grad_norm": 0.5128482580184937,
      "learning_rate": 1.4143526744166371e-05,
      "loss": 0.0733,
      "step": 17305
    },
    {
      "epoch": 0.8787248083659069,
      "grad_norm": 0.43205496668815613,
      "learning_rate": 1.4141834610893953e-05,
      "loss": 0.0741,
      "step": 17310
    },
    {
      "epoch": 0.8789786283567694,
      "grad_norm": 0.36526092886924744,
      "learning_rate": 1.4140142477621539e-05,
      "loss": 0.0685,
      "step": 17315
    },
    {
      "epoch": 0.8792324483476318,
      "grad_norm": 1.4037833213806152,
      "learning_rate": 1.4138450344349122e-05,
      "loss": 0.0829,
      "step": 17320
    },
    {
      "epoch": 0.8794862683384943,
      "grad_norm": 0.5729219317436218,
      "learning_rate": 1.4136758211076707e-05,
      "loss": 0.0733,
      "step": 17325
    },
    {
      "epoch": 0.8797400883293568,
      "grad_norm": 0.421556681394577,
      "learning_rate": 1.413506607780429e-05,
      "loss": 0.0655,
      "step": 17330
    },
    {
      "epoch": 0.8799939083202193,
      "grad_norm": 0.46088892221450806,
      "learning_rate": 1.4133373944531871e-05,
      "loss": 0.069,
      "step": 17335
    },
    {
      "epoch": 0.8802477283110818,
      "grad_norm": 0.29208675026893616,
      "learning_rate": 1.4131681811259456e-05,
      "loss": 0.0572,
      "step": 17340
    },
    {
      "epoch": 0.8805015483019443,
      "grad_norm": 0.5103023648262024,
      "learning_rate": 1.412998967798704e-05,
      "loss": 0.0623,
      "step": 17345
    },
    {
      "epoch": 0.8807553682928068,
      "grad_norm": 0.4015849232673645,
      "learning_rate": 1.4128297544714622e-05,
      "loss": 0.0709,
      "step": 17350
    },
    {
      "epoch": 0.8810091882836693,
      "grad_norm": 0.2928534150123596,
      "learning_rate": 1.4126605411442207e-05,
      "loss": 0.0688,
      "step": 17355
    },
    {
      "epoch": 0.8812630082745317,
      "grad_norm": 0.2520070970058441,
      "learning_rate": 1.4124913278169789e-05,
      "loss": 0.0692,
      "step": 17360
    },
    {
      "epoch": 0.8815168282653942,
      "grad_norm": 0.736379086971283,
      "learning_rate": 1.4123221144897372e-05,
      "loss": 0.077,
      "step": 17365
    },
    {
      "epoch": 0.8817706482562566,
      "grad_norm": 0.3595697283744812,
      "learning_rate": 1.4121529011624958e-05,
      "loss": 0.0589,
      "step": 17370
    },
    {
      "epoch": 0.8820244682471191,
      "grad_norm": 0.3573293685913086,
      "learning_rate": 1.411983687835254e-05,
      "loss": 0.071,
      "step": 17375
    },
    {
      "epoch": 0.8822782882379816,
      "grad_norm": 0.5311951637268066,
      "learning_rate": 1.4118144745080125e-05,
      "loss": 0.0707,
      "step": 17380
    },
    {
      "epoch": 0.8825321082288441,
      "grad_norm": 0.4468015432357788,
      "learning_rate": 1.4116452611807706e-05,
      "loss": 0.0693,
      "step": 17385
    },
    {
      "epoch": 0.8827859282197066,
      "grad_norm": 0.5135281085968018,
      "learning_rate": 1.411476047853529e-05,
      "loss": 0.0648,
      "step": 17390
    },
    {
      "epoch": 0.8830397482105691,
      "grad_norm": 0.36561042070388794,
      "learning_rate": 1.4113068345262875e-05,
      "loss": 0.0621,
      "step": 17395
    },
    {
      "epoch": 0.8832935682014316,
      "grad_norm": 0.6639304757118225,
      "learning_rate": 1.4111376211990457e-05,
      "loss": 0.065,
      "step": 17400
    },
    {
      "epoch": 0.8835473881922941,
      "grad_norm": 0.4147944450378418,
      "learning_rate": 1.410968407871804e-05,
      "loss": 0.0744,
      "step": 17405
    },
    {
      "epoch": 0.8838012081831566,
      "grad_norm": 0.4257824420928955,
      "learning_rate": 1.4107991945445624e-05,
      "loss": 0.0732,
      "step": 17410
    },
    {
      "epoch": 0.8840550281740189,
      "grad_norm": 0.30751651525497437,
      "learning_rate": 1.4106299812173208e-05,
      "loss": 0.0694,
      "step": 17415
    },
    {
      "epoch": 0.8843088481648814,
      "grad_norm": 0.3239120543003082,
      "learning_rate": 1.4104607678900793e-05,
      "loss": 0.0672,
      "step": 17420
    },
    {
      "epoch": 0.8845626681557439,
      "grad_norm": 0.4700276553630829,
      "learning_rate": 1.4102915545628375e-05,
      "loss": 0.0714,
      "step": 17425
    },
    {
      "epoch": 0.8848164881466064,
      "grad_norm": 0.3352143466472626,
      "learning_rate": 1.4101223412355958e-05,
      "loss": 0.0731,
      "step": 17430
    },
    {
      "epoch": 0.8850703081374689,
      "grad_norm": 0.34849241375923157,
      "learning_rate": 1.4099531279083542e-05,
      "loss": 0.067,
      "step": 17435
    },
    {
      "epoch": 0.8853241281283314,
      "grad_norm": 0.48389196395874023,
      "learning_rate": 1.4097839145811125e-05,
      "loss": 0.0659,
      "step": 17440
    },
    {
      "epoch": 0.8855779481191939,
      "grad_norm": 0.4253050684928894,
      "learning_rate": 1.4096147012538707e-05,
      "loss": 0.071,
      "step": 17445
    },
    {
      "epoch": 0.8858317681100564,
      "grad_norm": 0.34934577345848083,
      "learning_rate": 1.4094454879266293e-05,
      "loss": 0.068,
      "step": 17450
    },
    {
      "epoch": 0.8860855881009189,
      "grad_norm": 0.3193303644657135,
      "learning_rate": 1.4092762745993876e-05,
      "loss": 0.065,
      "step": 17455
    },
    {
      "epoch": 0.8863394080917814,
      "grad_norm": 0.5447885990142822,
      "learning_rate": 1.4091070612721458e-05,
      "loss": 0.0703,
      "step": 17460
    },
    {
      "epoch": 0.8865932280826437,
      "grad_norm": 0.4106573462486267,
      "learning_rate": 1.4089378479449043e-05,
      "loss": 0.0764,
      "step": 17465
    },
    {
      "epoch": 0.8868470480735062,
      "grad_norm": 0.35047876834869385,
      "learning_rate": 1.4087686346176625e-05,
      "loss": 0.0725,
      "step": 17470
    },
    {
      "epoch": 0.8871008680643687,
      "grad_norm": 0.5358198881149292,
      "learning_rate": 1.408599421290421e-05,
      "loss": 0.0748,
      "step": 17475
    },
    {
      "epoch": 0.8873546880552312,
      "grad_norm": 0.5286594033241272,
      "learning_rate": 1.4084302079631794e-05,
      "loss": 0.0695,
      "step": 17480
    },
    {
      "epoch": 0.8876085080460937,
      "grad_norm": 0.7509157657623291,
      "learning_rate": 1.4082609946359376e-05,
      "loss": 0.0677,
      "step": 17485
    },
    {
      "epoch": 0.8878623280369562,
      "grad_norm": 0.44565531611442566,
      "learning_rate": 1.4080917813086961e-05,
      "loss": 0.0675,
      "step": 17490
    },
    {
      "epoch": 0.8881161480278187,
      "grad_norm": 0.403942734003067,
      "learning_rate": 1.4079225679814543e-05,
      "loss": 0.0695,
      "step": 17495
    },
    {
      "epoch": 0.8883699680186812,
      "grad_norm": 0.653939425945282,
      "learning_rate": 1.4077533546542126e-05,
      "loss": 0.0751,
      "step": 17500
    },
    {
      "epoch": 0.8886237880095437,
      "grad_norm": 0.3768764138221741,
      "learning_rate": 1.4075841413269712e-05,
      "loss": 0.0654,
      "step": 17505
    },
    {
      "epoch": 0.8888776080004062,
      "grad_norm": 0.33701881766319275,
      "learning_rate": 1.4074149279997293e-05,
      "loss": 0.0663,
      "step": 17510
    },
    {
      "epoch": 0.8891314279912685,
      "grad_norm": 0.3610353171825409,
      "learning_rate": 1.4072457146724879e-05,
      "loss": 0.0632,
      "step": 17515
    },
    {
      "epoch": 0.889385247982131,
      "grad_norm": 0.5949681997299194,
      "learning_rate": 1.407076501345246e-05,
      "loss": 0.0694,
      "step": 17520
    },
    {
      "epoch": 0.8896390679729935,
      "grad_norm": 0.34294983744621277,
      "learning_rate": 1.4069072880180044e-05,
      "loss": 0.062,
      "step": 17525
    },
    {
      "epoch": 0.889892887963856,
      "grad_norm": 0.3811400830745697,
      "learning_rate": 1.406738074690763e-05,
      "loss": 0.061,
      "step": 17530
    },
    {
      "epoch": 0.8901467079547185,
      "grad_norm": 0.3286953270435333,
      "learning_rate": 1.4065688613635211e-05,
      "loss": 0.0635,
      "step": 17535
    },
    {
      "epoch": 0.890400527945581,
      "grad_norm": 0.4377914071083069,
      "learning_rate": 1.4063996480362793e-05,
      "loss": 0.0699,
      "step": 17540
    },
    {
      "epoch": 0.8906543479364435,
      "grad_norm": 0.4538041949272156,
      "learning_rate": 1.4062304347090378e-05,
      "loss": 0.0776,
      "step": 17545
    },
    {
      "epoch": 0.890908167927306,
      "grad_norm": 0.3821781277656555,
      "learning_rate": 1.4060612213817962e-05,
      "loss": 0.073,
      "step": 17550
    },
    {
      "epoch": 0.8911619879181685,
      "grad_norm": 0.5155601501464844,
      "learning_rate": 1.4058920080545544e-05,
      "loss": 0.0629,
      "step": 17555
    },
    {
      "epoch": 0.891415807909031,
      "grad_norm": 0.452945739030838,
      "learning_rate": 1.4057227947273129e-05,
      "loss": 0.0632,
      "step": 17560
    },
    {
      "epoch": 0.8916696278998933,
      "grad_norm": 0.48677828907966614,
      "learning_rate": 1.405553581400071e-05,
      "loss": 0.0751,
      "step": 17565
    },
    {
      "epoch": 0.8919234478907558,
      "grad_norm": 0.5773299932479858,
      "learning_rate": 1.4053843680728296e-05,
      "loss": 0.0657,
      "step": 17570
    },
    {
      "epoch": 0.8921772678816183,
      "grad_norm": 0.47708895802497864,
      "learning_rate": 1.405215154745588e-05,
      "loss": 0.0694,
      "step": 17575
    },
    {
      "epoch": 0.8924310878724808,
      "grad_norm": 0.3603571951389313,
      "learning_rate": 1.4050459414183461e-05,
      "loss": 0.07,
      "step": 17580
    },
    {
      "epoch": 0.8926849078633433,
      "grad_norm": 0.40807899832725525,
      "learning_rate": 1.4048767280911047e-05,
      "loss": 0.0733,
      "step": 17585
    },
    {
      "epoch": 0.8929387278542058,
      "grad_norm": 0.3264356255531311,
      "learning_rate": 1.4047075147638628e-05,
      "loss": 0.0724,
      "step": 17590
    },
    {
      "epoch": 0.8931925478450683,
      "grad_norm": 0.4679695665836334,
      "learning_rate": 1.4045383014366212e-05,
      "loss": 0.0772,
      "step": 17595
    },
    {
      "epoch": 0.8934463678359308,
      "grad_norm": 0.3341139853000641,
      "learning_rate": 1.4043690881093797e-05,
      "loss": 0.0745,
      "step": 17600
    },
    {
      "epoch": 0.8937001878267933,
      "grad_norm": 0.3257398307323456,
      "learning_rate": 1.4041998747821379e-05,
      "loss": 0.0736,
      "step": 17605
    },
    {
      "epoch": 0.8939540078176558,
      "grad_norm": 0.34771645069122314,
      "learning_rate": 1.4040306614548963e-05,
      "loss": 0.0663,
      "step": 17610
    },
    {
      "epoch": 0.8942078278085182,
      "grad_norm": 0.2789120674133301,
      "learning_rate": 1.4038614481276546e-05,
      "loss": 0.0727,
      "step": 17615
    },
    {
      "epoch": 0.8944616477993806,
      "grad_norm": 0.3300907611846924,
      "learning_rate": 1.403692234800413e-05,
      "loss": 0.0609,
      "step": 17620
    },
    {
      "epoch": 0.8947154677902431,
      "grad_norm": 0.4132726490497589,
      "learning_rate": 1.4035230214731715e-05,
      "loss": 0.0677,
      "step": 17625
    },
    {
      "epoch": 0.8949692877811056,
      "grad_norm": 0.47992974519729614,
      "learning_rate": 1.4033538081459297e-05,
      "loss": 0.0826,
      "step": 17630
    },
    {
      "epoch": 0.8952231077719681,
      "grad_norm": 0.3706139028072357,
      "learning_rate": 1.403184594818688e-05,
      "loss": 0.0766,
      "step": 17635
    },
    {
      "epoch": 0.8954769277628306,
      "grad_norm": 0.4228927791118622,
      "learning_rate": 1.4030153814914464e-05,
      "loss": 0.0827,
      "step": 17640
    },
    {
      "epoch": 0.8957307477536931,
      "grad_norm": 0.5431283116340637,
      "learning_rate": 1.4028461681642047e-05,
      "loss": 0.0683,
      "step": 17645
    },
    {
      "epoch": 0.8959845677445556,
      "grad_norm": 0.33838915824890137,
      "learning_rate": 1.402676954836963e-05,
      "loss": 0.058,
      "step": 17650
    },
    {
      "epoch": 0.8962383877354181,
      "grad_norm": 0.3902619481086731,
      "learning_rate": 1.4025077415097215e-05,
      "loss": 0.078,
      "step": 17655
    },
    {
      "epoch": 0.8964922077262806,
      "grad_norm": 0.38549157977104187,
      "learning_rate": 1.4023385281824798e-05,
      "loss": 0.0741,
      "step": 17660
    },
    {
      "epoch": 0.896746027717143,
      "grad_norm": 0.4042127728462219,
      "learning_rate": 1.4021693148552382e-05,
      "loss": 0.0687,
      "step": 17665
    },
    {
      "epoch": 0.8969998477080054,
      "grad_norm": 0.452392041683197,
      "learning_rate": 1.4020001015279965e-05,
      "loss": 0.0739,
      "step": 17670
    },
    {
      "epoch": 0.8972536676988679,
      "grad_norm": 0.45912966132164,
      "learning_rate": 1.4018308882007547e-05,
      "loss": 0.0725,
      "step": 17675
    },
    {
      "epoch": 0.8975074876897304,
      "grad_norm": 0.4095827043056488,
      "learning_rate": 1.4016616748735132e-05,
      "loss": 0.075,
      "step": 17680
    },
    {
      "epoch": 0.8977613076805929,
      "grad_norm": 0.3190585672855377,
      "learning_rate": 1.4014924615462716e-05,
      "loss": 0.0717,
      "step": 17685
    },
    {
      "epoch": 0.8980151276714554,
      "grad_norm": 0.6912304759025574,
      "learning_rate": 1.4013232482190298e-05,
      "loss": 0.0708,
      "step": 17690
    },
    {
      "epoch": 0.8982689476623179,
      "grad_norm": 0.7174093723297119,
      "learning_rate": 1.4011540348917883e-05,
      "loss": 0.069,
      "step": 17695
    },
    {
      "epoch": 0.8985227676531804,
      "grad_norm": 0.30754783749580383,
      "learning_rate": 1.4009848215645465e-05,
      "loss": 0.0618,
      "step": 17700
    },
    {
      "epoch": 0.8987765876440429,
      "grad_norm": 0.7452945113182068,
      "learning_rate": 1.4008156082373048e-05,
      "loss": 0.0766,
      "step": 17705
    },
    {
      "epoch": 0.8990304076349053,
      "grad_norm": 0.3052619695663452,
      "learning_rate": 1.4006463949100633e-05,
      "loss": 0.0701,
      "step": 17710
    },
    {
      "epoch": 0.8992842276257678,
      "grad_norm": 0.6337426900863647,
      "learning_rate": 1.4004771815828215e-05,
      "loss": 0.0775,
      "step": 17715
    },
    {
      "epoch": 0.8995380476166303,
      "grad_norm": 0.35096266865730286,
      "learning_rate": 1.40030796825558e-05,
      "loss": 0.0629,
      "step": 17720
    },
    {
      "epoch": 0.8997918676074927,
      "grad_norm": 0.40677130222320557,
      "learning_rate": 1.4001387549283382e-05,
      "loss": 0.0777,
      "step": 17725
    },
    {
      "epoch": 0.9000456875983552,
      "grad_norm": 0.4063386023044586,
      "learning_rate": 1.3999695416010966e-05,
      "loss": 0.0565,
      "step": 17730
    },
    {
      "epoch": 0.9002995075892177,
      "grad_norm": 0.30389711260795593,
      "learning_rate": 1.399800328273855e-05,
      "loss": 0.0671,
      "step": 17735
    },
    {
      "epoch": 0.9005533275800802,
      "grad_norm": 0.26976916193962097,
      "learning_rate": 1.3996311149466133e-05,
      "loss": 0.0717,
      "step": 17740
    },
    {
      "epoch": 0.9008071475709427,
      "grad_norm": 0.4910055994987488,
      "learning_rate": 1.3994619016193715e-05,
      "loss": 0.065,
      "step": 17745
    },
    {
      "epoch": 0.9010609675618052,
      "grad_norm": 0.3902299702167511,
      "learning_rate": 1.39929268829213e-05,
      "loss": 0.0636,
      "step": 17750
    },
    {
      "epoch": 0.9013147875526677,
      "grad_norm": 0.9779106378555298,
      "learning_rate": 1.3991234749648884e-05,
      "loss": 0.0642,
      "step": 17755
    },
    {
      "epoch": 0.9015686075435301,
      "grad_norm": 0.48715925216674805,
      "learning_rate": 1.3989542616376467e-05,
      "loss": 0.0749,
      "step": 17760
    },
    {
      "epoch": 0.9018224275343926,
      "grad_norm": 0.9629639387130737,
      "learning_rate": 1.398785048310405e-05,
      "loss": 0.0692,
      "step": 17765
    },
    {
      "epoch": 0.9020762475252551,
      "grad_norm": 0.39406028389930725,
      "learning_rate": 1.3986158349831633e-05,
      "loss": 0.0673,
      "step": 17770
    },
    {
      "epoch": 0.9023300675161176,
      "grad_norm": 0.3680139482021332,
      "learning_rate": 1.3984466216559218e-05,
      "loss": 0.0706,
      "step": 17775
    },
    {
      "epoch": 0.90258388750698,
      "grad_norm": 0.3493126928806305,
      "learning_rate": 1.3982774083286801e-05,
      "loss": 0.0711,
      "step": 17780
    },
    {
      "epoch": 0.9028377074978425,
      "grad_norm": 0.35266679525375366,
      "learning_rate": 1.3981081950014383e-05,
      "loss": 0.058,
      "step": 17785
    },
    {
      "epoch": 0.903091527488705,
      "grad_norm": 0.5064133405685425,
      "learning_rate": 1.3979389816741969e-05,
      "loss": 0.09,
      "step": 17790
    },
    {
      "epoch": 0.9033453474795675,
      "grad_norm": 0.6205065250396729,
      "learning_rate": 1.397769768346955e-05,
      "loss": 0.0656,
      "step": 17795
    },
    {
      "epoch": 0.90359916747043,
      "grad_norm": 0.36125099658966064,
      "learning_rate": 1.3976005550197134e-05,
      "loss": 0.0662,
      "step": 17800
    },
    {
      "epoch": 0.9038529874612925,
      "grad_norm": 0.369780570268631,
      "learning_rate": 1.3974313416924719e-05,
      "loss": 0.0625,
      "step": 17805
    },
    {
      "epoch": 0.9041068074521549,
      "grad_norm": 0.3109322786331177,
      "learning_rate": 1.3972621283652301e-05,
      "loss": 0.0709,
      "step": 17810
    },
    {
      "epoch": 0.9043606274430174,
      "grad_norm": 0.44117799401283264,
      "learning_rate": 1.3970929150379886e-05,
      "loss": 0.0566,
      "step": 17815
    },
    {
      "epoch": 0.9046144474338799,
      "grad_norm": 0.41141143441200256,
      "learning_rate": 1.3969237017107468e-05,
      "loss": 0.066,
      "step": 17820
    },
    {
      "epoch": 0.9048682674247424,
      "grad_norm": 0.38892972469329834,
      "learning_rate": 1.3967544883835052e-05,
      "loss": 0.0785,
      "step": 17825
    },
    {
      "epoch": 0.9051220874156048,
      "grad_norm": 0.42113691568374634,
      "learning_rate": 1.3965852750562637e-05,
      "loss": 0.0747,
      "step": 17830
    },
    {
      "epoch": 0.9053759074064673,
      "grad_norm": 0.33473747968673706,
      "learning_rate": 1.3964160617290219e-05,
      "loss": 0.0732,
      "step": 17835
    },
    {
      "epoch": 0.9056297273973298,
      "grad_norm": 0.4112972915172577,
      "learning_rate": 1.3962468484017802e-05,
      "loss": 0.0689,
      "step": 17840
    },
    {
      "epoch": 0.9058835473881923,
      "grad_norm": 0.31149786710739136,
      "learning_rate": 1.3960776350745386e-05,
      "loss": 0.0665,
      "step": 17845
    },
    {
      "epoch": 0.9061373673790548,
      "grad_norm": 0.4009893238544464,
      "learning_rate": 1.395908421747297e-05,
      "loss": 0.0708,
      "step": 17850
    },
    {
      "epoch": 0.9063911873699173,
      "grad_norm": 0.3744799792766571,
      "learning_rate": 1.3957392084200551e-05,
      "loss": 0.0738,
      "step": 17855
    },
    {
      "epoch": 0.9066450073607797,
      "grad_norm": 0.3893011212348938,
      "learning_rate": 1.3955699950928136e-05,
      "loss": 0.0616,
      "step": 17860
    },
    {
      "epoch": 0.9068988273516422,
      "grad_norm": 0.35065850615501404,
      "learning_rate": 1.395400781765572e-05,
      "loss": 0.0651,
      "step": 17865
    },
    {
      "epoch": 0.9071526473425047,
      "grad_norm": 0.5612497925758362,
      "learning_rate": 1.3952315684383304e-05,
      "loss": 0.0835,
      "step": 17870
    },
    {
      "epoch": 0.9074064673333672,
      "grad_norm": 0.39559072256088257,
      "learning_rate": 1.3950623551110887e-05,
      "loss": 0.0682,
      "step": 17875
    },
    {
      "epoch": 0.9076602873242297,
      "grad_norm": 0.35949650406837463,
      "learning_rate": 1.3948931417838469e-05,
      "loss": 0.0785,
      "step": 17880
    },
    {
      "epoch": 0.9079141073150921,
      "grad_norm": 0.5224626064300537,
      "learning_rate": 1.3947239284566054e-05,
      "loss": 0.085,
      "step": 17885
    },
    {
      "epoch": 0.9081679273059546,
      "grad_norm": 0.3599477708339691,
      "learning_rate": 1.3945547151293638e-05,
      "loss": 0.0684,
      "step": 17890
    },
    {
      "epoch": 0.9084217472968171,
      "grad_norm": 0.6261371970176697,
      "learning_rate": 1.394385501802122e-05,
      "loss": 0.068,
      "step": 17895
    },
    {
      "epoch": 0.9086755672876796,
      "grad_norm": 0.3289092183113098,
      "learning_rate": 1.3942162884748805e-05,
      "loss": 0.0606,
      "step": 17900
    },
    {
      "epoch": 0.9089293872785421,
      "grad_norm": 0.39295879006385803,
      "learning_rate": 1.3940470751476387e-05,
      "loss": 0.0604,
      "step": 17905
    },
    {
      "epoch": 0.9091832072694045,
      "grad_norm": 0.40853872895240784,
      "learning_rate": 1.3938778618203972e-05,
      "loss": 0.0763,
      "step": 17910
    },
    {
      "epoch": 0.909437027260267,
      "grad_norm": 0.587341845035553,
      "learning_rate": 1.3937086484931554e-05,
      "loss": 0.0724,
      "step": 17915
    },
    {
      "epoch": 0.9096908472511295,
      "grad_norm": 0.5507720708847046,
      "learning_rate": 1.3935394351659137e-05,
      "loss": 0.0695,
      "step": 17920
    },
    {
      "epoch": 0.909944667241992,
      "grad_norm": 0.31212714314460754,
      "learning_rate": 1.3933702218386723e-05,
      "loss": 0.0757,
      "step": 17925
    },
    {
      "epoch": 0.9101984872328545,
      "grad_norm": 0.31327691674232483,
      "learning_rate": 1.3932010085114304e-05,
      "loss": 0.0721,
      "step": 17930
    },
    {
      "epoch": 0.910452307223717,
      "grad_norm": 0.3152126669883728,
      "learning_rate": 1.3930317951841888e-05,
      "loss": 0.0568,
      "step": 17935
    },
    {
      "epoch": 0.9107061272145794,
      "grad_norm": 0.4510830342769623,
      "learning_rate": 1.3928625818569471e-05,
      "loss": 0.0672,
      "step": 17940
    },
    {
      "epoch": 0.9109599472054419,
      "grad_norm": 0.3216569721698761,
      "learning_rate": 1.3926933685297055e-05,
      "loss": 0.0676,
      "step": 17945
    },
    {
      "epoch": 0.9112137671963044,
      "grad_norm": 0.6665675044059753,
      "learning_rate": 1.3925241552024637e-05,
      "loss": 0.0646,
      "step": 17950
    },
    {
      "epoch": 0.9114675871871669,
      "grad_norm": 0.37302878499031067,
      "learning_rate": 1.3923549418752222e-05,
      "loss": 0.0681,
      "step": 17955
    },
    {
      "epoch": 0.9117214071780293,
      "grad_norm": 0.30719074606895447,
      "learning_rate": 1.3921857285479806e-05,
      "loss": 0.0694,
      "step": 17960
    },
    {
      "epoch": 0.9119752271688918,
      "grad_norm": 0.38624605536460876,
      "learning_rate": 1.392016515220739e-05,
      "loss": 0.0725,
      "step": 17965
    },
    {
      "epoch": 0.9122290471597543,
      "grad_norm": 0.299686461687088,
      "learning_rate": 1.3918473018934973e-05,
      "loss": 0.0604,
      "step": 17970
    },
    {
      "epoch": 0.9124828671506168,
      "grad_norm": 0.38363954424858093,
      "learning_rate": 1.3916780885662555e-05,
      "loss": 0.0683,
      "step": 17975
    },
    {
      "epoch": 0.9127366871414793,
      "grad_norm": 0.3559637665748596,
      "learning_rate": 1.391508875239014e-05,
      "loss": 0.0681,
      "step": 17980
    },
    {
      "epoch": 0.9129905071323418,
      "grad_norm": 0.41254445910453796,
      "learning_rate": 1.3913396619117723e-05,
      "loss": 0.0647,
      "step": 17985
    },
    {
      "epoch": 0.9132443271232042,
      "grad_norm": 0.3159063756465912,
      "learning_rate": 1.3911704485845305e-05,
      "loss": 0.0687,
      "step": 17990
    },
    {
      "epoch": 0.9134981471140667,
      "grad_norm": 0.5465823411941528,
      "learning_rate": 1.391001235257289e-05,
      "loss": 0.0702,
      "step": 17995
    },
    {
      "epoch": 0.9137519671049292,
      "grad_norm": 0.37820661067962646,
      "learning_rate": 1.3908320219300472e-05,
      "loss": 0.0766,
      "step": 18000
    },
    {
      "epoch": 0.9140057870957916,
      "grad_norm": 0.35360532999038696,
      "learning_rate": 1.3906628086028058e-05,
      "loss": 0.0726,
      "step": 18005
    },
    {
      "epoch": 0.9142596070866541,
      "grad_norm": 0.3606135845184326,
      "learning_rate": 1.3904935952755641e-05,
      "loss": 0.0647,
      "step": 18010
    },
    {
      "epoch": 0.9145134270775166,
      "grad_norm": 0.3642440438270569,
      "learning_rate": 1.3903243819483223e-05,
      "loss": 0.0769,
      "step": 18015
    },
    {
      "epoch": 0.9147672470683791,
      "grad_norm": 0.49041110277175903,
      "learning_rate": 1.3901551686210808e-05,
      "loss": 0.0738,
      "step": 18020
    },
    {
      "epoch": 0.9150210670592416,
      "grad_norm": 0.38325151801109314,
      "learning_rate": 1.389985955293839e-05,
      "loss": 0.062,
      "step": 18025
    },
    {
      "epoch": 0.9152748870501041,
      "grad_norm": 0.4725360870361328,
      "learning_rate": 1.3898167419665974e-05,
      "loss": 0.0713,
      "step": 18030
    },
    {
      "epoch": 0.9155287070409666,
      "grad_norm": 0.7474817037582397,
      "learning_rate": 1.3896475286393559e-05,
      "loss": 0.0604,
      "step": 18035
    },
    {
      "epoch": 0.915782527031829,
      "grad_norm": 0.3981219530105591,
      "learning_rate": 1.389478315312114e-05,
      "loss": 0.0713,
      "step": 18040
    },
    {
      "epoch": 0.9160363470226915,
      "grad_norm": 0.32454854249954224,
      "learning_rate": 1.3893091019848724e-05,
      "loss": 0.0672,
      "step": 18045
    },
    {
      "epoch": 0.916290167013554,
      "grad_norm": 0.3997170329093933,
      "learning_rate": 1.3891398886576308e-05,
      "loss": 0.0828,
      "step": 18050
    },
    {
      "epoch": 0.9165439870044164,
      "grad_norm": 0.28583142161369324,
      "learning_rate": 1.3889706753303891e-05,
      "loss": 0.0685,
      "step": 18055
    },
    {
      "epoch": 0.9167978069952789,
      "grad_norm": 0.48778054118156433,
      "learning_rate": 1.3888014620031477e-05,
      "loss": 0.0628,
      "step": 18060
    },
    {
      "epoch": 0.9170516269861414,
      "grad_norm": 0.38761603832244873,
      "learning_rate": 1.3886322486759058e-05,
      "loss": 0.0718,
      "step": 18065
    },
    {
      "epoch": 0.9173054469770039,
      "grad_norm": 0.5895673632621765,
      "learning_rate": 1.3884630353486642e-05,
      "loss": 0.0774,
      "step": 18070
    },
    {
      "epoch": 0.9175592669678664,
      "grad_norm": 0.30503496527671814,
      "learning_rate": 1.3882938220214225e-05,
      "loss": 0.0607,
      "step": 18075
    },
    {
      "epoch": 0.9178130869587289,
      "grad_norm": 0.4478660225868225,
      "learning_rate": 1.3881246086941809e-05,
      "loss": 0.0708,
      "step": 18080
    },
    {
      "epoch": 0.9180669069495914,
      "grad_norm": 0.509867787361145,
      "learning_rate": 1.3879553953669391e-05,
      "loss": 0.0738,
      "step": 18085
    },
    {
      "epoch": 0.9183207269404539,
      "grad_norm": 1.060240387916565,
      "learning_rate": 1.3877861820396976e-05,
      "loss": 0.0762,
      "step": 18090
    },
    {
      "epoch": 0.9185745469313163,
      "grad_norm": 0.31515824794769287,
      "learning_rate": 1.387616968712456e-05,
      "loss": 0.0693,
      "step": 18095
    },
    {
      "epoch": 0.9188283669221788,
      "grad_norm": 0.32389432191848755,
      "learning_rate": 1.3874477553852141e-05,
      "loss": 0.0649,
      "step": 18100
    },
    {
      "epoch": 0.9190821869130412,
      "grad_norm": 0.4039955139160156,
      "learning_rate": 1.3872785420579727e-05,
      "loss": 0.0619,
      "step": 18105
    },
    {
      "epoch": 0.9193360069039037,
      "grad_norm": 0.29883208870887756,
      "learning_rate": 1.3871093287307309e-05,
      "loss": 0.0692,
      "step": 18110
    },
    {
      "epoch": 0.9195898268947662,
      "grad_norm": 0.49272438883781433,
      "learning_rate": 1.3869401154034894e-05,
      "loss": 0.0791,
      "step": 18115
    },
    {
      "epoch": 0.9198436468856287,
      "grad_norm": 0.6225094795227051,
      "learning_rate": 1.3867709020762476e-05,
      "loss": 0.0692,
      "step": 18120
    },
    {
      "epoch": 0.9200974668764912,
      "grad_norm": 0.3257209062576294,
      "learning_rate": 1.386601688749006e-05,
      "loss": 0.0773,
      "step": 18125
    },
    {
      "epoch": 0.9203512868673537,
      "grad_norm": 0.5255199074745178,
      "learning_rate": 1.3864324754217644e-05,
      "loss": 0.0753,
      "step": 18130
    },
    {
      "epoch": 0.9206051068582162,
      "grad_norm": 0.42838171124458313,
      "learning_rate": 1.3862632620945226e-05,
      "loss": 0.0654,
      "step": 18135
    },
    {
      "epoch": 0.9208589268490787,
      "grad_norm": 0.37420716881752014,
      "learning_rate": 1.386094048767281e-05,
      "loss": 0.0669,
      "step": 18140
    },
    {
      "epoch": 0.9211127468399412,
      "grad_norm": 0.4622999429702759,
      "learning_rate": 1.3859248354400393e-05,
      "loss": 0.0679,
      "step": 18145
    },
    {
      "epoch": 0.9213665668308036,
      "grad_norm": 0.4860808849334717,
      "learning_rate": 1.3857556221127977e-05,
      "loss": 0.0785,
      "step": 18150
    },
    {
      "epoch": 0.921620386821666,
      "grad_norm": 0.32169151306152344,
      "learning_rate": 1.3855864087855562e-05,
      "loss": 0.0637,
      "step": 18155
    },
    {
      "epoch": 0.9218742068125285,
      "grad_norm": 0.37703365087509155,
      "learning_rate": 1.3854171954583144e-05,
      "loss": 0.065,
      "step": 18160
    },
    {
      "epoch": 0.922128026803391,
      "grad_norm": 0.3887845277786255,
      "learning_rate": 1.3852479821310728e-05,
      "loss": 0.0599,
      "step": 18165
    },
    {
      "epoch": 0.9223818467942535,
      "grad_norm": 0.35805755853652954,
      "learning_rate": 1.3850787688038311e-05,
      "loss": 0.0732,
      "step": 18170
    },
    {
      "epoch": 0.922635666785116,
      "grad_norm": 0.33947518467903137,
      "learning_rate": 1.3849095554765895e-05,
      "loss": 0.0667,
      "step": 18175
    },
    {
      "epoch": 0.9228894867759785,
      "grad_norm": 0.4389743208885193,
      "learning_rate": 1.3847403421493477e-05,
      "loss": 0.0693,
      "step": 18180
    },
    {
      "epoch": 0.923143306766841,
      "grad_norm": 0.44879239797592163,
      "learning_rate": 1.3845711288221062e-05,
      "loss": 0.0618,
      "step": 18185
    },
    {
      "epoch": 0.9233971267577035,
      "grad_norm": 0.3365269601345062,
      "learning_rate": 1.3844019154948645e-05,
      "loss": 0.0759,
      "step": 18190
    },
    {
      "epoch": 0.923650946748566,
      "grad_norm": 0.24301739037036896,
      "learning_rate": 1.3842327021676227e-05,
      "loss": 0.0658,
      "step": 18195
    },
    {
      "epoch": 0.9239047667394285,
      "grad_norm": 0.324535995721817,
      "learning_rate": 1.3840634888403812e-05,
      "loss": 0.0572,
      "step": 18200
    },
    {
      "epoch": 0.9241585867302908,
      "grad_norm": 0.3125174641609192,
      "learning_rate": 1.3838942755131394e-05,
      "loss": 0.0705,
      "step": 18205
    },
    {
      "epoch": 0.9244124067211533,
      "grad_norm": 0.41187185049057007,
      "learning_rate": 1.383725062185898e-05,
      "loss": 0.0654,
      "step": 18210
    },
    {
      "epoch": 0.9246662267120158,
      "grad_norm": 0.4254433512687683,
      "learning_rate": 1.3835558488586563e-05,
      "loss": 0.0832,
      "step": 18215
    },
    {
      "epoch": 0.9249200467028783,
      "grad_norm": 0.37248459458351135,
      "learning_rate": 1.3833866355314145e-05,
      "loss": 0.0635,
      "step": 18220
    },
    {
      "epoch": 0.9251738666937408,
      "grad_norm": 0.34162911772727966,
      "learning_rate": 1.383217422204173e-05,
      "loss": 0.0626,
      "step": 18225
    },
    {
      "epoch": 0.9254276866846033,
      "grad_norm": 0.4066483676433563,
      "learning_rate": 1.3830482088769312e-05,
      "loss": 0.0745,
      "step": 18230
    },
    {
      "epoch": 0.9256815066754658,
      "grad_norm": 0.600093424320221,
      "learning_rate": 1.3828789955496896e-05,
      "loss": 0.0673,
      "step": 18235
    },
    {
      "epoch": 0.9259353266663283,
      "grad_norm": 0.47302451729774475,
      "learning_rate": 1.382709782222448e-05,
      "loss": 0.0653,
      "step": 18240
    },
    {
      "epoch": 0.9261891466571908,
      "grad_norm": 0.41447651386260986,
      "learning_rate": 1.3825405688952063e-05,
      "loss": 0.0709,
      "step": 18245
    },
    {
      "epoch": 0.9264429666480531,
      "grad_norm": 0.31857216358184814,
      "learning_rate": 1.3823713555679648e-05,
      "loss": 0.0702,
      "step": 18250
    },
    {
      "epoch": 0.9266967866389156,
      "grad_norm": 0.39774346351623535,
      "learning_rate": 1.382202142240723e-05,
      "loss": 0.0757,
      "step": 18255
    },
    {
      "epoch": 0.9269506066297781,
      "grad_norm": 0.3512285053730011,
      "learning_rate": 1.3820329289134813e-05,
      "loss": 0.0664,
      "step": 18260
    },
    {
      "epoch": 0.9272044266206406,
      "grad_norm": 0.42229223251342773,
      "learning_rate": 1.3818637155862398e-05,
      "loss": 0.0662,
      "step": 18265
    },
    {
      "epoch": 0.9274582466115031,
      "grad_norm": 0.4002987742424011,
      "learning_rate": 1.381694502258998e-05,
      "loss": 0.0679,
      "step": 18270
    },
    {
      "epoch": 0.9277120666023656,
      "grad_norm": 0.4806728661060333,
      "learning_rate": 1.3815252889317564e-05,
      "loss": 0.0645,
      "step": 18275
    },
    {
      "epoch": 0.9279658865932281,
      "grad_norm": 0.29817402362823486,
      "learning_rate": 1.3813560756045147e-05,
      "loss": 0.0686,
      "step": 18280
    },
    {
      "epoch": 0.9282197065840906,
      "grad_norm": 0.4930202066898346,
      "learning_rate": 1.3811868622772731e-05,
      "loss": 0.0744,
      "step": 18285
    },
    {
      "epoch": 0.9284735265749531,
      "grad_norm": 0.4029638171195984,
      "learning_rate": 1.3810176489500313e-05,
      "loss": 0.0637,
      "step": 18290
    },
    {
      "epoch": 0.9287273465658156,
      "grad_norm": 0.45114755630493164,
      "learning_rate": 1.3808484356227898e-05,
      "loss": 0.0668,
      "step": 18295
    },
    {
      "epoch": 0.928981166556678,
      "grad_norm": 0.3982153832912445,
      "learning_rate": 1.380679222295548e-05,
      "loss": 0.0613,
      "step": 18300
    },
    {
      "epoch": 0.9292349865475404,
      "grad_norm": 0.3455192446708679,
      "learning_rate": 1.3805100089683065e-05,
      "loss": 0.0707,
      "step": 18305
    },
    {
      "epoch": 0.9294888065384029,
      "grad_norm": 0.3606429398059845,
      "learning_rate": 1.3803407956410649e-05,
      "loss": 0.0708,
      "step": 18310
    },
    {
      "epoch": 0.9297426265292654,
      "grad_norm": 0.45667099952697754,
      "learning_rate": 1.380171582313823e-05,
      "loss": 0.0642,
      "step": 18315
    },
    {
      "epoch": 0.9299964465201279,
      "grad_norm": 0.38719412684440613,
      "learning_rate": 1.3800023689865816e-05,
      "loss": 0.0648,
      "step": 18320
    },
    {
      "epoch": 0.9302502665109904,
      "grad_norm": 0.5505213737487793,
      "learning_rate": 1.3798331556593398e-05,
      "loss": 0.0653,
      "step": 18325
    },
    {
      "epoch": 0.9305040865018529,
      "grad_norm": 0.5225023627281189,
      "learning_rate": 1.3796639423320981e-05,
      "loss": 0.0715,
      "step": 18330
    },
    {
      "epoch": 0.9307579064927154,
      "grad_norm": 0.2744523286819458,
      "learning_rate": 1.3794947290048566e-05,
      "loss": 0.0596,
      "step": 18335
    },
    {
      "epoch": 0.9310117264835779,
      "grad_norm": 0.4220918118953705,
      "learning_rate": 1.3793255156776148e-05,
      "loss": 0.0703,
      "step": 18340
    },
    {
      "epoch": 0.9312655464744404,
      "grad_norm": 0.4151606261730194,
      "learning_rate": 1.3791563023503732e-05,
      "loss": 0.0629,
      "step": 18345
    },
    {
      "epoch": 0.9315193664653028,
      "grad_norm": 0.43558046221733093,
      "learning_rate": 1.3789870890231315e-05,
      "loss": 0.0554,
      "step": 18350
    },
    {
      "epoch": 0.9317731864561652,
      "grad_norm": 0.3388653099536896,
      "learning_rate": 1.3788178756958899e-05,
      "loss": 0.0702,
      "step": 18355
    },
    {
      "epoch": 0.9320270064470277,
      "grad_norm": 0.5925946235656738,
      "learning_rate": 1.3786486623686484e-05,
      "loss": 0.0682,
      "step": 18360
    },
    {
      "epoch": 0.9322808264378902,
      "grad_norm": 1.0079941749572754,
      "learning_rate": 1.3784794490414066e-05,
      "loss": 0.0663,
      "step": 18365
    },
    {
      "epoch": 0.9325346464287527,
      "grad_norm": 0.38128307461738586,
      "learning_rate": 1.378310235714165e-05,
      "loss": 0.0671,
      "step": 18370
    },
    {
      "epoch": 0.9327884664196152,
      "grad_norm": 0.3320787847042084,
      "learning_rate": 1.3781410223869233e-05,
      "loss": 0.0654,
      "step": 18375
    },
    {
      "epoch": 0.9330422864104777,
      "grad_norm": 0.2928427755832672,
      "learning_rate": 1.3779718090596817e-05,
      "loss": 0.0761,
      "step": 18380
    },
    {
      "epoch": 0.9332961064013402,
      "grad_norm": 0.4131205081939697,
      "learning_rate": 1.3778025957324398e-05,
      "loss": 0.0728,
      "step": 18385
    },
    {
      "epoch": 0.9335499263922027,
      "grad_norm": 0.38338300585746765,
      "learning_rate": 1.3776333824051984e-05,
      "loss": 0.0645,
      "step": 18390
    },
    {
      "epoch": 0.9338037463830652,
      "grad_norm": 0.387997031211853,
      "learning_rate": 1.3774641690779567e-05,
      "loss": 0.0662,
      "step": 18395
    },
    {
      "epoch": 0.9340575663739276,
      "grad_norm": 0.384552925825119,
      "learning_rate": 1.377294955750715e-05,
      "loss": 0.0722,
      "step": 18400
    },
    {
      "epoch": 0.93431138636479,
      "grad_norm": 0.36367639899253845,
      "learning_rate": 1.3771257424234734e-05,
      "loss": 0.0649,
      "step": 18405
    },
    {
      "epoch": 0.9345652063556525,
      "grad_norm": 0.7512829303741455,
      "learning_rate": 1.3769565290962316e-05,
      "loss": 0.0625,
      "step": 18410
    },
    {
      "epoch": 0.934819026346515,
      "grad_norm": 0.38222187757492065,
      "learning_rate": 1.3767873157689901e-05,
      "loss": 0.072,
      "step": 18415
    },
    {
      "epoch": 0.9350728463373775,
      "grad_norm": 0.6945295929908752,
      "learning_rate": 1.3766181024417485e-05,
      "loss": 0.0635,
      "step": 18420
    },
    {
      "epoch": 0.93532666632824,
      "grad_norm": 0.4692384600639343,
      "learning_rate": 1.3764488891145067e-05,
      "loss": 0.0697,
      "step": 18425
    },
    {
      "epoch": 0.9355804863191025,
      "grad_norm": 0.3983794152736664,
      "learning_rate": 1.3762796757872652e-05,
      "loss": 0.0651,
      "step": 18430
    },
    {
      "epoch": 0.935834306309965,
      "grad_norm": 0.450428307056427,
      "learning_rate": 1.3761104624600234e-05,
      "loss": 0.0644,
      "step": 18435
    },
    {
      "epoch": 0.9360881263008275,
      "grad_norm": 0.26266250014305115,
      "learning_rate": 1.3759412491327817e-05,
      "loss": 0.0562,
      "step": 18440
    },
    {
      "epoch": 0.93634194629169,
      "grad_norm": 0.4074264168739319,
      "learning_rate": 1.3757720358055403e-05,
      "loss": 0.0627,
      "step": 18445
    },
    {
      "epoch": 0.9365957662825524,
      "grad_norm": 0.41876572370529175,
      "learning_rate": 1.3756028224782985e-05,
      "loss": 0.0605,
      "step": 18450
    },
    {
      "epoch": 0.9368495862734149,
      "grad_norm": 0.38248318433761597,
      "learning_rate": 1.375433609151057e-05,
      "loss": 0.0615,
      "step": 18455
    },
    {
      "epoch": 0.9371034062642773,
      "grad_norm": 0.38324958086013794,
      "learning_rate": 1.3752643958238152e-05,
      "loss": 0.0664,
      "step": 18460
    },
    {
      "epoch": 0.9373572262551398,
      "grad_norm": 0.5330802202224731,
      "learning_rate": 1.3750951824965735e-05,
      "loss": 0.0593,
      "step": 18465
    },
    {
      "epoch": 0.9376110462460023,
      "grad_norm": 1.0437511205673218,
      "learning_rate": 1.374925969169332e-05,
      "loss": 0.064,
      "step": 18470
    },
    {
      "epoch": 0.9378648662368648,
      "grad_norm": 0.9003655910491943,
      "learning_rate": 1.3747567558420902e-05,
      "loss": 0.0683,
      "step": 18475
    },
    {
      "epoch": 0.9381186862277273,
      "grad_norm": 0.32290807366371155,
      "learning_rate": 1.3745875425148484e-05,
      "loss": 0.068,
      "step": 18480
    },
    {
      "epoch": 0.9383725062185898,
      "grad_norm": 0.6567671895027161,
      "learning_rate": 1.374418329187607e-05,
      "loss": 0.0661,
      "step": 18485
    },
    {
      "epoch": 0.9386263262094523,
      "grad_norm": 0.35361889004707336,
      "learning_rate": 1.3742491158603653e-05,
      "loss": 0.0689,
      "step": 18490
    },
    {
      "epoch": 0.9388801462003148,
      "grad_norm": 0.3336738049983978,
      "learning_rate": 1.3740799025331235e-05,
      "loss": 0.0594,
      "step": 18495
    },
    {
      "epoch": 0.9391339661911772,
      "grad_norm": 0.3773046135902405,
      "learning_rate": 1.373910689205882e-05,
      "loss": 0.0704,
      "step": 18500
    },
    {
      "epoch": 0.9393877861820397,
      "grad_norm": 0.3520534038543701,
      "learning_rate": 1.3737414758786402e-05,
      "loss": 0.0704,
      "step": 18505
    },
    {
      "epoch": 0.9396416061729022,
      "grad_norm": 0.3274901807308197,
      "learning_rate": 1.3735722625513987e-05,
      "loss": 0.071,
      "step": 18510
    },
    {
      "epoch": 0.9398954261637646,
      "grad_norm": 0.25946927070617676,
      "learning_rate": 1.373403049224157e-05,
      "loss": 0.0698,
      "step": 18515
    },
    {
      "epoch": 0.9401492461546271,
      "grad_norm": 0.5171555876731873,
      "learning_rate": 1.3732338358969152e-05,
      "loss": 0.0553,
      "step": 18520
    },
    {
      "epoch": 0.9404030661454896,
      "grad_norm": 0.32266131043434143,
      "learning_rate": 1.3730646225696738e-05,
      "loss": 0.0538,
      "step": 18525
    },
    {
      "epoch": 0.9406568861363521,
      "grad_norm": 0.43915310502052307,
      "learning_rate": 1.372895409242432e-05,
      "loss": 0.0583,
      "step": 18530
    },
    {
      "epoch": 0.9409107061272146,
      "grad_norm": 0.5458141565322876,
      "learning_rate": 1.3727261959151903e-05,
      "loss": 0.0679,
      "step": 18535
    },
    {
      "epoch": 0.9411645261180771,
      "grad_norm": 0.33626362681388855,
      "learning_rate": 1.3725569825879488e-05,
      "loss": 0.0617,
      "step": 18540
    },
    {
      "epoch": 0.9414183461089395,
      "grad_norm": 0.3278324007987976,
      "learning_rate": 1.372387769260707e-05,
      "loss": 0.0599,
      "step": 18545
    },
    {
      "epoch": 0.941672166099802,
      "grad_norm": 0.3287067115306854,
      "learning_rate": 1.3722185559334655e-05,
      "loss": 0.0635,
      "step": 18550
    },
    {
      "epoch": 0.9419259860906645,
      "grad_norm": 0.3660200834274292,
      "learning_rate": 1.3720493426062237e-05,
      "loss": 0.055,
      "step": 18555
    },
    {
      "epoch": 0.942179806081527,
      "grad_norm": 0.41835132241249084,
      "learning_rate": 1.371880129278982e-05,
      "loss": 0.0736,
      "step": 18560
    },
    {
      "epoch": 0.9424336260723895,
      "grad_norm": 0.44712701439857483,
      "learning_rate": 1.3717109159517406e-05,
      "loss": 0.0607,
      "step": 18565
    },
    {
      "epoch": 0.9426874460632519,
      "grad_norm": 0.33872178196907043,
      "learning_rate": 1.3715417026244988e-05,
      "loss": 0.0619,
      "step": 18570
    },
    {
      "epoch": 0.9429412660541144,
      "grad_norm": 0.6011540293693542,
      "learning_rate": 1.3713724892972571e-05,
      "loss": 0.0646,
      "step": 18575
    },
    {
      "epoch": 0.9431950860449769,
      "grad_norm": 0.5028777122497559,
      "learning_rate": 1.3712032759700155e-05,
      "loss": 0.0685,
      "step": 18580
    },
    {
      "epoch": 0.9434489060358394,
      "grad_norm": 0.595824122428894,
      "learning_rate": 1.3710340626427739e-05,
      "loss": 0.0637,
      "step": 18585
    },
    {
      "epoch": 0.9437027260267019,
      "grad_norm": 0.33471375703811646,
      "learning_rate": 1.370864849315532e-05,
      "loss": 0.0629,
      "step": 18590
    },
    {
      "epoch": 0.9439565460175643,
      "grad_norm": 0.3110968768596649,
      "learning_rate": 1.3706956359882906e-05,
      "loss": 0.0587,
      "step": 18595
    },
    {
      "epoch": 0.9442103660084268,
      "grad_norm": 0.303251713514328,
      "learning_rate": 1.370526422661049e-05,
      "loss": 0.0684,
      "step": 18600
    },
    {
      "epoch": 0.9444641859992893,
      "grad_norm": 0.5813922882080078,
      "learning_rate": 1.3703572093338073e-05,
      "loss": 0.0623,
      "step": 18605
    },
    {
      "epoch": 0.9447180059901518,
      "grad_norm": 0.4020741283893585,
      "learning_rate": 1.3701879960065656e-05,
      "loss": 0.065,
      "step": 18610
    },
    {
      "epoch": 0.9449718259810143,
      "grad_norm": 0.504642903804779,
      "learning_rate": 1.3700187826793238e-05,
      "loss": 0.0552,
      "step": 18615
    },
    {
      "epoch": 0.9452256459718767,
      "grad_norm": 0.4510534703731537,
      "learning_rate": 1.3698495693520823e-05,
      "loss": 0.0653,
      "step": 18620
    },
    {
      "epoch": 0.9454794659627392,
      "grad_norm": 0.40231838822364807,
      "learning_rate": 1.3696803560248407e-05,
      "loss": 0.0604,
      "step": 18625
    },
    {
      "epoch": 0.9457332859536017,
      "grad_norm": 0.303325891494751,
      "learning_rate": 1.3695111426975989e-05,
      "loss": 0.0552,
      "step": 18630
    },
    {
      "epoch": 0.9459871059444642,
      "grad_norm": 0.2996688187122345,
      "learning_rate": 1.3693419293703574e-05,
      "loss": 0.0658,
      "step": 18635
    },
    {
      "epoch": 0.9462409259353267,
      "grad_norm": 0.57343989610672,
      "learning_rate": 1.3691727160431156e-05,
      "loss": 0.0678,
      "step": 18640
    },
    {
      "epoch": 0.9464947459261891,
      "grad_norm": 0.36376577615737915,
      "learning_rate": 1.3690035027158741e-05,
      "loss": 0.072,
      "step": 18645
    },
    {
      "epoch": 0.9467485659170516,
      "grad_norm": 0.4058065414428711,
      "learning_rate": 1.3688342893886325e-05,
      "loss": 0.0642,
      "step": 18650
    },
    {
      "epoch": 0.9470023859079141,
      "grad_norm": 0.5225498676300049,
      "learning_rate": 1.3686650760613906e-05,
      "loss": 0.0687,
      "step": 18655
    },
    {
      "epoch": 0.9472562058987766,
      "grad_norm": 0.3555619716644287,
      "learning_rate": 1.3684958627341492e-05,
      "loss": 0.0575,
      "step": 18660
    },
    {
      "epoch": 0.9475100258896391,
      "grad_norm": 0.32786718010902405,
      "learning_rate": 1.3683266494069074e-05,
      "loss": 0.0603,
      "step": 18665
    },
    {
      "epoch": 0.9477638458805016,
      "grad_norm": 0.497963547706604,
      "learning_rate": 1.3681574360796657e-05,
      "loss": 0.0707,
      "step": 18670
    },
    {
      "epoch": 0.948017665871364,
      "grad_norm": 0.3656896650791168,
      "learning_rate": 1.3679882227524242e-05,
      "loss": 0.0636,
      "step": 18675
    },
    {
      "epoch": 0.9482714858622265,
      "grad_norm": 0.38868141174316406,
      "learning_rate": 1.3678190094251824e-05,
      "loss": 0.0619,
      "step": 18680
    },
    {
      "epoch": 0.948525305853089,
      "grad_norm": 0.4414658844470978,
      "learning_rate": 1.3676497960979406e-05,
      "loss": 0.0726,
      "step": 18685
    },
    {
      "epoch": 0.9487791258439515,
      "grad_norm": 0.38483738899230957,
      "learning_rate": 1.3674805827706991e-05,
      "loss": 0.068,
      "step": 18690
    },
    {
      "epoch": 0.9490329458348139,
      "grad_norm": 0.33117255568504333,
      "learning_rate": 1.3673113694434575e-05,
      "loss": 0.0637,
      "step": 18695
    },
    {
      "epoch": 0.9492867658256764,
      "grad_norm": 0.633201539516449,
      "learning_rate": 1.367142156116216e-05,
      "loss": 0.0634,
      "step": 18700
    },
    {
      "epoch": 0.9495405858165389,
      "grad_norm": 0.4020320177078247,
      "learning_rate": 1.3669729427889742e-05,
      "loss": 0.0544,
      "step": 18705
    },
    {
      "epoch": 0.9497944058074014,
      "grad_norm": 0.42026662826538086,
      "learning_rate": 1.3668037294617324e-05,
      "loss": 0.0612,
      "step": 18710
    },
    {
      "epoch": 0.9500482257982639,
      "grad_norm": 2.030712842941284,
      "learning_rate": 1.3666345161344909e-05,
      "loss": 0.0656,
      "step": 18715
    },
    {
      "epoch": 0.9503020457891264,
      "grad_norm": 0.44346386194229126,
      "learning_rate": 1.3664653028072493e-05,
      "loss": 0.0668,
      "step": 18720
    },
    {
      "epoch": 0.9505558657799889,
      "grad_norm": 0.41186749935150146,
      "learning_rate": 1.3662960894800074e-05,
      "loss": 0.0753,
      "step": 18725
    },
    {
      "epoch": 0.9508096857708513,
      "grad_norm": 0.3317037522792816,
      "learning_rate": 1.366126876152766e-05,
      "loss": 0.0677,
      "step": 18730
    },
    {
      "epoch": 0.9510635057617138,
      "grad_norm": 0.3373609781265259,
      "learning_rate": 1.3659576628255241e-05,
      "loss": 0.0704,
      "step": 18735
    },
    {
      "epoch": 0.9513173257525763,
      "grad_norm": 0.3054006099700928,
      "learning_rate": 1.3657884494982825e-05,
      "loss": 0.0564,
      "step": 18740
    },
    {
      "epoch": 0.9515711457434387,
      "grad_norm": 0.3507018983364105,
      "learning_rate": 1.365619236171041e-05,
      "loss": 0.0682,
      "step": 18745
    },
    {
      "epoch": 0.9518249657343012,
      "grad_norm": 0.501518189907074,
      "learning_rate": 1.3654500228437992e-05,
      "loss": 0.0721,
      "step": 18750
    },
    {
      "epoch": 0.9520787857251637,
      "grad_norm": 0.3085997998714447,
      "learning_rate": 1.3652808095165577e-05,
      "loss": 0.06,
      "step": 18755
    },
    {
      "epoch": 0.9523326057160262,
      "grad_norm": 0.700425386428833,
      "learning_rate": 1.365111596189316e-05,
      "loss": 0.0705,
      "step": 18760
    },
    {
      "epoch": 0.9525864257068887,
      "grad_norm": 0.3953579068183899,
      "learning_rate": 1.3649423828620743e-05,
      "loss": 0.0609,
      "step": 18765
    },
    {
      "epoch": 0.9528402456977512,
      "grad_norm": 0.48132267594337463,
      "learning_rate": 1.3647731695348328e-05,
      "loss": 0.0641,
      "step": 18770
    },
    {
      "epoch": 0.9530940656886137,
      "grad_norm": 0.38885220885276794,
      "learning_rate": 1.364603956207591e-05,
      "loss": 0.0662,
      "step": 18775
    },
    {
      "epoch": 0.9533478856794761,
      "grad_norm": 0.30875876545906067,
      "learning_rate": 1.3644347428803493e-05,
      "loss": 0.0597,
      "step": 18780
    },
    {
      "epoch": 0.9536017056703386,
      "grad_norm": 0.324533611536026,
      "learning_rate": 1.3642655295531077e-05,
      "loss": 0.0548,
      "step": 18785
    },
    {
      "epoch": 0.9538555256612011,
      "grad_norm": 0.33613696694374084,
      "learning_rate": 1.364096316225866e-05,
      "loss": 0.0601,
      "step": 18790
    },
    {
      "epoch": 0.9541093456520635,
      "grad_norm": 0.5050367712974548,
      "learning_rate": 1.3639271028986246e-05,
      "loss": 0.0637,
      "step": 18795
    },
    {
      "epoch": 0.954363165642926,
      "grad_norm": 0.2775934338569641,
      "learning_rate": 1.3637578895713828e-05,
      "loss": 0.072,
      "step": 18800
    },
    {
      "epoch": 0.9546169856337885,
      "grad_norm": 0.33112195134162903,
      "learning_rate": 1.3635886762441411e-05,
      "loss": 0.0624,
      "step": 18805
    },
    {
      "epoch": 0.954870805624651,
      "grad_norm": 0.4370911121368408,
      "learning_rate": 1.3634194629168995e-05,
      "loss": 0.0684,
      "step": 18810
    },
    {
      "epoch": 0.9551246256155135,
      "grad_norm": 0.2732505798339844,
      "learning_rate": 1.3632502495896578e-05,
      "loss": 0.0636,
      "step": 18815
    },
    {
      "epoch": 0.955378445606376,
      "grad_norm": 0.45320791006088257,
      "learning_rate": 1.363081036262416e-05,
      "loss": 0.0745,
      "step": 18820
    },
    {
      "epoch": 0.9556322655972385,
      "grad_norm": 0.44382214546203613,
      "learning_rate": 1.3629118229351745e-05,
      "loss": 0.0564,
      "step": 18825
    },
    {
      "epoch": 0.955886085588101,
      "grad_norm": 0.33199283480644226,
      "learning_rate": 1.3627426096079329e-05,
      "loss": 0.069,
      "step": 18830
    },
    {
      "epoch": 0.9561399055789634,
      "grad_norm": 0.48582470417022705,
      "learning_rate": 1.362573396280691e-05,
      "loss": 0.0689,
      "step": 18835
    },
    {
      "epoch": 0.9563937255698258,
      "grad_norm": 0.4178053140640259,
      "learning_rate": 1.3624041829534496e-05,
      "loss": 0.0692,
      "step": 18840
    },
    {
      "epoch": 0.9566475455606883,
      "grad_norm": 0.39388084411621094,
      "learning_rate": 1.3622349696262078e-05,
      "loss": 0.0658,
      "step": 18845
    },
    {
      "epoch": 0.9569013655515508,
      "grad_norm": 0.3513961136341095,
      "learning_rate": 1.3620657562989663e-05,
      "loss": 0.0729,
      "step": 18850
    },
    {
      "epoch": 0.9571551855424133,
      "grad_norm": 0.31259411573410034,
      "learning_rate": 1.3618965429717247e-05,
      "loss": 0.0679,
      "step": 18855
    },
    {
      "epoch": 0.9574090055332758,
      "grad_norm": 0.5756545066833496,
      "learning_rate": 1.3617273296444828e-05,
      "loss": 0.0665,
      "step": 18860
    },
    {
      "epoch": 0.9576628255241383,
      "grad_norm": 0.5503848791122437,
      "learning_rate": 1.3615581163172414e-05,
      "loss": 0.0535,
      "step": 18865
    },
    {
      "epoch": 0.9579166455150008,
      "grad_norm": 0.3748461604118347,
      "learning_rate": 1.3613889029899996e-05,
      "loss": 0.07,
      "step": 18870
    },
    {
      "epoch": 0.9581704655058633,
      "grad_norm": 0.25519129633903503,
      "learning_rate": 1.3612196896627579e-05,
      "loss": 0.0668,
      "step": 18875
    },
    {
      "epoch": 0.9584242854967258,
      "grad_norm": 0.46383601427078247,
      "learning_rate": 1.3610504763355164e-05,
      "loss": 0.0691,
      "step": 18880
    },
    {
      "epoch": 0.9586781054875883,
      "grad_norm": 0.4589533805847168,
      "learning_rate": 1.3608812630082746e-05,
      "loss": 0.0734,
      "step": 18885
    },
    {
      "epoch": 0.9589319254784506,
      "grad_norm": 0.9147213697433472,
      "learning_rate": 1.3607120496810331e-05,
      "loss": 0.0678,
      "step": 18890
    },
    {
      "epoch": 0.9591857454693131,
      "grad_norm": 0.37536412477493286,
      "learning_rate": 1.3605428363537913e-05,
      "loss": 0.0653,
      "step": 18895
    },
    {
      "epoch": 0.9594395654601756,
      "grad_norm": 0.7372077703475952,
      "learning_rate": 1.3603736230265497e-05,
      "loss": 0.0673,
      "step": 18900
    },
    {
      "epoch": 0.9596933854510381,
      "grad_norm": 0.4121319651603699,
      "learning_rate": 1.360204409699308e-05,
      "loss": 0.0643,
      "step": 18905
    },
    {
      "epoch": 0.9599472054419006,
      "grad_norm": 0.2821780741214752,
      "learning_rate": 1.3600351963720664e-05,
      "loss": 0.0634,
      "step": 18910
    },
    {
      "epoch": 0.9602010254327631,
      "grad_norm": 0.2763786017894745,
      "learning_rate": 1.3598659830448246e-05,
      "loss": 0.0655,
      "step": 18915
    },
    {
      "epoch": 0.9604548454236256,
      "grad_norm": 0.3872879445552826,
      "learning_rate": 1.3596967697175831e-05,
      "loss": 0.0695,
      "step": 18920
    },
    {
      "epoch": 0.9607086654144881,
      "grad_norm": 0.3432711660861969,
      "learning_rate": 1.3595275563903414e-05,
      "loss": 0.0618,
      "step": 18925
    },
    {
      "epoch": 0.9609624854053506,
      "grad_norm": 0.40918609499931335,
      "learning_rate": 1.3593583430630996e-05,
      "loss": 0.0663,
      "step": 18930
    },
    {
      "epoch": 0.961216305396213,
      "grad_norm": 0.2839658558368683,
      "learning_rate": 1.3591891297358582e-05,
      "loss": 0.0609,
      "step": 18935
    },
    {
      "epoch": 0.9614701253870754,
      "grad_norm": 0.45697519183158875,
      "learning_rate": 1.3590199164086163e-05,
      "loss": 0.0682,
      "step": 18940
    },
    {
      "epoch": 0.9617239453779379,
      "grad_norm": 0.4115654528141022,
      "learning_rate": 1.3588507030813749e-05,
      "loss": 0.0755,
      "step": 18945
    },
    {
      "epoch": 0.9619777653688004,
      "grad_norm": 0.3328510820865631,
      "learning_rate": 1.3586814897541332e-05,
      "loss": 0.0726,
      "step": 18950
    },
    {
      "epoch": 0.9622315853596629,
      "grad_norm": 0.392888605594635,
      "learning_rate": 1.3585122764268914e-05,
      "loss": 0.0692,
      "step": 18955
    },
    {
      "epoch": 0.9624854053505254,
      "grad_norm": 0.7154536843299866,
      "learning_rate": 1.35834306309965e-05,
      "loss": 0.0709,
      "step": 18960
    },
    {
      "epoch": 0.9627392253413879,
      "grad_norm": 0.38823920488357544,
      "learning_rate": 1.3581738497724081e-05,
      "loss": 0.0581,
      "step": 18965
    },
    {
      "epoch": 0.9629930453322504,
      "grad_norm": 1.6257457733154297,
      "learning_rate": 1.3580046364451665e-05,
      "loss": 0.0627,
      "step": 18970
    },
    {
      "epoch": 0.9632468653231129,
      "grad_norm": 0.3897399604320526,
      "learning_rate": 1.357835423117925e-05,
      "loss": 0.0612,
      "step": 18975
    },
    {
      "epoch": 0.9635006853139754,
      "grad_norm": 0.33613255620002747,
      "learning_rate": 1.3576662097906832e-05,
      "loss": 0.0535,
      "step": 18980
    },
    {
      "epoch": 0.9637545053048379,
      "grad_norm": 0.3078048527240753,
      "learning_rate": 1.3574969964634415e-05,
      "loss": 0.0664,
      "step": 18985
    },
    {
      "epoch": 0.9640083252957002,
      "grad_norm": 0.8129367828369141,
      "learning_rate": 1.3573277831361999e-05,
      "loss": 0.0686,
      "step": 18990
    },
    {
      "epoch": 0.9642621452865627,
      "grad_norm": 0.28624656796455383,
      "learning_rate": 1.3571585698089582e-05,
      "loss": 0.0629,
      "step": 18995
    },
    {
      "epoch": 0.9645159652774252,
      "grad_norm": 0.40177884697914124,
      "learning_rate": 1.3569893564817168e-05,
      "loss": 0.0631,
      "step": 19000
    },
    {
      "epoch": 0.9647697852682877,
      "grad_norm": 0.8333194255828857,
      "learning_rate": 1.356820143154475e-05,
      "loss": 0.0636,
      "step": 19005
    },
    {
      "epoch": 0.9650236052591502,
      "grad_norm": 0.6080228090286255,
      "learning_rate": 1.3566509298272333e-05,
      "loss": 0.0579,
      "step": 19010
    },
    {
      "epoch": 0.9652774252500127,
      "grad_norm": 0.3721805512905121,
      "learning_rate": 1.3564817164999917e-05,
      "loss": 0.0657,
      "step": 19015
    },
    {
      "epoch": 0.9655312452408752,
      "grad_norm": 0.36625728011131287,
      "learning_rate": 1.35631250317275e-05,
      "loss": 0.065,
      "step": 19020
    },
    {
      "epoch": 0.9657850652317377,
      "grad_norm": 0.4012068510055542,
      "learning_rate": 1.3561432898455082e-05,
      "loss": 0.0647,
      "step": 19025
    },
    {
      "epoch": 0.9660388852226002,
      "grad_norm": 0.32912421226501465,
      "learning_rate": 1.3559740765182667e-05,
      "loss": 0.0632,
      "step": 19030
    },
    {
      "epoch": 0.9662927052134627,
      "grad_norm": 0.6822962164878845,
      "learning_rate": 1.355804863191025e-05,
      "loss": 0.0625,
      "step": 19035
    },
    {
      "epoch": 0.966546525204325,
      "grad_norm": 0.3175319731235504,
      "learning_rate": 1.3556356498637834e-05,
      "loss": 0.068,
      "step": 19040
    },
    {
      "epoch": 0.9668003451951875,
      "grad_norm": 0.4144521653652191,
      "learning_rate": 1.3554664365365418e-05,
      "loss": 0.0693,
      "step": 19045
    },
    {
      "epoch": 0.96705416518605,
      "grad_norm": 0.42917636036872864,
      "learning_rate": 1.3552972232093e-05,
      "loss": 0.0584,
      "step": 19050
    },
    {
      "epoch": 0.9673079851769125,
      "grad_norm": 0.3734082877635956,
      "learning_rate": 1.3551280098820585e-05,
      "loss": 0.0753,
      "step": 19055
    },
    {
      "epoch": 0.967561805167775,
      "grad_norm": 0.6813828945159912,
      "learning_rate": 1.3549587965548168e-05,
      "loss": 0.0681,
      "step": 19060
    },
    {
      "epoch": 0.9678156251586375,
      "grad_norm": 0.5187677145004272,
      "learning_rate": 1.354789583227575e-05,
      "loss": 0.063,
      "step": 19065
    },
    {
      "epoch": 0.9680694451495,
      "grad_norm": 0.37740230560302734,
      "learning_rate": 1.3546203699003336e-05,
      "loss": 0.0598,
      "step": 19070
    },
    {
      "epoch": 0.9683232651403625,
      "grad_norm": 0.35848960280418396,
      "learning_rate": 1.3544511565730917e-05,
      "loss": 0.0598,
      "step": 19075
    },
    {
      "epoch": 0.968577085131225,
      "grad_norm": 0.36278432607650757,
      "learning_rate": 1.3542819432458501e-05,
      "loss": 0.0645,
      "step": 19080
    },
    {
      "epoch": 0.9688309051220874,
      "grad_norm": 0.37709158658981323,
      "learning_rate": 1.3541127299186085e-05,
      "loss": 0.0654,
      "step": 19085
    },
    {
      "epoch": 0.9690847251129499,
      "grad_norm": 0.41565874218940735,
      "learning_rate": 1.3539435165913668e-05,
      "loss": 0.067,
      "step": 19090
    },
    {
      "epoch": 0.9693385451038123,
      "grad_norm": 0.405154287815094,
      "learning_rate": 1.3537743032641253e-05,
      "loss": 0.06,
      "step": 19095
    },
    {
      "epoch": 0.9695923650946748,
      "grad_norm": 0.4619700610637665,
      "learning_rate": 1.3536050899368835e-05,
      "loss": 0.0692,
      "step": 19100
    },
    {
      "epoch": 0.9698461850855373,
      "grad_norm": 0.43478238582611084,
      "learning_rate": 1.3534358766096419e-05,
      "loss": 0.0603,
      "step": 19105
    },
    {
      "epoch": 0.9701000050763998,
      "grad_norm": 0.36731022596359253,
      "learning_rate": 1.3532666632824002e-05,
      "loss": 0.06,
      "step": 19110
    },
    {
      "epoch": 0.9703538250672623,
      "grad_norm": 1.0390390157699585,
      "learning_rate": 1.3530974499551586e-05,
      "loss": 0.0686,
      "step": 19115
    },
    {
      "epoch": 0.9706076450581248,
      "grad_norm": 0.3847677707672119,
      "learning_rate": 1.3529282366279168e-05,
      "loss": 0.0622,
      "step": 19120
    },
    {
      "epoch": 0.9708614650489873,
      "grad_norm": 0.302962064743042,
      "learning_rate": 1.3527590233006753e-05,
      "loss": 0.0576,
      "step": 19125
    },
    {
      "epoch": 0.9711152850398498,
      "grad_norm": 0.48355162143707275,
      "learning_rate": 1.3525898099734336e-05,
      "loss": 0.0691,
      "step": 19130
    },
    {
      "epoch": 0.9713691050307122,
      "grad_norm": 0.40184906125068665,
      "learning_rate": 1.352420596646192e-05,
      "loss": 0.064,
      "step": 19135
    },
    {
      "epoch": 0.9716229250215747,
      "grad_norm": 0.3403814136981964,
      "learning_rate": 1.3522513833189504e-05,
      "loss": 0.0654,
      "step": 19140
    },
    {
      "epoch": 0.9718767450124371,
      "grad_norm": 0.37238359451293945,
      "learning_rate": 1.3520821699917085e-05,
      "loss": 0.068,
      "step": 19145
    },
    {
      "epoch": 0.9721305650032996,
      "grad_norm": 0.30097347497940063,
      "learning_rate": 1.351912956664467e-05,
      "loss": 0.0694,
      "step": 19150
    },
    {
      "epoch": 0.9723843849941621,
      "grad_norm": 0.25011131167411804,
      "learning_rate": 1.3517437433372254e-05,
      "loss": 0.0593,
      "step": 19155
    },
    {
      "epoch": 0.9726382049850246,
      "grad_norm": 0.3264051377773285,
      "learning_rate": 1.3515745300099836e-05,
      "loss": 0.0596,
      "step": 19160
    },
    {
      "epoch": 0.9728920249758871,
      "grad_norm": 0.4142516553401947,
      "learning_rate": 1.3514053166827421e-05,
      "loss": 0.0748,
      "step": 19165
    },
    {
      "epoch": 0.9731458449667496,
      "grad_norm": 0.4116646349430084,
      "learning_rate": 1.3512361033555003e-05,
      "loss": 0.0582,
      "step": 19170
    },
    {
      "epoch": 0.9733996649576121,
      "grad_norm": 0.33693939447402954,
      "learning_rate": 1.3510668900282587e-05,
      "loss": 0.0742,
      "step": 19175
    },
    {
      "epoch": 0.9736534849484746,
      "grad_norm": 0.3353005051612854,
      "learning_rate": 1.3508976767010172e-05,
      "loss": 0.0641,
      "step": 19180
    },
    {
      "epoch": 0.973907304939337,
      "grad_norm": 0.4305810332298279,
      "learning_rate": 1.3507284633737754e-05,
      "loss": 0.0645,
      "step": 19185
    },
    {
      "epoch": 0.9741611249301995,
      "grad_norm": 0.3571843206882477,
      "learning_rate": 1.3505592500465339e-05,
      "loss": 0.066,
      "step": 19190
    },
    {
      "epoch": 0.974414944921062,
      "grad_norm": 0.34298276901245117,
      "learning_rate": 1.350390036719292e-05,
      "loss": 0.0736,
      "step": 19195
    },
    {
      "epoch": 0.9746687649119244,
      "grad_norm": 0.4371688961982727,
      "learning_rate": 1.3502208233920504e-05,
      "loss": 0.0638,
      "step": 19200
    },
    {
      "epoch": 0.9749225849027869,
      "grad_norm": 0.43059366941452026,
      "learning_rate": 1.350051610064809e-05,
      "loss": 0.0683,
      "step": 19205
    },
    {
      "epoch": 0.9751764048936494,
      "grad_norm": 0.33127647638320923,
      "learning_rate": 1.3498823967375671e-05,
      "loss": 0.066,
      "step": 19210
    },
    {
      "epoch": 0.9754302248845119,
      "grad_norm": 0.4141124188899994,
      "learning_rate": 1.3497131834103255e-05,
      "loss": 0.0569,
      "step": 19215
    },
    {
      "epoch": 0.9756840448753744,
      "grad_norm": 0.45645686984062195,
      "learning_rate": 1.3495439700830839e-05,
      "loss": 0.0676,
      "step": 19220
    },
    {
      "epoch": 0.9759378648662369,
      "grad_norm": 0.44985485076904297,
      "learning_rate": 1.3493747567558422e-05,
      "loss": 0.0654,
      "step": 19225
    },
    {
      "epoch": 0.9761916848570994,
      "grad_norm": 0.3672463893890381,
      "learning_rate": 1.3492055434286004e-05,
      "loss": 0.0711,
      "step": 19230
    },
    {
      "epoch": 0.9764455048479618,
      "grad_norm": 1.25873601436615,
      "learning_rate": 1.349036330101359e-05,
      "loss": 0.0723,
      "step": 19235
    },
    {
      "epoch": 0.9766993248388243,
      "grad_norm": 0.5634270310401917,
      "learning_rate": 1.3488671167741173e-05,
      "loss": 0.0616,
      "step": 19240
    },
    {
      "epoch": 0.9769531448296868,
      "grad_norm": 0.2947288155555725,
      "learning_rate": 1.3486979034468756e-05,
      "loss": 0.0618,
      "step": 19245
    },
    {
      "epoch": 0.9772069648205493,
      "grad_norm": 0.5857254266738892,
      "learning_rate": 1.348528690119634e-05,
      "loss": 0.0682,
      "step": 19250
    },
    {
      "epoch": 0.9774607848114117,
      "grad_norm": 0.300155907869339,
      "learning_rate": 1.3483594767923922e-05,
      "loss": 0.0563,
      "step": 19255
    },
    {
      "epoch": 0.9777146048022742,
      "grad_norm": 0.521580159664154,
      "learning_rate": 1.3481902634651507e-05,
      "loss": 0.0677,
      "step": 19260
    },
    {
      "epoch": 0.9779684247931367,
      "grad_norm": 0.3018982708454132,
      "learning_rate": 1.348021050137909e-05,
      "loss": 0.0695,
      "step": 19265
    },
    {
      "epoch": 0.9782222447839992,
      "grad_norm": 0.4247705638408661,
      "learning_rate": 1.3478518368106672e-05,
      "loss": 0.0648,
      "step": 19270
    },
    {
      "epoch": 0.9784760647748617,
      "grad_norm": 0.3755514621734619,
      "learning_rate": 1.3476826234834258e-05,
      "loss": 0.0713,
      "step": 19275
    },
    {
      "epoch": 0.9787298847657242,
      "grad_norm": 0.650913417339325,
      "learning_rate": 1.347513410156184e-05,
      "loss": 0.0591,
      "step": 19280
    },
    {
      "epoch": 0.9789837047565866,
      "grad_norm": 0.32536032795906067,
      "learning_rate": 1.3473441968289425e-05,
      "loss": 0.0794,
      "step": 19285
    },
    {
      "epoch": 0.9792375247474491,
      "grad_norm": 0.5901187658309937,
      "learning_rate": 1.3471749835017006e-05,
      "loss": 0.0784,
      "step": 19290
    },
    {
      "epoch": 0.9794913447383116,
      "grad_norm": 0.3043603003025055,
      "learning_rate": 1.347005770174459e-05,
      "loss": 0.0748,
      "step": 19295
    },
    {
      "epoch": 0.9797451647291741,
      "grad_norm": 0.36308273673057556,
      "learning_rate": 1.3468365568472175e-05,
      "loss": 0.0697,
      "step": 19300
    },
    {
      "epoch": 0.9799989847200365,
      "grad_norm": 0.34357786178588867,
      "learning_rate": 1.3466673435199757e-05,
      "loss": 0.0694,
      "step": 19305
    },
    {
      "epoch": 0.980252804710899,
      "grad_norm": 0.9105756878852844,
      "learning_rate": 1.346498130192734e-05,
      "loss": 0.0545,
      "step": 19310
    },
    {
      "epoch": 0.9805066247017615,
      "grad_norm": 0.3355824947357178,
      "learning_rate": 1.3463289168654924e-05,
      "loss": 0.0552,
      "step": 19315
    },
    {
      "epoch": 0.980760444692624,
      "grad_norm": 0.3532446026802063,
      "learning_rate": 1.3461597035382508e-05,
      "loss": 0.0681,
      "step": 19320
    },
    {
      "epoch": 0.9810142646834865,
      "grad_norm": 0.3839724659919739,
      "learning_rate": 1.345990490211009e-05,
      "loss": 0.0596,
      "step": 19325
    },
    {
      "epoch": 0.981268084674349,
      "grad_norm": 0.4662843346595764,
      "learning_rate": 1.3458212768837675e-05,
      "loss": 0.0633,
      "step": 19330
    },
    {
      "epoch": 0.9815219046652114,
      "grad_norm": 0.4850928485393524,
      "learning_rate": 1.3456520635565258e-05,
      "loss": 0.06,
      "step": 19335
    },
    {
      "epoch": 0.9817757246560739,
      "grad_norm": 0.5426207184791565,
      "learning_rate": 1.3454828502292842e-05,
      "loss": 0.0711,
      "step": 19340
    },
    {
      "epoch": 0.9820295446469364,
      "grad_norm": 0.24476711452007294,
      "learning_rate": 1.3453136369020425e-05,
      "loss": 0.0687,
      "step": 19345
    },
    {
      "epoch": 0.9822833646377989,
      "grad_norm": 0.39669865369796753,
      "learning_rate": 1.3451444235748007e-05,
      "loss": 0.0648,
      "step": 19350
    },
    {
      "epoch": 0.9825371846286614,
      "grad_norm": 0.329333633184433,
      "learning_rate": 1.3449752102475593e-05,
      "loss": 0.0623,
      "step": 19355
    },
    {
      "epoch": 0.9827910046195238,
      "grad_norm": 0.5927011966705322,
      "learning_rate": 1.3448059969203176e-05,
      "loss": 0.0579,
      "step": 19360
    },
    {
      "epoch": 0.9830448246103863,
      "grad_norm": 0.4235691726207733,
      "learning_rate": 1.3446367835930758e-05,
      "loss": 0.0576,
      "step": 19365
    },
    {
      "epoch": 0.9832986446012488,
      "grad_norm": 0.3612593412399292,
      "learning_rate": 1.3444675702658343e-05,
      "loss": 0.0659,
      "step": 19370
    },
    {
      "epoch": 0.9835524645921113,
      "grad_norm": 0.4044787287712097,
      "learning_rate": 1.3442983569385925e-05,
      "loss": 0.0665,
      "step": 19375
    },
    {
      "epoch": 0.9838062845829737,
      "grad_norm": 0.32001495361328125,
      "learning_rate": 1.344129143611351e-05,
      "loss": 0.058,
      "step": 19380
    },
    {
      "epoch": 0.9840601045738362,
      "grad_norm": 0.2624494731426239,
      "learning_rate": 1.3439599302841094e-05,
      "loss": 0.0622,
      "step": 19385
    },
    {
      "epoch": 0.9843139245646987,
      "grad_norm": 0.3039698004722595,
      "learning_rate": 1.3437907169568676e-05,
      "loss": 0.0539,
      "step": 19390
    },
    {
      "epoch": 0.9845677445555612,
      "grad_norm": 1.1381529569625854,
      "learning_rate": 1.3436215036296261e-05,
      "loss": 0.0671,
      "step": 19395
    },
    {
      "epoch": 0.9848215645464237,
      "grad_norm": 0.3349166810512543,
      "learning_rate": 1.3434522903023843e-05,
      "loss": 0.0663,
      "step": 19400
    },
    {
      "epoch": 0.9850753845372862,
      "grad_norm": 0.31104400753974915,
      "learning_rate": 1.3432830769751426e-05,
      "loss": 0.053,
      "step": 19405
    },
    {
      "epoch": 0.9853292045281486,
      "grad_norm": 0.3512427806854248,
      "learning_rate": 1.3431138636479012e-05,
      "loss": 0.0741,
      "step": 19410
    },
    {
      "epoch": 0.9855830245190111,
      "grad_norm": 0.5050990581512451,
      "learning_rate": 1.3429446503206593e-05,
      "loss": 0.0678,
      "step": 19415
    },
    {
      "epoch": 0.9858368445098736,
      "grad_norm": 0.4575447142124176,
      "learning_rate": 1.3427754369934177e-05,
      "loss": 0.0707,
      "step": 19420
    },
    {
      "epoch": 0.9860906645007361,
      "grad_norm": 0.464824378490448,
      "learning_rate": 1.342606223666176e-05,
      "loss": 0.0573,
      "step": 19425
    },
    {
      "epoch": 0.9863444844915985,
      "grad_norm": 0.2967434823513031,
      "learning_rate": 1.3424370103389344e-05,
      "loss": 0.0627,
      "step": 19430
    },
    {
      "epoch": 0.986598304482461,
      "grad_norm": 0.425357848405838,
      "learning_rate": 1.342267797011693e-05,
      "loss": 0.0534,
      "step": 19435
    },
    {
      "epoch": 0.9868521244733235,
      "grad_norm": 0.437012255191803,
      "learning_rate": 1.3420985836844511e-05,
      "loss": 0.0685,
      "step": 19440
    },
    {
      "epoch": 0.987105944464186,
      "grad_norm": 0.37560611963272095,
      "learning_rate": 1.3419293703572095e-05,
      "loss": 0.0605,
      "step": 19445
    },
    {
      "epoch": 0.9873597644550485,
      "grad_norm": 0.35586923360824585,
      "learning_rate": 1.3417601570299678e-05,
      "loss": 0.0641,
      "step": 19450
    },
    {
      "epoch": 0.987613584445911,
      "grad_norm": 0.3776257038116455,
      "learning_rate": 1.3415909437027262e-05,
      "loss": 0.0571,
      "step": 19455
    },
    {
      "epoch": 0.9878674044367735,
      "grad_norm": 1.247651219367981,
      "learning_rate": 1.3414217303754844e-05,
      "loss": 0.0635,
      "step": 19460
    },
    {
      "epoch": 0.988121224427636,
      "grad_norm": 0.5398990511894226,
      "learning_rate": 1.3412525170482429e-05,
      "loss": 0.0712,
      "step": 19465
    },
    {
      "epoch": 0.9883750444184984,
      "grad_norm": 0.49520477652549744,
      "learning_rate": 1.341083303721001e-05,
      "loss": 0.0661,
      "step": 19470
    },
    {
      "epoch": 0.9886288644093609,
      "grad_norm": 0.38991716504096985,
      "learning_rate": 1.3409140903937594e-05,
      "loss": 0.0766,
      "step": 19475
    },
    {
      "epoch": 0.9888826844002233,
      "grad_norm": 0.2874346971511841,
      "learning_rate": 1.340744877066518e-05,
      "loss": 0.0654,
      "step": 19480
    },
    {
      "epoch": 0.9891365043910858,
      "grad_norm": 0.5209989547729492,
      "learning_rate": 1.3405756637392761e-05,
      "loss": 0.0577,
      "step": 19485
    },
    {
      "epoch": 0.9893903243819483,
      "grad_norm": 0.7382110953330994,
      "learning_rate": 1.3404064504120347e-05,
      "loss": 0.0635,
      "step": 19490
    },
    {
      "epoch": 0.9896441443728108,
      "grad_norm": 0.41847217082977295,
      "learning_rate": 1.3402372370847928e-05,
      "loss": 0.0525,
      "step": 19495
    },
    {
      "epoch": 0.9898979643636733,
      "grad_norm": 0.3458189070224762,
      "learning_rate": 1.3400680237575512e-05,
      "loss": 0.1819,
      "step": 19500
    },
    {
      "epoch": 0.9901517843545358,
      "grad_norm": 0.3702060580253601,
      "learning_rate": 1.3398988104303097e-05,
      "loss": 0.0635,
      "step": 19505
    },
    {
      "epoch": 0.9904056043453983,
      "grad_norm": 0.3774377405643463,
      "learning_rate": 1.3397295971030679e-05,
      "loss": 0.0685,
      "step": 19510
    },
    {
      "epoch": 0.9906594243362608,
      "grad_norm": 0.296116441488266,
      "learning_rate": 1.3395603837758263e-05,
      "loss": 0.0704,
      "step": 19515
    },
    {
      "epoch": 0.9909132443271232,
      "grad_norm": 0.5571689605712891,
      "learning_rate": 1.3393911704485846e-05,
      "loss": 0.0602,
      "step": 19520
    },
    {
      "epoch": 0.9911670643179857,
      "grad_norm": 0.36025819182395935,
      "learning_rate": 1.339221957121343e-05,
      "loss": 0.0623,
      "step": 19525
    },
    {
      "epoch": 0.9914208843088481,
      "grad_norm": 0.39932265877723694,
      "learning_rate": 1.3390527437941015e-05,
      "loss": 0.0564,
      "step": 19530
    },
    {
      "epoch": 0.9916747042997106,
      "grad_norm": 0.35256311297416687,
      "learning_rate": 1.3388835304668597e-05,
      "loss": 0.0751,
      "step": 19535
    },
    {
      "epoch": 0.9919285242905731,
      "grad_norm": 0.4102122187614441,
      "learning_rate": 1.338714317139618e-05,
      "loss": 0.0593,
      "step": 19540
    },
    {
      "epoch": 0.9921823442814356,
      "grad_norm": 0.3455532193183899,
      "learning_rate": 1.3385451038123764e-05,
      "loss": 0.0598,
      "step": 19545
    },
    {
      "epoch": 0.9924361642722981,
      "grad_norm": 0.3834126889705658,
      "learning_rate": 1.3383758904851347e-05,
      "loss": 0.0608,
      "step": 19550
    },
    {
      "epoch": 0.9926899842631606,
      "grad_norm": 0.4701562821865082,
      "learning_rate": 1.338206677157893e-05,
      "loss": 0.057,
      "step": 19555
    },
    {
      "epoch": 0.9929438042540231,
      "grad_norm": 0.5309749245643616,
      "learning_rate": 1.3380374638306514e-05,
      "loss": 0.0592,
      "step": 19560
    },
    {
      "epoch": 0.9931976242448856,
      "grad_norm": 0.4886894226074219,
      "learning_rate": 1.3378682505034098e-05,
      "loss": 0.0625,
      "step": 19565
    },
    {
      "epoch": 0.993451444235748,
      "grad_norm": 0.5212421417236328,
      "learning_rate": 1.337699037176168e-05,
      "loss": 0.0668,
      "step": 19570
    },
    {
      "epoch": 0.9937052642266105,
      "grad_norm": 0.3621155917644501,
      "learning_rate": 1.3375298238489265e-05,
      "loss": 0.0623,
      "step": 19575
    },
    {
      "epoch": 0.9939590842174729,
      "grad_norm": 0.47136250138282776,
      "learning_rate": 1.3373606105216847e-05,
      "loss": 0.0728,
      "step": 19580
    },
    {
      "epoch": 0.9942129042083354,
      "grad_norm": 0.31008508801460266,
      "learning_rate": 1.3371913971944432e-05,
      "loss": 0.0583,
      "step": 19585
    },
    {
      "epoch": 0.9944667241991979,
      "grad_norm": 0.47124168276786804,
      "learning_rate": 1.3370221838672016e-05,
      "loss": 0.0659,
      "step": 19590
    },
    {
      "epoch": 0.9947205441900604,
      "grad_norm": 0.40367239713668823,
      "learning_rate": 1.3368529705399598e-05,
      "loss": 0.059,
      "step": 19595
    },
    {
      "epoch": 0.9949743641809229,
      "grad_norm": 0.37501493096351624,
      "learning_rate": 1.3366837572127183e-05,
      "loss": 0.0666,
      "step": 19600
    },
    {
      "epoch": 0.9952281841717854,
      "grad_norm": 0.6273606419563293,
      "learning_rate": 1.3365145438854765e-05,
      "loss": 0.0662,
      "step": 19605
    },
    {
      "epoch": 0.9954820041626479,
      "grad_norm": 0.3741655945777893,
      "learning_rate": 1.3363453305582348e-05,
      "loss": 0.0577,
      "step": 19610
    },
    {
      "epoch": 0.9957358241535104,
      "grad_norm": 0.33114105463027954,
      "learning_rate": 1.3361761172309933e-05,
      "loss": 0.0658,
      "step": 19615
    },
    {
      "epoch": 0.9959896441443729,
      "grad_norm": 0.35361751914024353,
      "learning_rate": 1.3360069039037515e-05,
      "loss": 0.063,
      "step": 19620
    },
    {
      "epoch": 0.9962434641352353,
      "grad_norm": 0.41206908226013184,
      "learning_rate": 1.3358376905765099e-05,
      "loss": 0.0629,
      "step": 19625
    },
    {
      "epoch": 0.9964972841260977,
      "grad_norm": 0.4646819233894348,
      "learning_rate": 1.3356684772492682e-05,
      "loss": 0.0647,
      "step": 19630
    },
    {
      "epoch": 0.9967511041169602,
      "grad_norm": 0.3486287295818329,
      "learning_rate": 1.3354992639220266e-05,
      "loss": 0.0571,
      "step": 19635
    },
    {
      "epoch": 0.9970049241078227,
      "grad_norm": 0.3921375274658203,
      "learning_rate": 1.3353300505947851e-05,
      "loss": 0.057,
      "step": 19640
    },
    {
      "epoch": 0.9972587440986852,
      "grad_norm": 0.3869142532348633,
      "learning_rate": 1.3351608372675433e-05,
      "loss": 0.0695,
      "step": 19645
    },
    {
      "epoch": 0.9975125640895477,
      "grad_norm": 0.652873694896698,
      "learning_rate": 1.3349916239403015e-05,
      "loss": 0.0608,
      "step": 19650
    },
    {
      "epoch": 0.9977663840804102,
      "grad_norm": 0.35025376081466675,
      "learning_rate": 1.33482241061306e-05,
      "loss": 0.0675,
      "step": 19655
    },
    {
      "epoch": 0.9980202040712727,
      "grad_norm": 0.45848169922828674,
      "learning_rate": 1.3346531972858184e-05,
      "loss": 0.0656,
      "step": 19660
    },
    {
      "epoch": 0.9982740240621352,
      "grad_norm": 0.3314003050327301,
      "learning_rate": 1.3344839839585766e-05,
      "loss": 0.0591,
      "step": 19665
    },
    {
      "epoch": 0.9985278440529977,
      "grad_norm": 0.3548816740512848,
      "learning_rate": 1.334314770631335e-05,
      "loss": 0.0668,
      "step": 19670
    },
    {
      "epoch": 0.99878166404386,
      "grad_norm": 0.46631526947021484,
      "learning_rate": 1.3341455573040933e-05,
      "loss": 0.0563,
      "step": 19675
    },
    {
      "epoch": 0.9990354840347225,
      "grad_norm": 0.5142710208892822,
      "learning_rate": 1.3339763439768518e-05,
      "loss": 0.0609,
      "step": 19680
    },
    {
      "epoch": 0.999289304025585,
      "grad_norm": 0.965015172958374,
      "learning_rate": 1.3338071306496101e-05,
      "loss": 0.0591,
      "step": 19685
    },
    {
      "epoch": 0.9995431240164475,
      "grad_norm": 0.568295955657959,
      "learning_rate": 1.3336379173223683e-05,
      "loss": 0.0636,
      "step": 19690
    },
    {
      "epoch": 0.99979694400731,
      "grad_norm": 0.3382091820240021,
      "learning_rate": 1.3334687039951268e-05,
      "loss": 0.0609,
      "step": 19695
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.160739004611969,
      "eval_runtime": 1784.4225,
      "eval_samples_per_second": 70.051,
      "eval_steps_per_second": 2.19,
      "step": 19699
    },
    {
      "epoch": 1.0000507639981724,
      "grad_norm": 0.3930205702781677,
      "learning_rate": 1.333299490667885e-05,
      "loss": 0.068,
      "step": 19700
    },
    {
      "epoch": 1.000304583989035,
      "grad_norm": 0.3845120370388031,
      "learning_rate": 1.3331302773406434e-05,
      "loss": 0.0642,
      "step": 19705
    },
    {
      "epoch": 1.0005584039798974,
      "grad_norm": 0.4073793292045593,
      "learning_rate": 1.3329610640134019e-05,
      "loss": 0.0605,
      "step": 19710
    },
    {
      "epoch": 1.00081222397076,
      "grad_norm": 0.34505850076675415,
      "learning_rate": 1.3327918506861601e-05,
      "loss": 0.0636,
      "step": 19715
    },
    {
      "epoch": 1.0010660439616224,
      "grad_norm": 0.6583669781684875,
      "learning_rate": 1.3326226373589185e-05,
      "loss": 0.0639,
      "step": 19720
    },
    {
      "epoch": 1.001319863952485,
      "grad_norm": 0.5834250450134277,
      "learning_rate": 1.3324534240316768e-05,
      "loss": 0.0641,
      "step": 19725
    },
    {
      "epoch": 1.0015736839433473,
      "grad_norm": 0.3789755403995514,
      "learning_rate": 1.3322842107044352e-05,
      "loss": 0.0593,
      "step": 19730
    },
    {
      "epoch": 1.00182750393421,
      "grad_norm": 0.3655293881893158,
      "learning_rate": 1.3321149973771937e-05,
      "loss": 0.0675,
      "step": 19735
    },
    {
      "epoch": 1.0020813239250723,
      "grad_norm": 0.4233531355857849,
      "learning_rate": 1.3319457840499519e-05,
      "loss": 0.0625,
      "step": 19740
    },
    {
      "epoch": 1.002335143915935,
      "grad_norm": 0.3656198978424072,
      "learning_rate": 1.3317765707227102e-05,
      "loss": 0.0637,
      "step": 19745
    },
    {
      "epoch": 1.0025889639067973,
      "grad_norm": 0.3142138123512268,
      "learning_rate": 1.3316073573954686e-05,
      "loss": 0.0648,
      "step": 19750
    },
    {
      "epoch": 1.0028427838976597,
      "grad_norm": 0.37569764256477356,
      "learning_rate": 1.331438144068227e-05,
      "loss": 0.0681,
      "step": 19755
    },
    {
      "epoch": 1.0030966038885223,
      "grad_norm": 0.4830641746520996,
      "learning_rate": 1.3312689307409851e-05,
      "loss": 0.0621,
      "step": 19760
    },
    {
      "epoch": 1.0033504238793847,
      "grad_norm": 0.3205713629722595,
      "learning_rate": 1.3310997174137436e-05,
      "loss": 0.0579,
      "step": 19765
    },
    {
      "epoch": 1.0036042438702473,
      "grad_norm": 0.4066654145717621,
      "learning_rate": 1.330930504086502e-05,
      "loss": 0.0529,
      "step": 19770
    },
    {
      "epoch": 1.0038580638611097,
      "grad_norm": 0.35887086391448975,
      "learning_rate": 1.3307612907592604e-05,
      "loss": 0.0603,
      "step": 19775
    },
    {
      "epoch": 1.0041118838519723,
      "grad_norm": 0.4492895007133484,
      "learning_rate": 1.3305920774320187e-05,
      "loss": 0.0569,
      "step": 19780
    },
    {
      "epoch": 1.0043657038428346,
      "grad_norm": 0.38979437947273254,
      "learning_rate": 1.3304228641047769e-05,
      "loss": 0.071,
      "step": 19785
    },
    {
      "epoch": 1.0046195238336972,
      "grad_norm": 0.36487793922424316,
      "learning_rate": 1.3302536507775354e-05,
      "loss": 0.0599,
      "step": 19790
    },
    {
      "epoch": 1.0048733438245596,
      "grad_norm": 0.4509921073913574,
      "learning_rate": 1.3300844374502938e-05,
      "loss": 0.0717,
      "step": 19795
    },
    {
      "epoch": 1.005127163815422,
      "grad_norm": 0.39418506622314453,
      "learning_rate": 1.329915224123052e-05,
      "loss": 0.0608,
      "step": 19800
    },
    {
      "epoch": 1.0053809838062846,
      "grad_norm": 0.44550445675849915,
      "learning_rate": 1.3297460107958105e-05,
      "loss": 0.0711,
      "step": 19805
    },
    {
      "epoch": 1.005634803797147,
      "grad_norm": 0.40730229020118713,
      "learning_rate": 1.3295767974685687e-05,
      "loss": 0.0635,
      "step": 19810
    },
    {
      "epoch": 1.0058886237880096,
      "grad_norm": 0.41011321544647217,
      "learning_rate": 1.329407584141327e-05,
      "loss": 0.0596,
      "step": 19815
    },
    {
      "epoch": 1.006142443778872,
      "grad_norm": 0.36048024892807007,
      "learning_rate": 1.3292383708140855e-05,
      "loss": 0.0565,
      "step": 19820
    },
    {
      "epoch": 1.0063962637697346,
      "grad_norm": 0.49804311990737915,
      "learning_rate": 1.3290691574868437e-05,
      "loss": 0.0658,
      "step": 19825
    },
    {
      "epoch": 1.006650083760597,
      "grad_norm": 0.3198188841342926,
      "learning_rate": 1.3288999441596023e-05,
      "loss": 0.0631,
      "step": 19830
    },
    {
      "epoch": 1.0069039037514596,
      "grad_norm": 0.4254019856452942,
      "learning_rate": 1.3287307308323604e-05,
      "loss": 0.0609,
      "step": 19835
    },
    {
      "epoch": 1.007157723742322,
      "grad_norm": 0.31280845403671265,
      "learning_rate": 1.3285615175051188e-05,
      "loss": 0.065,
      "step": 19840
    },
    {
      "epoch": 1.0074115437331845,
      "grad_norm": 0.4120943546295166,
      "learning_rate": 1.3283923041778773e-05,
      "loss": 0.0609,
      "step": 19845
    },
    {
      "epoch": 1.007665363724047,
      "grad_norm": 0.39241835474967957,
      "learning_rate": 1.3282230908506355e-05,
      "loss": 0.0664,
      "step": 19850
    },
    {
      "epoch": 1.0079191837149093,
      "grad_norm": 0.5152608752250671,
      "learning_rate": 1.3280538775233937e-05,
      "loss": 0.0623,
      "step": 19855
    },
    {
      "epoch": 1.008173003705772,
      "grad_norm": 0.7134572267532349,
      "learning_rate": 1.3278846641961522e-05,
      "loss": 0.068,
      "step": 19860
    },
    {
      "epoch": 1.0084268236966343,
      "grad_norm": 0.3215194344520569,
      "learning_rate": 1.3277154508689106e-05,
      "loss": 0.0608,
      "step": 19865
    },
    {
      "epoch": 1.0086806436874969,
      "grad_norm": 0.5526114702224731,
      "learning_rate": 1.3275462375416687e-05,
      "loss": 0.0632,
      "step": 19870
    },
    {
      "epoch": 1.0089344636783593,
      "grad_norm": 0.4783251881599426,
      "learning_rate": 1.3273770242144273e-05,
      "loss": 0.0695,
      "step": 19875
    },
    {
      "epoch": 1.0091882836692219,
      "grad_norm": 0.48075810074806213,
      "learning_rate": 1.3272078108871855e-05,
      "loss": 0.0576,
      "step": 19880
    },
    {
      "epoch": 1.0094421036600842,
      "grad_norm": 0.3602225184440613,
      "learning_rate": 1.327038597559944e-05,
      "loss": 0.0565,
      "step": 19885
    },
    {
      "epoch": 1.0096959236509468,
      "grad_norm": 0.3769897222518921,
      "learning_rate": 1.3268693842327023e-05,
      "loss": 0.0665,
      "step": 19890
    },
    {
      "epoch": 1.0099497436418092,
      "grad_norm": 0.3689655065536499,
      "learning_rate": 1.3267001709054605e-05,
      "loss": 0.0638,
      "step": 19895
    },
    {
      "epoch": 1.0102035636326716,
      "grad_norm": 0.26339930295944214,
      "learning_rate": 1.326530957578219e-05,
      "loss": 0.0568,
      "step": 19900
    },
    {
      "epoch": 1.0104573836235342,
      "grad_norm": 0.5178040862083435,
      "learning_rate": 1.3263617442509772e-05,
      "loss": 0.061,
      "step": 19905
    },
    {
      "epoch": 1.0107112036143966,
      "grad_norm": 0.4020889103412628,
      "learning_rate": 1.3261925309237356e-05,
      "loss": 0.0614,
      "step": 19910
    },
    {
      "epoch": 1.0109650236052592,
      "grad_norm": 0.2972166836261749,
      "learning_rate": 1.3260233175964941e-05,
      "loss": 0.0644,
      "step": 19915
    },
    {
      "epoch": 1.0112188435961216,
      "grad_norm": 0.36977848410606384,
      "learning_rate": 1.3258541042692523e-05,
      "loss": 0.0621,
      "step": 19920
    },
    {
      "epoch": 1.0114726635869842,
      "grad_norm": 0.46586596965789795,
      "learning_rate": 1.3256848909420108e-05,
      "loss": 0.0593,
      "step": 19925
    },
    {
      "epoch": 1.0117264835778466,
      "grad_norm": 0.34465721249580383,
      "learning_rate": 1.325515677614769e-05,
      "loss": 0.0635,
      "step": 19930
    },
    {
      "epoch": 1.0119803035687092,
      "grad_norm": 0.27624374628067017,
      "learning_rate": 1.3253464642875274e-05,
      "loss": 0.0585,
      "step": 19935
    },
    {
      "epoch": 1.0122341235595715,
      "grad_norm": 0.41285547614097595,
      "learning_rate": 1.3251772509602859e-05,
      "loss": 0.0718,
      "step": 19940
    },
    {
      "epoch": 1.012487943550434,
      "grad_norm": 0.3415983319282532,
      "learning_rate": 1.325008037633044e-05,
      "loss": 0.0716,
      "step": 19945
    },
    {
      "epoch": 1.0127417635412965,
      "grad_norm": 0.35713669657707214,
      "learning_rate": 1.3248388243058024e-05,
      "loss": 0.0605,
      "step": 19950
    },
    {
      "epoch": 1.012995583532159,
      "grad_norm": 0.31129032373428345,
      "learning_rate": 1.3246696109785608e-05,
      "loss": 0.0633,
      "step": 19955
    },
    {
      "epoch": 1.0132494035230215,
      "grad_norm": 0.9359402656555176,
      "learning_rate": 1.3245003976513191e-05,
      "loss": 0.0641,
      "step": 19960
    },
    {
      "epoch": 1.0135032235138839,
      "grad_norm": 0.3182302713394165,
      "learning_rate": 1.3243311843240773e-05,
      "loss": 0.0627,
      "step": 19965
    },
    {
      "epoch": 1.0137570435047465,
      "grad_norm": 0.3964100778102875,
      "learning_rate": 1.3241619709968358e-05,
      "loss": 0.0568,
      "step": 19970
    },
    {
      "epoch": 1.0140108634956089,
      "grad_norm": 0.35237252712249756,
      "learning_rate": 1.3239927576695942e-05,
      "loss": 0.0679,
      "step": 19975
    },
    {
      "epoch": 1.0142646834864715,
      "grad_norm": 0.37263184785842896,
      "learning_rate": 1.3238235443423525e-05,
      "loss": 0.0585,
      "step": 19980
    },
    {
      "epoch": 1.0145185034773339,
      "grad_norm": 0.44940632581710815,
      "learning_rate": 1.3236543310151109e-05,
      "loss": 0.0632,
      "step": 19985
    },
    {
      "epoch": 1.0147723234681965,
      "grad_norm": 0.7039437890052795,
      "learning_rate": 1.323485117687869e-05,
      "loss": 0.0697,
      "step": 19990
    },
    {
      "epoch": 1.0150261434590588,
      "grad_norm": 0.7221899628639221,
      "learning_rate": 1.3233159043606276e-05,
      "loss": 0.0585,
      "step": 19995
    },
    {
      "epoch": 1.0152799634499212,
      "grad_norm": 0.25587931275367737,
      "learning_rate": 1.323146691033386e-05,
      "loss": 0.0663,
      "step": 20000
    },
    {
      "epoch": 1.0155337834407838,
      "grad_norm": 0.34087198972702026,
      "learning_rate": 1.3229774777061441e-05,
      "loss": 0.0739,
      "step": 20005
    },
    {
      "epoch": 1.0157876034316462,
      "grad_norm": 0.49121344089508057,
      "learning_rate": 1.3228082643789027e-05,
      "loss": 0.0589,
      "step": 20010
    },
    {
      "epoch": 1.0160414234225088,
      "grad_norm": 0.39154985547065735,
      "learning_rate": 1.3226390510516609e-05,
      "loss": 0.0561,
      "step": 20015
    },
    {
      "epoch": 1.0162952434133712,
      "grad_norm": 0.3109443187713623,
      "learning_rate": 1.3224698377244194e-05,
      "loss": 0.0565,
      "step": 20020
    },
    {
      "epoch": 1.0165490634042338,
      "grad_norm": 0.349619597196579,
      "learning_rate": 1.3223006243971777e-05,
      "loss": 0.0612,
      "step": 20025
    },
    {
      "epoch": 1.0168028833950962,
      "grad_norm": 0.6276240348815918,
      "learning_rate": 1.322131411069936e-05,
      "loss": 0.0633,
      "step": 20030
    },
    {
      "epoch": 1.0170567033859588,
      "grad_norm": 0.4947071373462677,
      "learning_rate": 1.3219621977426944e-05,
      "loss": 0.0532,
      "step": 20035
    },
    {
      "epoch": 1.0173105233768212,
      "grad_norm": 0.987997829914093,
      "learning_rate": 1.3217929844154526e-05,
      "loss": 0.0661,
      "step": 20040
    },
    {
      "epoch": 1.0175643433676835,
      "grad_norm": 0.5403960347175598,
      "learning_rate": 1.321623771088211e-05,
      "loss": 0.0588,
      "step": 20045
    },
    {
      "epoch": 1.0178181633585461,
      "grad_norm": 0.41313982009887695,
      "learning_rate": 1.3214545577609695e-05,
      "loss": 0.0599,
      "step": 20050
    },
    {
      "epoch": 1.0180719833494085,
      "grad_norm": 0.3229277729988098,
      "learning_rate": 1.3212853444337277e-05,
      "loss": 0.0602,
      "step": 20055
    },
    {
      "epoch": 1.0183258033402711,
      "grad_norm": 0.5154731273651123,
      "learning_rate": 1.3211161311064859e-05,
      "loss": 0.0557,
      "step": 20060
    },
    {
      "epoch": 1.0185796233311335,
      "grad_norm": 0.3665696978569031,
      "learning_rate": 1.3209469177792444e-05,
      "loss": 0.0642,
      "step": 20065
    },
    {
      "epoch": 1.018833443321996,
      "grad_norm": 0.35480642318725586,
      "learning_rate": 1.3207777044520028e-05,
      "loss": 0.0621,
      "step": 20070
    },
    {
      "epoch": 1.0190872633128585,
      "grad_norm": 0.3062871992588043,
      "learning_rate": 1.3206084911247611e-05,
      "loss": 0.0646,
      "step": 20075
    },
    {
      "epoch": 1.019341083303721,
      "grad_norm": 0.37780115008354187,
      "learning_rate": 1.3204392777975195e-05,
      "loss": 0.0559,
      "step": 20080
    },
    {
      "epoch": 1.0195949032945835,
      "grad_norm": 0.5313156843185425,
      "learning_rate": 1.3202700644702776e-05,
      "loss": 0.0635,
      "step": 20085
    },
    {
      "epoch": 1.019848723285446,
      "grad_norm": 0.4488339424133301,
      "learning_rate": 1.3201008511430362e-05,
      "loss": 0.0559,
      "step": 20090
    },
    {
      "epoch": 1.0201025432763084,
      "grad_norm": 0.3646913766860962,
      "learning_rate": 1.3199316378157945e-05,
      "loss": 0.0573,
      "step": 20095
    },
    {
      "epoch": 1.0203563632671708,
      "grad_norm": 0.2983665466308594,
      "learning_rate": 1.3197624244885527e-05,
      "loss": 0.0694,
      "step": 20100
    },
    {
      "epoch": 1.0206101832580334,
      "grad_norm": 0.4676768481731415,
      "learning_rate": 1.3195932111613112e-05,
      "loss": 0.0594,
      "step": 20105
    },
    {
      "epoch": 1.0208640032488958,
      "grad_norm": 0.37798216938972473,
      "learning_rate": 1.3194239978340694e-05,
      "loss": 0.0671,
      "step": 20110
    },
    {
      "epoch": 1.0211178232397584,
      "grad_norm": 0.28777554631233215,
      "learning_rate": 1.3192547845068278e-05,
      "loss": 0.0513,
      "step": 20115
    },
    {
      "epoch": 1.0213716432306208,
      "grad_norm": 0.5031976699829102,
      "learning_rate": 1.3190855711795863e-05,
      "loss": 0.0551,
      "step": 20120
    },
    {
      "epoch": 1.0216254632214834,
      "grad_norm": 0.8326206803321838,
      "learning_rate": 1.3189163578523445e-05,
      "loss": 0.0711,
      "step": 20125
    },
    {
      "epoch": 1.0218792832123458,
      "grad_norm": 0.35230034589767456,
      "learning_rate": 1.318747144525103e-05,
      "loss": 0.0674,
      "step": 20130
    },
    {
      "epoch": 1.0221331032032084,
      "grad_norm": 0.8358342051506042,
      "learning_rate": 1.3185779311978612e-05,
      "loss": 0.0586,
      "step": 20135
    },
    {
      "epoch": 1.0223869231940708,
      "grad_norm": 1.732825517654419,
      "learning_rate": 1.3184087178706195e-05,
      "loss": 0.0644,
      "step": 20140
    },
    {
      "epoch": 1.0226407431849331,
      "grad_norm": 0.36032116413116455,
      "learning_rate": 1.318239504543378e-05,
      "loss": 0.0567,
      "step": 20145
    },
    {
      "epoch": 1.0228945631757957,
      "grad_norm": 0.36123108863830566,
      "learning_rate": 1.3180702912161363e-05,
      "loss": 0.0582,
      "step": 20150
    },
    {
      "epoch": 1.0231483831666581,
      "grad_norm": 0.32960960268974304,
      "learning_rate": 1.3179010778888946e-05,
      "loss": 0.0617,
      "step": 20155
    },
    {
      "epoch": 1.0234022031575207,
      "grad_norm": 0.4045203626155853,
      "learning_rate": 1.317731864561653e-05,
      "loss": 0.059,
      "step": 20160
    },
    {
      "epoch": 1.023656023148383,
      "grad_norm": 0.38749194145202637,
      "learning_rate": 1.3175626512344113e-05,
      "loss": 0.0567,
      "step": 20165
    },
    {
      "epoch": 1.0239098431392457,
      "grad_norm": 0.37718096375465393,
      "learning_rate": 1.3173934379071698e-05,
      "loss": 0.0693,
      "step": 20170
    },
    {
      "epoch": 1.024163663130108,
      "grad_norm": 0.9238191843032837,
      "learning_rate": 1.317224224579928e-05,
      "loss": 0.0565,
      "step": 20175
    },
    {
      "epoch": 1.0244174831209707,
      "grad_norm": 0.41635268926620483,
      "learning_rate": 1.3170550112526864e-05,
      "loss": 0.0659,
      "step": 20180
    },
    {
      "epoch": 1.024671303111833,
      "grad_norm": 0.41010648012161255,
      "learning_rate": 1.3168857979254447e-05,
      "loss": 0.0517,
      "step": 20185
    },
    {
      "epoch": 1.0249251231026957,
      "grad_norm": 0.40220311284065247,
      "learning_rate": 1.3167165845982031e-05,
      "loss": 0.0689,
      "step": 20190
    },
    {
      "epoch": 1.025178943093558,
      "grad_norm": 0.40923091769218445,
      "learning_rate": 1.3165473712709613e-05,
      "loss": 0.0595,
      "step": 20195
    },
    {
      "epoch": 1.0254327630844204,
      "grad_norm": 0.3411441445350647,
      "learning_rate": 1.3163781579437198e-05,
      "loss": 0.0644,
      "step": 20200
    },
    {
      "epoch": 1.025686583075283,
      "grad_norm": 0.3687761425971985,
      "learning_rate": 1.3162089446164782e-05,
      "loss": 0.0579,
      "step": 20205
    },
    {
      "epoch": 1.0259404030661454,
      "grad_norm": 0.3332912027835846,
      "learning_rate": 1.3160397312892363e-05,
      "loss": 0.0651,
      "step": 20210
    },
    {
      "epoch": 1.026194223057008,
      "grad_norm": 0.851193904876709,
      "learning_rate": 1.3158705179619949e-05,
      "loss": 0.0657,
      "step": 20215
    },
    {
      "epoch": 1.0264480430478704,
      "grad_norm": 0.4673822224140167,
      "learning_rate": 1.315701304634753e-05,
      "loss": 0.0572,
      "step": 20220
    },
    {
      "epoch": 1.026701863038733,
      "grad_norm": 0.33037447929382324,
      "learning_rate": 1.3155320913075116e-05,
      "loss": 0.0683,
      "step": 20225
    },
    {
      "epoch": 1.0269556830295954,
      "grad_norm": 0.6725794672966003,
      "learning_rate": 1.31536287798027e-05,
      "loss": 0.0666,
      "step": 20230
    },
    {
      "epoch": 1.027209503020458,
      "grad_norm": 0.6721415519714355,
      "learning_rate": 1.3151936646530281e-05,
      "loss": 0.0791,
      "step": 20235
    },
    {
      "epoch": 1.0274633230113204,
      "grad_norm": 0.39361828565597534,
      "learning_rate": 1.3150244513257866e-05,
      "loss": 0.0651,
      "step": 20240
    },
    {
      "epoch": 1.0277171430021828,
      "grad_norm": 0.30990535020828247,
      "learning_rate": 1.3148552379985448e-05,
      "loss": 0.0637,
      "step": 20245
    },
    {
      "epoch": 1.0279709629930454,
      "grad_norm": 0.44756659865379333,
      "learning_rate": 1.3146860246713032e-05,
      "loss": 0.0678,
      "step": 20250
    },
    {
      "epoch": 1.0282247829839077,
      "grad_norm": 0.3761730194091797,
      "learning_rate": 1.3145168113440615e-05,
      "loss": 0.0621,
      "step": 20255
    },
    {
      "epoch": 1.0284786029747703,
      "grad_norm": 0.29764240980148315,
      "learning_rate": 1.3143475980168199e-05,
      "loss": 0.0572,
      "step": 20260
    },
    {
      "epoch": 1.0287324229656327,
      "grad_norm": 0.31058117747306824,
      "learning_rate": 1.3141783846895784e-05,
      "loss": 0.0497,
      "step": 20265
    },
    {
      "epoch": 1.0289862429564953,
      "grad_norm": 0.677664577960968,
      "learning_rate": 1.3140091713623366e-05,
      "loss": 0.0689,
      "step": 20270
    },
    {
      "epoch": 1.0292400629473577,
      "grad_norm": 0.3639600872993469,
      "learning_rate": 1.313839958035095e-05,
      "loss": 0.0561,
      "step": 20275
    },
    {
      "epoch": 1.0294938829382203,
      "grad_norm": 0.29225844144821167,
      "learning_rate": 1.3136707447078533e-05,
      "loss": 0.0551,
      "step": 20280
    },
    {
      "epoch": 1.0297477029290827,
      "grad_norm": 0.43538472056388855,
      "learning_rate": 1.3135015313806117e-05,
      "loss": 0.0612,
      "step": 20285
    },
    {
      "epoch": 1.030001522919945,
      "grad_norm": 0.3201233744621277,
      "learning_rate": 1.3133323180533698e-05,
      "loss": 0.0577,
      "step": 20290
    },
    {
      "epoch": 1.0302553429108077,
      "grad_norm": 0.3105303645133972,
      "learning_rate": 1.3131631047261284e-05,
      "loss": 0.0593,
      "step": 20295
    },
    {
      "epoch": 1.03050916290167,
      "grad_norm": 0.4399048686027527,
      "learning_rate": 1.3129938913988867e-05,
      "loss": 0.0688,
      "step": 20300
    },
    {
      "epoch": 1.0307629828925327,
      "grad_norm": 0.45303788781166077,
      "learning_rate": 1.3128246780716449e-05,
      "loss": 0.0586,
      "step": 20305
    },
    {
      "epoch": 1.031016802883395,
      "grad_norm": 0.5587612986564636,
      "learning_rate": 1.3126554647444034e-05,
      "loss": 0.0668,
      "step": 20310
    },
    {
      "epoch": 1.0312706228742576,
      "grad_norm": 0.39321210980415344,
      "learning_rate": 1.3124862514171616e-05,
      "loss": 0.0588,
      "step": 20315
    },
    {
      "epoch": 1.03152444286512,
      "grad_norm": 0.2924768328666687,
      "learning_rate": 1.3123170380899201e-05,
      "loss": 0.0569,
      "step": 20320
    },
    {
      "epoch": 1.0317782628559826,
      "grad_norm": 0.32587504386901855,
      "learning_rate": 1.3121478247626785e-05,
      "loss": 0.0588,
      "step": 20325
    },
    {
      "epoch": 1.032032082846845,
      "grad_norm": 0.34906765818595886,
      "learning_rate": 1.3119786114354367e-05,
      "loss": 0.0669,
      "step": 20330
    },
    {
      "epoch": 1.0322859028377076,
      "grad_norm": 0.45067837834358215,
      "learning_rate": 1.3118093981081952e-05,
      "loss": 0.0627,
      "step": 20335
    },
    {
      "epoch": 1.03253972282857,
      "grad_norm": 0.43164461851119995,
      "learning_rate": 1.3116401847809534e-05,
      "loss": 0.0544,
      "step": 20340
    },
    {
      "epoch": 1.0327935428194324,
      "grad_norm": 0.343434602022171,
      "learning_rate": 1.3114709714537117e-05,
      "loss": 0.0596,
      "step": 20345
    },
    {
      "epoch": 1.033047362810295,
      "grad_norm": 0.43846994638442993,
      "learning_rate": 1.3113017581264703e-05,
      "loss": 0.0646,
      "step": 20350
    },
    {
      "epoch": 1.0333011828011573,
      "grad_norm": 0.3330884277820587,
      "learning_rate": 1.3111325447992285e-05,
      "loss": 0.0568,
      "step": 20355
    },
    {
      "epoch": 1.03355500279202,
      "grad_norm": 0.5818292498588562,
      "learning_rate": 1.3109633314719868e-05,
      "loss": 0.0613,
      "step": 20360
    },
    {
      "epoch": 1.0338088227828823,
      "grad_norm": 0.37332767248153687,
      "learning_rate": 1.3107941181447452e-05,
      "loss": 0.06,
      "step": 20365
    },
    {
      "epoch": 1.034062642773745,
      "grad_norm": 0.3650437891483307,
      "learning_rate": 1.3106249048175035e-05,
      "loss": 0.059,
      "step": 20370
    },
    {
      "epoch": 1.0343164627646073,
      "grad_norm": 0.800799548625946,
      "learning_rate": 1.310455691490262e-05,
      "loss": 0.0664,
      "step": 20375
    },
    {
      "epoch": 1.03457028275547,
      "grad_norm": 0.4584999680519104,
      "learning_rate": 1.3102864781630202e-05,
      "loss": 0.06,
      "step": 20380
    },
    {
      "epoch": 1.0348241027463323,
      "grad_norm": 0.4071432948112488,
      "learning_rate": 1.3101172648357786e-05,
      "loss": 0.0786,
      "step": 20385
    },
    {
      "epoch": 1.0350779227371947,
      "grad_norm": 0.3009105622768402,
      "learning_rate": 1.309948051508537e-05,
      "loss": 0.0629,
      "step": 20390
    },
    {
      "epoch": 1.0353317427280573,
      "grad_norm": 0.573149561882019,
      "learning_rate": 1.3097788381812953e-05,
      "loss": 0.069,
      "step": 20395
    },
    {
      "epoch": 1.0355855627189197,
      "grad_norm": 0.8304667472839355,
      "learning_rate": 1.3096096248540535e-05,
      "loss": 0.0595,
      "step": 20400
    },
    {
      "epoch": 1.0358393827097823,
      "grad_norm": 0.4500064551830292,
      "learning_rate": 1.309440411526812e-05,
      "loss": 0.0642,
      "step": 20405
    },
    {
      "epoch": 1.0360932027006446,
      "grad_norm": 0.6188081502914429,
      "learning_rate": 1.3092711981995704e-05,
      "loss": 0.0581,
      "step": 20410
    },
    {
      "epoch": 1.0363470226915072,
      "grad_norm": 0.4538813829421997,
      "learning_rate": 1.3091019848723287e-05,
      "loss": 0.0652,
      "step": 20415
    },
    {
      "epoch": 1.0366008426823696,
      "grad_norm": 0.269491583108902,
      "learning_rate": 1.308932771545087e-05,
      "loss": 0.0541,
      "step": 20420
    },
    {
      "epoch": 1.0368546626732322,
      "grad_norm": 0.36252927780151367,
      "learning_rate": 1.3087635582178452e-05,
      "loss": 0.0587,
      "step": 20425
    },
    {
      "epoch": 1.0371084826640946,
      "grad_norm": 0.4088985025882721,
      "learning_rate": 1.3085943448906038e-05,
      "loss": 0.0726,
      "step": 20430
    },
    {
      "epoch": 1.037362302654957,
      "grad_norm": 0.4039810001850128,
      "learning_rate": 1.3084251315633621e-05,
      "loss": 0.0561,
      "step": 20435
    },
    {
      "epoch": 1.0376161226458196,
      "grad_norm": 0.36475130915641785,
      "learning_rate": 1.3082559182361203e-05,
      "loss": 0.0712,
      "step": 20440
    },
    {
      "epoch": 1.037869942636682,
      "grad_norm": 0.4266735017299652,
      "learning_rate": 1.3080867049088788e-05,
      "loss": 0.0593,
      "step": 20445
    },
    {
      "epoch": 1.0381237626275446,
      "grad_norm": 0.44501543045043945,
      "learning_rate": 1.307917491581637e-05,
      "loss": 0.0605,
      "step": 20450
    },
    {
      "epoch": 1.038377582618407,
      "grad_norm": 0.6565629839897156,
      "learning_rate": 1.3077482782543954e-05,
      "loss": 0.057,
      "step": 20455
    },
    {
      "epoch": 1.0386314026092696,
      "grad_norm": 0.33326682448387146,
      "learning_rate": 1.3075790649271537e-05,
      "loss": 0.081,
      "step": 20460
    },
    {
      "epoch": 1.038885222600132,
      "grad_norm": 0.3565109074115753,
      "learning_rate": 1.307409851599912e-05,
      "loss": 0.0618,
      "step": 20465
    },
    {
      "epoch": 1.0391390425909945,
      "grad_norm": 0.30127477645874023,
      "learning_rate": 1.3072406382726706e-05,
      "loss": 0.0595,
      "step": 20470
    },
    {
      "epoch": 1.039392862581857,
      "grad_norm": 0.39654025435447693,
      "learning_rate": 1.3070714249454288e-05,
      "loss": 0.0554,
      "step": 20475
    },
    {
      "epoch": 1.0396466825727195,
      "grad_norm": 0.4261183440685272,
      "learning_rate": 1.3069022116181871e-05,
      "loss": 0.0744,
      "step": 20480
    },
    {
      "epoch": 1.039900502563582,
      "grad_norm": 0.4532530605792999,
      "learning_rate": 1.3067329982909455e-05,
      "loss": 0.0598,
      "step": 20485
    },
    {
      "epoch": 1.0401543225544443,
      "grad_norm": 0.49771350622177124,
      "learning_rate": 1.3065637849637039e-05,
      "loss": 0.0674,
      "step": 20490
    },
    {
      "epoch": 1.040408142545307,
      "grad_norm": 0.49165821075439453,
      "learning_rate": 1.306394571636462e-05,
      "loss": 0.0655,
      "step": 20495
    },
    {
      "epoch": 1.0406619625361693,
      "grad_norm": 0.3307662010192871,
      "learning_rate": 1.3062253583092206e-05,
      "loss": 0.0587,
      "step": 20500
    },
    {
      "epoch": 1.0409157825270319,
      "grad_norm": 0.3834070861339569,
      "learning_rate": 1.306056144981979e-05,
      "loss": 0.0599,
      "step": 20505
    },
    {
      "epoch": 1.0411696025178943,
      "grad_norm": 0.4196530878543854,
      "learning_rate": 1.3058869316547373e-05,
      "loss": 0.058,
      "step": 20510
    },
    {
      "epoch": 1.0414234225087569,
      "grad_norm": 0.376595675945282,
      "learning_rate": 1.3057177183274956e-05,
      "loss": 0.0598,
      "step": 20515
    },
    {
      "epoch": 1.0416772424996192,
      "grad_norm": 0.4179707467556,
      "learning_rate": 1.3055485050002538e-05,
      "loss": 0.0647,
      "step": 20520
    },
    {
      "epoch": 1.0419310624904818,
      "grad_norm": 0.35341501235961914,
      "learning_rate": 1.3053792916730123e-05,
      "loss": 0.0506,
      "step": 20525
    },
    {
      "epoch": 1.0421848824813442,
      "grad_norm": 0.3597331941127777,
      "learning_rate": 1.3052100783457707e-05,
      "loss": 0.0671,
      "step": 20530
    },
    {
      "epoch": 1.0424387024722068,
      "grad_norm": 0.35214442014694214,
      "learning_rate": 1.3050408650185289e-05,
      "loss": 0.0501,
      "step": 20535
    },
    {
      "epoch": 1.0426925224630692,
      "grad_norm": 1.1911040544509888,
      "learning_rate": 1.3048716516912874e-05,
      "loss": 0.0683,
      "step": 20540
    },
    {
      "epoch": 1.0429463424539316,
      "grad_norm": 0.40806475281715393,
      "learning_rate": 1.3047024383640456e-05,
      "loss": 0.0652,
      "step": 20545
    },
    {
      "epoch": 1.0432001624447942,
      "grad_norm": 0.3692667782306671,
      "learning_rate": 1.304533225036804e-05,
      "loss": 0.0665,
      "step": 20550
    },
    {
      "epoch": 1.0434539824356566,
      "grad_norm": 0.580768883228302,
      "learning_rate": 1.3043640117095625e-05,
      "loss": 0.0652,
      "step": 20555
    },
    {
      "epoch": 1.0437078024265192,
      "grad_norm": 0.32445693016052246,
      "learning_rate": 1.3041947983823206e-05,
      "loss": 0.0599,
      "step": 20560
    },
    {
      "epoch": 1.0439616224173816,
      "grad_norm": 0.5763265490531921,
      "learning_rate": 1.3040255850550792e-05,
      "loss": 0.0694,
      "step": 20565
    },
    {
      "epoch": 1.0442154424082442,
      "grad_norm": 0.33058294653892517,
      "learning_rate": 1.3038563717278374e-05,
      "loss": 0.0563,
      "step": 20570
    },
    {
      "epoch": 1.0444692623991065,
      "grad_norm": 0.26986175775527954,
      "learning_rate": 1.3036871584005957e-05,
      "loss": 0.0542,
      "step": 20575
    },
    {
      "epoch": 1.0447230823899691,
      "grad_norm": 0.3680388331413269,
      "learning_rate": 1.3035179450733542e-05,
      "loss": 0.0613,
      "step": 20580
    },
    {
      "epoch": 1.0449769023808315,
      "grad_norm": 0.3555222153663635,
      "learning_rate": 1.3033487317461124e-05,
      "loss": 0.0608,
      "step": 20585
    },
    {
      "epoch": 1.045230722371694,
      "grad_norm": 0.29103875160217285,
      "learning_rate": 1.3031795184188708e-05,
      "loss": 0.0479,
      "step": 20590
    },
    {
      "epoch": 1.0454845423625565,
      "grad_norm": 0.36284422874450684,
      "learning_rate": 1.3030103050916291e-05,
      "loss": 0.0546,
      "step": 20595
    },
    {
      "epoch": 1.0457383623534189,
      "grad_norm": 0.2850485146045685,
      "learning_rate": 1.3028410917643875e-05,
      "loss": 0.0633,
      "step": 20600
    },
    {
      "epoch": 1.0459921823442815,
      "grad_norm": 0.37116739153862,
      "learning_rate": 1.3026718784371457e-05,
      "loss": 0.0618,
      "step": 20605
    },
    {
      "epoch": 1.0462460023351439,
      "grad_norm": 0.41948822140693665,
      "learning_rate": 1.3025026651099042e-05,
      "loss": 0.0711,
      "step": 20610
    },
    {
      "epoch": 1.0464998223260065,
      "grad_norm": 0.3508945107460022,
      "learning_rate": 1.3023334517826625e-05,
      "loss": 0.0549,
      "step": 20615
    },
    {
      "epoch": 1.0467536423168688,
      "grad_norm": 0.46695828437805176,
      "learning_rate": 1.3021642384554209e-05,
      "loss": 0.0566,
      "step": 20620
    },
    {
      "epoch": 1.0470074623077315,
      "grad_norm": 0.4364718496799469,
      "learning_rate": 1.3019950251281793e-05,
      "loss": 0.0669,
      "step": 20625
    },
    {
      "epoch": 1.0472612822985938,
      "grad_norm": 0.26235929131507874,
      "learning_rate": 1.3018258118009374e-05,
      "loss": 0.06,
      "step": 20630
    },
    {
      "epoch": 1.0475151022894562,
      "grad_norm": 0.36236390471458435,
      "learning_rate": 1.301656598473696e-05,
      "loss": 0.0602,
      "step": 20635
    },
    {
      "epoch": 1.0477689222803188,
      "grad_norm": 0.5523089170455933,
      "learning_rate": 1.3014873851464541e-05,
      "loss": 0.0604,
      "step": 20640
    },
    {
      "epoch": 1.0480227422711812,
      "grad_norm": 2.3140509128570557,
      "learning_rate": 1.3013181718192125e-05,
      "loss": 0.0675,
      "step": 20645
    },
    {
      "epoch": 1.0482765622620438,
      "grad_norm": 0.2716338634490967,
      "learning_rate": 1.301148958491971e-05,
      "loss": 0.0613,
      "step": 20650
    },
    {
      "epoch": 1.0485303822529062,
      "grad_norm": 0.38538867235183716,
      "learning_rate": 1.3009797451647292e-05,
      "loss": 0.0613,
      "step": 20655
    },
    {
      "epoch": 1.0487842022437688,
      "grad_norm": 0.8486253619194031,
      "learning_rate": 1.3008105318374877e-05,
      "loss": 0.0683,
      "step": 20660
    },
    {
      "epoch": 1.0490380222346312,
      "grad_norm": 0.8067624568939209,
      "learning_rate": 1.300641318510246e-05,
      "loss": 0.0569,
      "step": 20665
    },
    {
      "epoch": 1.0492918422254938,
      "grad_norm": 0.36104530096054077,
      "learning_rate": 1.3004721051830043e-05,
      "loss": 0.0571,
      "step": 20670
    },
    {
      "epoch": 1.0495456622163561,
      "grad_norm": 0.3068177402019501,
      "learning_rate": 1.3003028918557628e-05,
      "loss": 0.0574,
      "step": 20675
    },
    {
      "epoch": 1.0497994822072187,
      "grad_norm": 0.41249674558639526,
      "learning_rate": 1.300133678528521e-05,
      "loss": 0.0616,
      "step": 20680
    },
    {
      "epoch": 1.0500533021980811,
      "grad_norm": 0.3544803857803345,
      "learning_rate": 1.2999644652012793e-05,
      "loss": 0.0664,
      "step": 20685
    },
    {
      "epoch": 1.0503071221889435,
      "grad_norm": 0.2459775060415268,
      "learning_rate": 1.2997952518740377e-05,
      "loss": 0.0602,
      "step": 20690
    },
    {
      "epoch": 1.0505609421798061,
      "grad_norm": 0.5989146828651428,
      "learning_rate": 1.299626038546796e-05,
      "loss": 0.0552,
      "step": 20695
    },
    {
      "epoch": 1.0508147621706685,
      "grad_norm": 0.44064995646476746,
      "learning_rate": 1.2994568252195542e-05,
      "loss": 0.0494,
      "step": 20700
    },
    {
      "epoch": 1.051068582161531,
      "grad_norm": 0.34163224697113037,
      "learning_rate": 1.2992876118923128e-05,
      "loss": 0.053,
      "step": 20705
    },
    {
      "epoch": 1.0513224021523935,
      "grad_norm": 0.3708654046058655,
      "learning_rate": 1.2991183985650711e-05,
      "loss": 0.054,
      "step": 20710
    },
    {
      "epoch": 1.051576222143256,
      "grad_norm": 0.2843742072582245,
      "learning_rate": 1.2989491852378295e-05,
      "loss": 0.058,
      "step": 20715
    },
    {
      "epoch": 1.0518300421341185,
      "grad_norm": 0.4104609489440918,
      "learning_rate": 1.2987799719105878e-05,
      "loss": 0.0633,
      "step": 20720
    },
    {
      "epoch": 1.052083862124981,
      "grad_norm": 0.32387226819992065,
      "learning_rate": 1.298610758583346e-05,
      "loss": 0.0572,
      "step": 20725
    },
    {
      "epoch": 1.0523376821158434,
      "grad_norm": 0.3362037241458893,
      "learning_rate": 1.2984415452561045e-05,
      "loss": 0.0565,
      "step": 20730
    },
    {
      "epoch": 1.0525915021067058,
      "grad_norm": 0.32576224207878113,
      "learning_rate": 1.2982723319288629e-05,
      "loss": 0.0654,
      "step": 20735
    },
    {
      "epoch": 1.0528453220975684,
      "grad_norm": 0.4229300618171692,
      "learning_rate": 1.298103118601621e-05,
      "loss": 0.0646,
      "step": 20740
    },
    {
      "epoch": 1.0530991420884308,
      "grad_norm": 0.4319082200527191,
      "learning_rate": 1.2979339052743796e-05,
      "loss": 0.0541,
      "step": 20745
    },
    {
      "epoch": 1.0533529620792934,
      "grad_norm": 0.421505868434906,
      "learning_rate": 1.2977646919471378e-05,
      "loss": 0.0597,
      "step": 20750
    },
    {
      "epoch": 1.0536067820701558,
      "grad_norm": 0.3763205409049988,
      "learning_rate": 1.2975954786198961e-05,
      "loss": 0.0492,
      "step": 20755
    },
    {
      "epoch": 1.0538606020610184,
      "grad_norm": 0.3560442626476288,
      "learning_rate": 1.2974262652926547e-05,
      "loss": 0.0588,
      "step": 20760
    },
    {
      "epoch": 1.0541144220518808,
      "grad_norm": 0.4291508197784424,
      "learning_rate": 1.2972570519654128e-05,
      "loss": 0.072,
      "step": 20765
    },
    {
      "epoch": 1.0543682420427434,
      "grad_norm": 0.3134581446647644,
      "learning_rate": 1.2970878386381714e-05,
      "loss": 0.0573,
      "step": 20770
    },
    {
      "epoch": 1.0546220620336058,
      "grad_norm": 0.3855588138103485,
      "learning_rate": 1.2969186253109295e-05,
      "loss": 0.0657,
      "step": 20775
    },
    {
      "epoch": 1.0548758820244681,
      "grad_norm": 0.2989584803581238,
      "learning_rate": 1.2967494119836879e-05,
      "loss": 0.0587,
      "step": 20780
    },
    {
      "epoch": 1.0551297020153307,
      "grad_norm": 0.4255363345146179,
      "learning_rate": 1.2965801986564464e-05,
      "loss": 0.0619,
      "step": 20785
    },
    {
      "epoch": 1.0553835220061931,
      "grad_norm": 0.4935469329357147,
      "learning_rate": 1.2964109853292046e-05,
      "loss": 0.0663,
      "step": 20790
    },
    {
      "epoch": 1.0556373419970557,
      "grad_norm": 0.2790074348449707,
      "learning_rate": 1.296241772001963e-05,
      "loss": 0.0493,
      "step": 20795
    },
    {
      "epoch": 1.055891161987918,
      "grad_norm": 0.5771297216415405,
      "learning_rate": 1.2960725586747213e-05,
      "loss": 0.0635,
      "step": 20800
    },
    {
      "epoch": 1.0561449819787807,
      "grad_norm": 0.49115124344825745,
      "learning_rate": 1.2959033453474797e-05,
      "loss": 0.0605,
      "step": 20805
    },
    {
      "epoch": 1.056398801969643,
      "grad_norm": 0.3402343690395355,
      "learning_rate": 1.2957341320202382e-05,
      "loss": 0.058,
      "step": 20810
    },
    {
      "epoch": 1.0566526219605057,
      "grad_norm": 0.39330241084098816,
      "learning_rate": 1.2955649186929964e-05,
      "loss": 0.0623,
      "step": 20815
    },
    {
      "epoch": 1.056906441951368,
      "grad_norm": 0.29627296328544617,
      "learning_rate": 1.2953957053657546e-05,
      "loss": 0.0552,
      "step": 20820
    },
    {
      "epoch": 1.0571602619422307,
      "grad_norm": 0.35211825370788574,
      "learning_rate": 1.2952264920385131e-05,
      "loss": 0.0632,
      "step": 20825
    },
    {
      "epoch": 1.057414081933093,
      "grad_norm": 0.4592018723487854,
      "learning_rate": 1.2950572787112714e-05,
      "loss": 0.0626,
      "step": 20830
    },
    {
      "epoch": 1.0576679019239554,
      "grad_norm": 0.5759520530700684,
      "learning_rate": 1.2948880653840296e-05,
      "loss": 0.0812,
      "step": 20835
    },
    {
      "epoch": 1.057921721914818,
      "grad_norm": 0.431813508272171,
      "learning_rate": 1.2947188520567882e-05,
      "loss": 0.057,
      "step": 20840
    },
    {
      "epoch": 1.0581755419056804,
      "grad_norm": 0.810368001461029,
      "learning_rate": 1.2945496387295463e-05,
      "loss": 0.0601,
      "step": 20845
    },
    {
      "epoch": 1.058429361896543,
      "grad_norm": 0.3972325623035431,
      "learning_rate": 1.2943804254023047e-05,
      "loss": 0.0614,
      "step": 20850
    },
    {
      "epoch": 1.0586831818874054,
      "grad_norm": 0.34748002886772156,
      "learning_rate": 1.2942112120750632e-05,
      "loss": 0.0631,
      "step": 20855
    },
    {
      "epoch": 1.058937001878268,
      "grad_norm": 0.5244632363319397,
      "learning_rate": 1.2940419987478214e-05,
      "loss": 0.0664,
      "step": 20860
    },
    {
      "epoch": 1.0591908218691304,
      "grad_norm": 0.4019574522972107,
      "learning_rate": 1.29387278542058e-05,
      "loss": 0.0622,
      "step": 20865
    },
    {
      "epoch": 1.059444641859993,
      "grad_norm": 0.43360385298728943,
      "learning_rate": 1.2937035720933381e-05,
      "loss": 0.0605,
      "step": 20870
    },
    {
      "epoch": 1.0596984618508554,
      "grad_norm": 0.37314683198928833,
      "learning_rate": 1.2935343587660965e-05,
      "loss": 0.0598,
      "step": 20875
    },
    {
      "epoch": 1.0599522818417177,
      "grad_norm": 0.3321194350719452,
      "learning_rate": 1.293365145438855e-05,
      "loss": 0.0598,
      "step": 20880
    },
    {
      "epoch": 1.0602061018325803,
      "grad_norm": 0.5326045155525208,
      "learning_rate": 1.2931959321116132e-05,
      "loss": 0.0623,
      "step": 20885
    },
    {
      "epoch": 1.0604599218234427,
      "grad_norm": 0.31733155250549316,
      "learning_rate": 1.2930267187843715e-05,
      "loss": 0.0526,
      "step": 20890
    },
    {
      "epoch": 1.0607137418143053,
      "grad_norm": 0.34186238050460815,
      "learning_rate": 1.2928575054571299e-05,
      "loss": 0.0633,
      "step": 20895
    },
    {
      "epoch": 1.0609675618051677,
      "grad_norm": 0.4136981964111328,
      "learning_rate": 1.2926882921298882e-05,
      "loss": 0.0663,
      "step": 20900
    },
    {
      "epoch": 1.0612213817960303,
      "grad_norm": 0.41344696283340454,
      "learning_rate": 1.2925190788026468e-05,
      "loss": 0.0592,
      "step": 20905
    },
    {
      "epoch": 1.0614752017868927,
      "grad_norm": 0.5743115544319153,
      "learning_rate": 1.292349865475405e-05,
      "loss": 0.067,
      "step": 20910
    },
    {
      "epoch": 1.0617290217777553,
      "grad_norm": 0.4706607460975647,
      "learning_rate": 1.2921806521481633e-05,
      "loss": 0.0581,
      "step": 20915
    },
    {
      "epoch": 1.0619828417686177,
      "grad_norm": 0.39083805680274963,
      "learning_rate": 1.2920114388209217e-05,
      "loss": 0.0726,
      "step": 20920
    },
    {
      "epoch": 1.0622366617594803,
      "grad_norm": 0.3401205241680145,
      "learning_rate": 1.29184222549368e-05,
      "loss": 0.061,
      "step": 20925
    },
    {
      "epoch": 1.0624904817503427,
      "grad_norm": 0.33678048849105835,
      "learning_rate": 1.2916730121664382e-05,
      "loss": 0.0606,
      "step": 20930
    },
    {
      "epoch": 1.062744301741205,
      "grad_norm": 0.5099437832832336,
      "learning_rate": 1.2915037988391967e-05,
      "loss": 0.071,
      "step": 20935
    },
    {
      "epoch": 1.0629981217320676,
      "grad_norm": 0.32160258293151855,
      "learning_rate": 1.291334585511955e-05,
      "loss": 0.0643,
      "step": 20940
    },
    {
      "epoch": 1.06325194172293,
      "grad_norm": 0.4883134067058563,
      "learning_rate": 1.2911653721847133e-05,
      "loss": 0.0616,
      "step": 20945
    },
    {
      "epoch": 1.0635057617137926,
      "grad_norm": 0.3711860179901123,
      "learning_rate": 1.2909961588574718e-05,
      "loss": 0.0621,
      "step": 20950
    },
    {
      "epoch": 1.063759581704655,
      "grad_norm": 0.49837297201156616,
      "learning_rate": 1.29082694553023e-05,
      "loss": 0.0691,
      "step": 20955
    },
    {
      "epoch": 1.0640134016955176,
      "grad_norm": 0.6001039147377014,
      "learning_rate": 1.2906577322029885e-05,
      "loss": 0.0606,
      "step": 20960
    },
    {
      "epoch": 1.06426722168638,
      "grad_norm": 0.32044780254364014,
      "learning_rate": 1.2904885188757468e-05,
      "loss": 0.0534,
      "step": 20965
    },
    {
      "epoch": 1.0645210416772426,
      "grad_norm": 0.30035600066185,
      "learning_rate": 1.290319305548505e-05,
      "loss": 0.054,
      "step": 20970
    },
    {
      "epoch": 1.064774861668105,
      "grad_norm": 0.343683660030365,
      "learning_rate": 1.2901500922212636e-05,
      "loss": 0.0637,
      "step": 20975
    },
    {
      "epoch": 1.0650286816589674,
      "grad_norm": 0.3978451192378998,
      "learning_rate": 1.2899808788940217e-05,
      "loss": 0.06,
      "step": 20980
    },
    {
      "epoch": 1.06528250164983,
      "grad_norm": 0.40602362155914307,
      "learning_rate": 1.2898116655667801e-05,
      "loss": 0.0693,
      "step": 20985
    },
    {
      "epoch": 1.0655363216406923,
      "grad_norm": 0.4250233471393585,
      "learning_rate": 1.2896424522395386e-05,
      "loss": 0.0615,
      "step": 20990
    },
    {
      "epoch": 1.065790141631555,
      "grad_norm": 0.3212973177433014,
      "learning_rate": 1.2894732389122968e-05,
      "loss": 0.0448,
      "step": 20995
    },
    {
      "epoch": 1.0660439616224173,
      "grad_norm": 0.4593718945980072,
      "learning_rate": 1.289304025585055e-05,
      "loss": 0.0589,
      "step": 21000
    },
    {
      "epoch": 1.06629778161328,
      "grad_norm": 0.41860365867614746,
      "learning_rate": 1.2891348122578135e-05,
      "loss": 0.0582,
      "step": 21005
    },
    {
      "epoch": 1.0665516016041423,
      "grad_norm": 0.32869061827659607,
      "learning_rate": 1.2889655989305719e-05,
      "loss": 0.0559,
      "step": 21010
    },
    {
      "epoch": 1.066805421595005,
      "grad_norm": 0.4083346724510193,
      "learning_rate": 1.2887963856033304e-05,
      "loss": 0.0536,
      "step": 21015
    },
    {
      "epoch": 1.0670592415858673,
      "grad_norm": 0.4695943295955658,
      "learning_rate": 1.2886271722760886e-05,
      "loss": 0.0629,
      "step": 21020
    },
    {
      "epoch": 1.06731306157673,
      "grad_norm": 0.42070719599723816,
      "learning_rate": 1.2884579589488468e-05,
      "loss": 0.0628,
      "step": 21025
    },
    {
      "epoch": 1.0675668815675923,
      "grad_norm": 0.32895520329475403,
      "learning_rate": 1.2882887456216053e-05,
      "loss": 0.0565,
      "step": 21030
    },
    {
      "epoch": 1.0678207015584547,
      "grad_norm": 0.44509220123291016,
      "learning_rate": 1.2881195322943636e-05,
      "loss": 0.0595,
      "step": 21035
    },
    {
      "epoch": 1.0680745215493173,
      "grad_norm": 0.3205646276473999,
      "learning_rate": 1.2879503189671218e-05,
      "loss": 0.0543,
      "step": 21040
    },
    {
      "epoch": 1.0683283415401796,
      "grad_norm": 0.6537548899650574,
      "learning_rate": 1.2877811056398803e-05,
      "loss": 0.0676,
      "step": 21045
    },
    {
      "epoch": 1.0685821615310422,
      "grad_norm": 0.40213271975517273,
      "learning_rate": 1.2876118923126385e-05,
      "loss": 0.0665,
      "step": 21050
    },
    {
      "epoch": 1.0688359815219046,
      "grad_norm": 0.39872047305107117,
      "learning_rate": 1.287442678985397e-05,
      "loss": 0.0633,
      "step": 21055
    },
    {
      "epoch": 1.0690898015127672,
      "grad_norm": 0.8748286366462708,
      "learning_rate": 1.2872734656581554e-05,
      "loss": 0.0584,
      "step": 21060
    },
    {
      "epoch": 1.0693436215036296,
      "grad_norm": 0.28194040060043335,
      "learning_rate": 1.2871042523309136e-05,
      "loss": 0.0569,
      "step": 21065
    },
    {
      "epoch": 1.0695974414944922,
      "grad_norm": 0.30738887190818787,
      "learning_rate": 1.2869350390036721e-05,
      "loss": 0.0523,
      "step": 21070
    },
    {
      "epoch": 1.0698512614853546,
      "grad_norm": 0.4939640462398529,
      "learning_rate": 1.2867658256764303e-05,
      "loss": 0.059,
      "step": 21075
    },
    {
      "epoch": 1.070105081476217,
      "grad_norm": 0.3448868691921234,
      "learning_rate": 1.2865966123491887e-05,
      "loss": 0.0564,
      "step": 21080
    },
    {
      "epoch": 1.0703589014670796,
      "grad_norm": 0.3513730764389038,
      "learning_rate": 1.2864273990219472e-05,
      "loss": 0.0567,
      "step": 21085
    },
    {
      "epoch": 1.070612721457942,
      "grad_norm": 0.4272534251213074,
      "learning_rate": 1.2862581856947054e-05,
      "loss": 0.054,
      "step": 21090
    },
    {
      "epoch": 1.0708665414488046,
      "grad_norm": 0.2933310568332672,
      "learning_rate": 1.2860889723674637e-05,
      "loss": 0.0505,
      "step": 21095
    },
    {
      "epoch": 1.071120361439667,
      "grad_norm": 0.4470325708389282,
      "learning_rate": 1.285919759040222e-05,
      "loss": 0.0735,
      "step": 21100
    },
    {
      "epoch": 1.0713741814305295,
      "grad_norm": 0.3551653325557709,
      "learning_rate": 1.2857505457129804e-05,
      "loss": 0.0541,
      "step": 21105
    },
    {
      "epoch": 1.071628001421392,
      "grad_norm": 0.6195331811904907,
      "learning_rate": 1.285581332385739e-05,
      "loss": 0.0681,
      "step": 21110
    },
    {
      "epoch": 1.0718818214122545,
      "grad_norm": 0.38556531071662903,
      "learning_rate": 1.2854121190584971e-05,
      "loss": 0.0612,
      "step": 21115
    },
    {
      "epoch": 1.072135641403117,
      "grad_norm": 0.6047544479370117,
      "learning_rate": 1.2852429057312555e-05,
      "loss": 0.0597,
      "step": 21120
    },
    {
      "epoch": 1.0723894613939793,
      "grad_norm": 0.48388174176216125,
      "learning_rate": 1.2850736924040139e-05,
      "loss": 0.0641,
      "step": 21125
    },
    {
      "epoch": 1.0726432813848419,
      "grad_norm": 0.8064355850219727,
      "learning_rate": 1.2849044790767722e-05,
      "loss": 0.0678,
      "step": 21130
    },
    {
      "epoch": 1.0728971013757043,
      "grad_norm": 0.3345118463039398,
      "learning_rate": 1.2847352657495304e-05,
      "loss": 0.0635,
      "step": 21135
    },
    {
      "epoch": 1.0731509213665669,
      "grad_norm": 0.3172535300254822,
      "learning_rate": 1.2845660524222889e-05,
      "loss": 0.065,
      "step": 21140
    },
    {
      "epoch": 1.0734047413574292,
      "grad_norm": 0.8722877502441406,
      "learning_rate": 1.2843968390950473e-05,
      "loss": 0.0571,
      "step": 21145
    },
    {
      "epoch": 1.0736585613482919,
      "grad_norm": 0.42107436060905457,
      "learning_rate": 1.2842276257678056e-05,
      "loss": 0.0537,
      "step": 21150
    },
    {
      "epoch": 1.0739123813391542,
      "grad_norm": 0.41451701521873474,
      "learning_rate": 1.284058412440564e-05,
      "loss": 0.0595,
      "step": 21155
    },
    {
      "epoch": 1.0741662013300168,
      "grad_norm": 0.4234960377216339,
      "learning_rate": 1.2838891991133222e-05,
      "loss": 0.0581,
      "step": 21160
    },
    {
      "epoch": 1.0744200213208792,
      "grad_norm": 0.40103062987327576,
      "learning_rate": 1.2837199857860807e-05,
      "loss": 0.0581,
      "step": 21165
    },
    {
      "epoch": 1.0746738413117418,
      "grad_norm": 0.35604149103164673,
      "learning_rate": 1.283550772458839e-05,
      "loss": 0.057,
      "step": 21170
    },
    {
      "epoch": 1.0749276613026042,
      "grad_norm": 0.4039483666419983,
      "learning_rate": 1.2833815591315972e-05,
      "loss": 0.061,
      "step": 21175
    },
    {
      "epoch": 1.0751814812934666,
      "grad_norm": 0.36171674728393555,
      "learning_rate": 1.2832123458043558e-05,
      "loss": 0.0592,
      "step": 21180
    },
    {
      "epoch": 1.0754353012843292,
      "grad_norm": 0.525554358959198,
      "learning_rate": 1.283043132477114e-05,
      "loss": 0.0607,
      "step": 21185
    },
    {
      "epoch": 1.0756891212751916,
      "grad_norm": 0.3412221670150757,
      "learning_rate": 1.2828739191498723e-05,
      "loss": 0.0593,
      "step": 21190
    },
    {
      "epoch": 1.0759429412660542,
      "grad_norm": 0.35413116216659546,
      "learning_rate": 1.2827047058226308e-05,
      "loss": 0.0608,
      "step": 21195
    },
    {
      "epoch": 1.0761967612569165,
      "grad_norm": 0.7104817032814026,
      "learning_rate": 1.282535492495389e-05,
      "loss": 0.0676,
      "step": 21200
    },
    {
      "epoch": 1.0764505812477791,
      "grad_norm": 0.4611401855945587,
      "learning_rate": 1.2823662791681475e-05,
      "loss": 0.065,
      "step": 21205
    },
    {
      "epoch": 1.0767044012386415,
      "grad_norm": 0.32466113567352295,
      "learning_rate": 1.2821970658409057e-05,
      "loss": 0.0638,
      "step": 21210
    },
    {
      "epoch": 1.0769582212295041,
      "grad_norm": 0.37902674078941345,
      "learning_rate": 1.282027852513664e-05,
      "loss": 0.0586,
      "step": 21215
    },
    {
      "epoch": 1.0772120412203665,
      "grad_norm": 0.29567191004753113,
      "learning_rate": 1.2818586391864226e-05,
      "loss": 0.0583,
      "step": 21220
    },
    {
      "epoch": 1.077465861211229,
      "grad_norm": 0.24273955821990967,
      "learning_rate": 1.2816894258591808e-05,
      "loss": 0.0586,
      "step": 21225
    },
    {
      "epoch": 1.0777196812020915,
      "grad_norm": 0.4052215814590454,
      "learning_rate": 1.281520212531939e-05,
      "loss": 0.0519,
      "step": 21230
    },
    {
      "epoch": 1.0779735011929539,
      "grad_norm": 0.43652230501174927,
      "learning_rate": 1.2813509992046975e-05,
      "loss": 0.0588,
      "step": 21235
    },
    {
      "epoch": 1.0782273211838165,
      "grad_norm": 0.5343736410140991,
      "learning_rate": 1.2811817858774558e-05,
      "loss": 0.0683,
      "step": 21240
    },
    {
      "epoch": 1.0784811411746789,
      "grad_norm": 0.43747490644454956,
      "learning_rate": 1.281012572550214e-05,
      "loss": 0.0556,
      "step": 21245
    },
    {
      "epoch": 1.0787349611655415,
      "grad_norm": 0.28066039085388184,
      "learning_rate": 1.2808433592229725e-05,
      "loss": 0.058,
      "step": 21250
    },
    {
      "epoch": 1.0789887811564038,
      "grad_norm": 0.5007389783859253,
      "learning_rate": 1.2806741458957307e-05,
      "loss": 0.0506,
      "step": 21255
    },
    {
      "epoch": 1.0792426011472664,
      "grad_norm": 0.3657475411891937,
      "learning_rate": 1.2805049325684893e-05,
      "loss": 0.066,
      "step": 21260
    },
    {
      "epoch": 1.0794964211381288,
      "grad_norm": 0.3738833963871002,
      "learning_rate": 1.2803357192412476e-05,
      "loss": 0.0601,
      "step": 21265
    },
    {
      "epoch": 1.0797502411289912,
      "grad_norm": 0.35183340311050415,
      "learning_rate": 1.2801665059140058e-05,
      "loss": 0.0672,
      "step": 21270
    },
    {
      "epoch": 1.0800040611198538,
      "grad_norm": 0.47940659523010254,
      "learning_rate": 1.2799972925867643e-05,
      "loss": 0.0623,
      "step": 21275
    },
    {
      "epoch": 1.0802578811107162,
      "grad_norm": 0.5233344435691833,
      "learning_rate": 1.2798280792595225e-05,
      "loss": 0.0598,
      "step": 21280
    },
    {
      "epoch": 1.0805117011015788,
      "grad_norm": 0.30529236793518066,
      "learning_rate": 1.2796588659322809e-05,
      "loss": 0.0479,
      "step": 21285
    },
    {
      "epoch": 1.0807655210924412,
      "grad_norm": 0.4038148522377014,
      "learning_rate": 1.2794896526050394e-05,
      "loss": 0.0736,
      "step": 21290
    },
    {
      "epoch": 1.0810193410833038,
      "grad_norm": 0.3613746464252472,
      "learning_rate": 1.2793204392777976e-05,
      "loss": 0.0547,
      "step": 21295
    },
    {
      "epoch": 1.0812731610741662,
      "grad_norm": 0.3199770450592041,
      "learning_rate": 1.2791512259505561e-05,
      "loss": 0.0582,
      "step": 21300
    },
    {
      "epoch": 1.0815269810650288,
      "grad_norm": 0.6598812937736511,
      "learning_rate": 1.2789820126233143e-05,
      "loss": 0.0638,
      "step": 21305
    },
    {
      "epoch": 1.0817808010558911,
      "grad_norm": 0.3184927999973297,
      "learning_rate": 1.2788127992960726e-05,
      "loss": 0.0535,
      "step": 21310
    },
    {
      "epoch": 1.0820346210467537,
      "grad_norm": 0.3598698675632477,
      "learning_rate": 1.2786435859688312e-05,
      "loss": 0.0534,
      "step": 21315
    },
    {
      "epoch": 1.0822884410376161,
      "grad_norm": 0.4094015657901764,
      "learning_rate": 1.2784743726415893e-05,
      "loss": 0.0479,
      "step": 21320
    },
    {
      "epoch": 1.0825422610284785,
      "grad_norm": 0.9048264026641846,
      "learning_rate": 1.2783051593143477e-05,
      "loss": 0.0632,
      "step": 21325
    },
    {
      "epoch": 1.082796081019341,
      "grad_norm": 0.4209839701652527,
      "learning_rate": 1.278135945987106e-05,
      "loss": 0.0714,
      "step": 21330
    },
    {
      "epoch": 1.0830499010102035,
      "grad_norm": 0.4597996771335602,
      "learning_rate": 1.2779667326598644e-05,
      "loss": 0.0731,
      "step": 21335
    },
    {
      "epoch": 1.083303721001066,
      "grad_norm": 0.29909446835517883,
      "learning_rate": 1.2777975193326226e-05,
      "loss": 0.0644,
      "step": 21340
    },
    {
      "epoch": 1.0835575409919285,
      "grad_norm": 0.4894222617149353,
      "learning_rate": 1.2776283060053811e-05,
      "loss": 0.0624,
      "step": 21345
    },
    {
      "epoch": 1.083811360982791,
      "grad_norm": 0.38440021872520447,
      "learning_rate": 1.2774590926781395e-05,
      "loss": 0.0657,
      "step": 21350
    },
    {
      "epoch": 1.0840651809736535,
      "grad_norm": 0.4377933144569397,
      "learning_rate": 1.2772898793508978e-05,
      "loss": 0.0567,
      "step": 21355
    },
    {
      "epoch": 1.084319000964516,
      "grad_norm": 0.3647824227809906,
      "learning_rate": 1.2771206660236562e-05,
      "loss": 0.0674,
      "step": 21360
    },
    {
      "epoch": 1.0845728209553784,
      "grad_norm": 0.3673681616783142,
      "learning_rate": 1.2769514526964144e-05,
      "loss": 0.0582,
      "step": 21365
    },
    {
      "epoch": 1.084826640946241,
      "grad_norm": 0.7054483890533447,
      "learning_rate": 1.2767822393691729e-05,
      "loss": 0.0566,
      "step": 21370
    },
    {
      "epoch": 1.0850804609371034,
      "grad_norm": 0.4479354918003082,
      "learning_rate": 1.2766130260419312e-05,
      "loss": 0.0625,
      "step": 21375
    },
    {
      "epoch": 1.0853342809279658,
      "grad_norm": 0.34682953357696533,
      "learning_rate": 1.2764438127146894e-05,
      "loss": 0.0578,
      "step": 21380
    },
    {
      "epoch": 1.0855881009188284,
      "grad_norm": 0.4050367772579193,
      "learning_rate": 1.276274599387448e-05,
      "loss": 0.0532,
      "step": 21385
    },
    {
      "epoch": 1.0858419209096908,
      "grad_norm": 0.3523172438144684,
      "learning_rate": 1.2761053860602061e-05,
      "loss": 0.0578,
      "step": 21390
    },
    {
      "epoch": 1.0860957409005534,
      "grad_norm": 0.34507274627685547,
      "learning_rate": 1.2759361727329647e-05,
      "loss": 0.0622,
      "step": 21395
    },
    {
      "epoch": 1.0863495608914158,
      "grad_norm": 0.29645681381225586,
      "learning_rate": 1.275766959405723e-05,
      "loss": 0.0599,
      "step": 21400
    },
    {
      "epoch": 1.0866033808822784,
      "grad_norm": 0.36118993163108826,
      "learning_rate": 1.2755977460784812e-05,
      "loss": 0.0571,
      "step": 21405
    },
    {
      "epoch": 1.0868572008731407,
      "grad_norm": 0.5020495653152466,
      "learning_rate": 1.2754285327512397e-05,
      "loss": 0.0638,
      "step": 21410
    },
    {
      "epoch": 1.0871110208640034,
      "grad_norm": 0.32721999287605286,
      "learning_rate": 1.2752593194239979e-05,
      "loss": 0.0616,
      "step": 21415
    },
    {
      "epoch": 1.0873648408548657,
      "grad_norm": 0.38569194078445435,
      "learning_rate": 1.2750901060967563e-05,
      "loss": 0.0527,
      "step": 21420
    },
    {
      "epoch": 1.0876186608457281,
      "grad_norm": 0.6979565620422363,
      "learning_rate": 1.2749208927695148e-05,
      "loss": 0.0524,
      "step": 21425
    },
    {
      "epoch": 1.0878724808365907,
      "grad_norm": 0.4546845555305481,
      "learning_rate": 1.274751679442273e-05,
      "loss": 0.0601,
      "step": 21430
    },
    {
      "epoch": 1.088126300827453,
      "grad_norm": 0.46561935544013977,
      "learning_rate": 1.2745824661150311e-05,
      "loss": 0.0587,
      "step": 21435
    },
    {
      "epoch": 1.0883801208183157,
      "grad_norm": 0.32504963874816895,
      "learning_rate": 1.2744132527877897e-05,
      "loss": 0.0474,
      "step": 21440
    },
    {
      "epoch": 1.088633940809178,
      "grad_norm": 0.4292490482330322,
      "learning_rate": 1.274244039460548e-05,
      "loss": 0.054,
      "step": 21445
    },
    {
      "epoch": 1.0888877608000407,
      "grad_norm": 0.3140452802181244,
      "learning_rate": 1.2740748261333064e-05,
      "loss": 0.0709,
      "step": 21450
    },
    {
      "epoch": 1.089141580790903,
      "grad_norm": 0.4569678008556366,
      "learning_rate": 1.2739056128060647e-05,
      "loss": 0.064,
      "step": 21455
    },
    {
      "epoch": 1.0893954007817657,
      "grad_norm": 0.34585288166999817,
      "learning_rate": 1.273736399478823e-05,
      "loss": 0.056,
      "step": 21460
    },
    {
      "epoch": 1.089649220772628,
      "grad_norm": 0.35902848839759827,
      "learning_rate": 1.2735671861515814e-05,
      "loss": 0.0487,
      "step": 21465
    },
    {
      "epoch": 1.0899030407634904,
      "grad_norm": 0.40261754393577576,
      "learning_rate": 1.2733979728243398e-05,
      "loss": 0.0532,
      "step": 21470
    },
    {
      "epoch": 1.090156860754353,
      "grad_norm": 0.3371613323688507,
      "learning_rate": 1.273228759497098e-05,
      "loss": 0.0689,
      "step": 21475
    },
    {
      "epoch": 1.0904106807452154,
      "grad_norm": 0.3712048828601837,
      "learning_rate": 1.2730595461698565e-05,
      "loss": 0.064,
      "step": 21480
    },
    {
      "epoch": 1.090664500736078,
      "grad_norm": 0.39349299669265747,
      "learning_rate": 1.2728903328426147e-05,
      "loss": 0.0587,
      "step": 21485
    },
    {
      "epoch": 1.0909183207269404,
      "grad_norm": 0.6331197023391724,
      "learning_rate": 1.272721119515373e-05,
      "loss": 0.0638,
      "step": 21490
    },
    {
      "epoch": 1.091172140717803,
      "grad_norm": 0.28697672486305237,
      "learning_rate": 1.2725519061881316e-05,
      "loss": 0.0682,
      "step": 21495
    },
    {
      "epoch": 1.0914259607086654,
      "grad_norm": 0.48400163650512695,
      "learning_rate": 1.2723826928608898e-05,
      "loss": 0.0563,
      "step": 21500
    },
    {
      "epoch": 1.091679780699528,
      "grad_norm": 0.38245150446891785,
      "learning_rate": 1.2722134795336483e-05,
      "loss": 0.0541,
      "step": 21505
    },
    {
      "epoch": 1.0919336006903904,
      "grad_norm": 0.399664044380188,
      "learning_rate": 1.2720442662064065e-05,
      "loss": 0.0572,
      "step": 21510
    },
    {
      "epoch": 1.092187420681253,
      "grad_norm": 0.385906457901001,
      "learning_rate": 1.2718750528791648e-05,
      "loss": 0.0661,
      "step": 21515
    },
    {
      "epoch": 1.0924412406721153,
      "grad_norm": 0.5282519459724426,
      "learning_rate": 1.2717058395519233e-05,
      "loss": 0.05,
      "step": 21520
    },
    {
      "epoch": 1.0926950606629777,
      "grad_norm": 0.35903728008270264,
      "learning_rate": 1.2715366262246815e-05,
      "loss": 0.0623,
      "step": 21525
    },
    {
      "epoch": 1.0929488806538403,
      "grad_norm": 0.33200833201408386,
      "learning_rate": 1.2713674128974399e-05,
      "loss": 0.0585,
      "step": 21530
    },
    {
      "epoch": 1.0932027006447027,
      "grad_norm": 0.5667392611503601,
      "learning_rate": 1.2711981995701982e-05,
      "loss": 0.0561,
      "step": 21535
    },
    {
      "epoch": 1.0934565206355653,
      "grad_norm": 0.49172818660736084,
      "learning_rate": 1.2710289862429566e-05,
      "loss": 0.0512,
      "step": 21540
    },
    {
      "epoch": 1.0937103406264277,
      "grad_norm": 0.31876128911972046,
      "learning_rate": 1.2708597729157151e-05,
      "loss": 0.0592,
      "step": 21545
    },
    {
      "epoch": 1.0939641606172903,
      "grad_norm": 0.348209023475647,
      "learning_rate": 1.2706905595884733e-05,
      "loss": 0.054,
      "step": 21550
    },
    {
      "epoch": 1.0942179806081527,
      "grad_norm": 1.1828866004943848,
      "learning_rate": 1.2705213462612317e-05,
      "loss": 0.0622,
      "step": 21555
    },
    {
      "epoch": 1.0944718005990153,
      "grad_norm": 0.376403272151947,
      "learning_rate": 1.27035213293399e-05,
      "loss": 0.0693,
      "step": 21560
    },
    {
      "epoch": 1.0947256205898777,
      "grad_norm": 0.35962969064712524,
      "learning_rate": 1.2701829196067484e-05,
      "loss": 0.0499,
      "step": 21565
    },
    {
      "epoch": 1.09497944058074,
      "grad_norm": 0.3168092668056488,
      "learning_rate": 1.2700137062795066e-05,
      "loss": 0.0544,
      "step": 21570
    },
    {
      "epoch": 1.0952332605716026,
      "grad_norm": 0.7175357341766357,
      "learning_rate": 1.269844492952265e-05,
      "loss": 0.059,
      "step": 21575
    },
    {
      "epoch": 1.095487080562465,
      "grad_norm": 0.3884638845920563,
      "learning_rate": 1.2696752796250234e-05,
      "loss": 0.0632,
      "step": 21580
    },
    {
      "epoch": 1.0957409005533276,
      "grad_norm": 0.40249037742614746,
      "learning_rate": 1.2695060662977816e-05,
      "loss": 0.0603,
      "step": 21585
    },
    {
      "epoch": 1.09599472054419,
      "grad_norm": 0.5622670650482178,
      "learning_rate": 1.2693368529705401e-05,
      "loss": 0.0582,
      "step": 21590
    },
    {
      "epoch": 1.0962485405350526,
      "grad_norm": 0.42979860305786133,
      "learning_rate": 1.2691676396432983e-05,
      "loss": 0.058,
      "step": 21595
    },
    {
      "epoch": 1.096502360525915,
      "grad_norm": 0.36912596225738525,
      "learning_rate": 1.2689984263160568e-05,
      "loss": 0.0677,
      "step": 21600
    },
    {
      "epoch": 1.0967561805167776,
      "grad_norm": 0.5511152744293213,
      "learning_rate": 1.2688292129888152e-05,
      "loss": 0.0566,
      "step": 21605
    },
    {
      "epoch": 1.09701000050764,
      "grad_norm": 0.4276013672351837,
      "learning_rate": 1.2686599996615734e-05,
      "loss": 0.0757,
      "step": 21610
    },
    {
      "epoch": 1.0972638204985024,
      "grad_norm": 0.34152284264564514,
      "learning_rate": 1.2684907863343319e-05,
      "loss": 0.0543,
      "step": 21615
    },
    {
      "epoch": 1.097517640489365,
      "grad_norm": 0.3993591070175171,
      "learning_rate": 1.2683215730070901e-05,
      "loss": 0.0513,
      "step": 21620
    },
    {
      "epoch": 1.0977714604802273,
      "grad_norm": 0.2898729741573334,
      "learning_rate": 1.2681523596798484e-05,
      "loss": 0.0559,
      "step": 21625
    },
    {
      "epoch": 1.09802528047109,
      "grad_norm": 0.3158964514732361,
      "learning_rate": 1.2679831463526068e-05,
      "loss": 0.0575,
      "step": 21630
    },
    {
      "epoch": 1.0982791004619523,
      "grad_norm": 0.6591548323631287,
      "learning_rate": 1.2678139330253652e-05,
      "loss": 0.0614,
      "step": 21635
    },
    {
      "epoch": 1.098532920452815,
      "grad_norm": 0.39295467734336853,
      "learning_rate": 1.2676447196981237e-05,
      "loss": 0.0593,
      "step": 21640
    },
    {
      "epoch": 1.0987867404436773,
      "grad_norm": 0.3043166995048523,
      "learning_rate": 1.2674755063708819e-05,
      "loss": 0.0524,
      "step": 21645
    },
    {
      "epoch": 1.09904056043454,
      "grad_norm": 0.5459681153297424,
      "learning_rate": 1.2673062930436402e-05,
      "loss": 0.0599,
      "step": 21650
    },
    {
      "epoch": 1.0992943804254023,
      "grad_norm": 0.3792338967323303,
      "learning_rate": 1.2671370797163986e-05,
      "loss": 0.0498,
      "step": 21655
    },
    {
      "epoch": 1.0995482004162649,
      "grad_norm": 0.35836130380630493,
      "learning_rate": 1.266967866389157e-05,
      "loss": 0.0672,
      "step": 21660
    },
    {
      "epoch": 1.0998020204071273,
      "grad_norm": 0.3934837579727173,
      "learning_rate": 1.2667986530619151e-05,
      "loss": 0.069,
      "step": 21665
    },
    {
      "epoch": 1.1000558403979896,
      "grad_norm": 0.374671071767807,
      "learning_rate": 1.2666294397346736e-05,
      "loss": 0.0657,
      "step": 21670
    },
    {
      "epoch": 1.1003096603888523,
      "grad_norm": 0.3873213231563568,
      "learning_rate": 1.266460226407432e-05,
      "loss": 0.0505,
      "step": 21675
    },
    {
      "epoch": 1.1005634803797146,
      "grad_norm": 0.29623717069625854,
      "learning_rate": 1.2662910130801902e-05,
      "loss": 0.0536,
      "step": 21680
    },
    {
      "epoch": 1.1008173003705772,
      "grad_norm": 0.320719450712204,
      "learning_rate": 1.2661217997529487e-05,
      "loss": 0.0545,
      "step": 21685
    },
    {
      "epoch": 1.1010711203614396,
      "grad_norm": 0.2868250906467438,
      "learning_rate": 1.2659525864257069e-05,
      "loss": 0.0587,
      "step": 21690
    },
    {
      "epoch": 1.1013249403523022,
      "grad_norm": 0.42279836535453796,
      "learning_rate": 1.2657833730984654e-05,
      "loss": 0.0527,
      "step": 21695
    },
    {
      "epoch": 1.1015787603431646,
      "grad_norm": 0.3950606882572174,
      "learning_rate": 1.2656141597712238e-05,
      "loss": 0.0573,
      "step": 21700
    },
    {
      "epoch": 1.1018325803340272,
      "grad_norm": 0.5181793570518494,
      "learning_rate": 1.265444946443982e-05,
      "loss": 0.0526,
      "step": 21705
    },
    {
      "epoch": 1.1020864003248896,
      "grad_norm": 0.5141026973724365,
      "learning_rate": 1.2652757331167405e-05,
      "loss": 0.0693,
      "step": 21710
    },
    {
      "epoch": 1.1023402203157522,
      "grad_norm": 0.4179568290710449,
      "learning_rate": 1.2651065197894987e-05,
      "loss": 0.0558,
      "step": 21715
    },
    {
      "epoch": 1.1025940403066146,
      "grad_norm": 0.37919455766677856,
      "learning_rate": 1.264937306462257e-05,
      "loss": 0.0566,
      "step": 21720
    },
    {
      "epoch": 1.102847860297477,
      "grad_norm": 0.3605843782424927,
      "learning_rate": 1.2647680931350155e-05,
      "loss": 0.0587,
      "step": 21725
    },
    {
      "epoch": 1.1031016802883395,
      "grad_norm": 0.28280141949653625,
      "learning_rate": 1.2645988798077737e-05,
      "loss": 0.0612,
      "step": 21730
    },
    {
      "epoch": 1.103355500279202,
      "grad_norm": 0.35269805788993835,
      "learning_rate": 1.264429666480532e-05,
      "loss": 0.0536,
      "step": 21735
    },
    {
      "epoch": 1.1036093202700645,
      "grad_norm": 0.35521480441093445,
      "learning_rate": 1.2642604531532904e-05,
      "loss": 0.0533,
      "step": 21740
    },
    {
      "epoch": 1.103863140260927,
      "grad_norm": 0.4273066222667694,
      "learning_rate": 1.2640912398260488e-05,
      "loss": 0.0609,
      "step": 21745
    },
    {
      "epoch": 1.1041169602517895,
      "grad_norm": 0.6088941097259521,
      "learning_rate": 1.2639220264988073e-05,
      "loss": 0.0587,
      "step": 21750
    },
    {
      "epoch": 1.104370780242652,
      "grad_norm": 0.28236907720565796,
      "learning_rate": 1.2637528131715655e-05,
      "loss": 0.063,
      "step": 21755
    },
    {
      "epoch": 1.1046246002335143,
      "grad_norm": 0.723622739315033,
      "learning_rate": 1.2635835998443239e-05,
      "loss": 0.0613,
      "step": 21760
    },
    {
      "epoch": 1.1048784202243769,
      "grad_norm": 0.43080398440361023,
      "learning_rate": 1.2634143865170822e-05,
      "loss": 0.0547,
      "step": 21765
    },
    {
      "epoch": 1.1051322402152393,
      "grad_norm": 0.41854366660118103,
      "learning_rate": 1.2632451731898406e-05,
      "loss": 0.0544,
      "step": 21770
    },
    {
      "epoch": 1.1053860602061019,
      "grad_norm": 0.26487666368484497,
      "learning_rate": 1.2630759598625987e-05,
      "loss": 0.0583,
      "step": 21775
    },
    {
      "epoch": 1.1056398801969642,
      "grad_norm": 0.5939835906028748,
      "learning_rate": 1.2629067465353573e-05,
      "loss": 0.0576,
      "step": 21780
    },
    {
      "epoch": 1.1058937001878268,
      "grad_norm": 0.3684999942779541,
      "learning_rate": 1.2627375332081156e-05,
      "loss": 0.0511,
      "step": 21785
    },
    {
      "epoch": 1.1061475201786892,
      "grad_norm": 0.3227483630180359,
      "learning_rate": 1.262568319880874e-05,
      "loss": 0.0518,
      "step": 21790
    },
    {
      "epoch": 1.1064013401695518,
      "grad_norm": 0.3155810534954071,
      "learning_rate": 1.2623991065536323e-05,
      "loss": 0.0575,
      "step": 21795
    },
    {
      "epoch": 1.1066551601604142,
      "grad_norm": 0.35092318058013916,
      "learning_rate": 1.2622298932263905e-05,
      "loss": 0.0554,
      "step": 21800
    },
    {
      "epoch": 1.1069089801512768,
      "grad_norm": 0.5898188352584839,
      "learning_rate": 1.262060679899149e-05,
      "loss": 0.06,
      "step": 21805
    },
    {
      "epoch": 1.1071628001421392,
      "grad_norm": 0.430916428565979,
      "learning_rate": 1.2618914665719072e-05,
      "loss": 0.0549,
      "step": 21810
    },
    {
      "epoch": 1.1074166201330016,
      "grad_norm": 0.5769869685173035,
      "learning_rate": 1.2617222532446656e-05,
      "loss": 0.0618,
      "step": 21815
    },
    {
      "epoch": 1.1076704401238642,
      "grad_norm": 0.3023453950881958,
      "learning_rate": 1.2615530399174241e-05,
      "loss": 0.0608,
      "step": 21820
    },
    {
      "epoch": 1.1079242601147266,
      "grad_norm": 0.7524684071540833,
      "learning_rate": 1.2613838265901823e-05,
      "loss": 0.0546,
      "step": 21825
    },
    {
      "epoch": 1.1081780801055892,
      "grad_norm": 0.3695290684700012,
      "learning_rate": 1.2612146132629406e-05,
      "loss": 0.0685,
      "step": 21830
    },
    {
      "epoch": 1.1084319000964515,
      "grad_norm": 0.29292744398117065,
      "learning_rate": 1.261045399935699e-05,
      "loss": 0.0589,
      "step": 21835
    },
    {
      "epoch": 1.1086857200873141,
      "grad_norm": 0.43530234694480896,
      "learning_rate": 1.2608761866084574e-05,
      "loss": 0.0537,
      "step": 21840
    },
    {
      "epoch": 1.1089395400781765,
      "grad_norm": 0.34705910086631775,
      "learning_rate": 1.2607069732812159e-05,
      "loss": 0.0596,
      "step": 21845
    },
    {
      "epoch": 1.1091933600690391,
      "grad_norm": 0.655457079410553,
      "learning_rate": 1.260537759953974e-05,
      "loss": 0.0512,
      "step": 21850
    },
    {
      "epoch": 1.1094471800599015,
      "grad_norm": 0.3433069586753845,
      "learning_rate": 1.2603685466267324e-05,
      "loss": 0.0456,
      "step": 21855
    },
    {
      "epoch": 1.109701000050764,
      "grad_norm": 0.5595205426216125,
      "learning_rate": 1.2601993332994908e-05,
      "loss": 0.0587,
      "step": 21860
    },
    {
      "epoch": 1.1099548200416265,
      "grad_norm": 0.4585667550563812,
      "learning_rate": 1.2600301199722491e-05,
      "loss": 0.0613,
      "step": 21865
    },
    {
      "epoch": 1.1102086400324889,
      "grad_norm": 0.41729608178138733,
      "learning_rate": 1.2598609066450073e-05,
      "loss": 0.0583,
      "step": 21870
    },
    {
      "epoch": 1.1104624600233515,
      "grad_norm": 1.0790683031082153,
      "learning_rate": 1.2596916933177658e-05,
      "loss": 0.0584,
      "step": 21875
    },
    {
      "epoch": 1.1107162800142139,
      "grad_norm": 0.43448856472969055,
      "learning_rate": 1.2595224799905242e-05,
      "loss": 0.0539,
      "step": 21880
    },
    {
      "epoch": 1.1109701000050765,
      "grad_norm": 0.3212340474128723,
      "learning_rate": 1.2593532666632824e-05,
      "loss": 0.0467,
      "step": 21885
    },
    {
      "epoch": 1.1112239199959388,
      "grad_norm": 0.32932937145233154,
      "learning_rate": 1.2591840533360409e-05,
      "loss": 0.0648,
      "step": 21890
    },
    {
      "epoch": 1.1114777399868014,
      "grad_norm": 0.4850245416164398,
      "learning_rate": 1.259014840008799e-05,
      "loss": 0.0658,
      "step": 21895
    },
    {
      "epoch": 1.1117315599776638,
      "grad_norm": 0.3544910252094269,
      "learning_rate": 1.2588456266815576e-05,
      "loss": 0.0594,
      "step": 21900
    },
    {
      "epoch": 1.1119853799685264,
      "grad_norm": 0.40757372975349426,
      "learning_rate": 1.258676413354316e-05,
      "loss": 0.0556,
      "step": 21905
    },
    {
      "epoch": 1.1122391999593888,
      "grad_norm": 0.4513925313949585,
      "learning_rate": 1.2585072000270741e-05,
      "loss": 0.0583,
      "step": 21910
    },
    {
      "epoch": 1.1124930199502512,
      "grad_norm": 0.4681965410709381,
      "learning_rate": 1.2583379866998327e-05,
      "loss": 0.0628,
      "step": 21915
    },
    {
      "epoch": 1.1127468399411138,
      "grad_norm": 0.3026393949985504,
      "learning_rate": 1.2581687733725909e-05,
      "loss": 0.0507,
      "step": 21920
    },
    {
      "epoch": 1.1130006599319762,
      "grad_norm": 0.6231611371040344,
      "learning_rate": 1.2579995600453492e-05,
      "loss": 0.0661,
      "step": 21925
    },
    {
      "epoch": 1.1132544799228388,
      "grad_norm": 0.4058896005153656,
      "learning_rate": 1.2578303467181077e-05,
      "loss": 0.0585,
      "step": 21930
    },
    {
      "epoch": 1.1135082999137011,
      "grad_norm": 0.4088999330997467,
      "learning_rate": 1.257661133390866e-05,
      "loss": 0.0544,
      "step": 21935
    },
    {
      "epoch": 1.1137621199045638,
      "grad_norm": 0.34594523906707764,
      "learning_rate": 1.2574919200636244e-05,
      "loss": 0.0553,
      "step": 21940
    },
    {
      "epoch": 1.1140159398954261,
      "grad_norm": 0.38814792037010193,
      "learning_rate": 1.2573227067363826e-05,
      "loss": 0.0665,
      "step": 21945
    },
    {
      "epoch": 1.1142697598862887,
      "grad_norm": 0.2994110584259033,
      "learning_rate": 1.257153493409141e-05,
      "loss": 0.0508,
      "step": 21950
    },
    {
      "epoch": 1.1145235798771511,
      "grad_norm": 0.4700678288936615,
      "learning_rate": 1.2569842800818995e-05,
      "loss": 0.0538,
      "step": 21955
    },
    {
      "epoch": 1.1147773998680135,
      "grad_norm": 0.3245387375354767,
      "learning_rate": 1.2568150667546577e-05,
      "loss": 0.0723,
      "step": 21960
    },
    {
      "epoch": 1.115031219858876,
      "grad_norm": 0.32570117712020874,
      "learning_rate": 1.256645853427416e-05,
      "loss": 0.0629,
      "step": 21965
    },
    {
      "epoch": 1.1152850398497385,
      "grad_norm": 0.39165782928466797,
      "learning_rate": 1.2564766401001744e-05,
      "loss": 0.0596,
      "step": 21970
    },
    {
      "epoch": 1.115538859840601,
      "grad_norm": 0.3810831904411316,
      "learning_rate": 1.2563074267729328e-05,
      "loss": 0.0539,
      "step": 21975
    },
    {
      "epoch": 1.1157926798314635,
      "grad_norm": 0.2843420207500458,
      "learning_rate": 1.256138213445691e-05,
      "loss": 0.0539,
      "step": 21980
    },
    {
      "epoch": 1.116046499822326,
      "grad_norm": 0.28111520409584045,
      "learning_rate": 1.2559690001184495e-05,
      "loss": 0.058,
      "step": 21985
    },
    {
      "epoch": 1.1163003198131884,
      "grad_norm": 0.37228140234947205,
      "learning_rate": 1.2557997867912076e-05,
      "loss": 0.0589,
      "step": 21990
    },
    {
      "epoch": 1.116554139804051,
      "grad_norm": 0.3925822079181671,
      "learning_rate": 1.2556305734639662e-05,
      "loss": 0.0695,
      "step": 21995
    },
    {
      "epoch": 1.1168079597949134,
      "grad_norm": 0.7498750686645508,
      "learning_rate": 1.2554613601367245e-05,
      "loss": 0.0677,
      "step": 22000
    },
    {
      "epoch": 1.117061779785776,
      "grad_norm": 0.3801857531070709,
      "learning_rate": 1.2552921468094827e-05,
      "loss": 0.0603,
      "step": 22005
    },
    {
      "epoch": 1.1173155997766384,
      "grad_norm": 0.35116496682167053,
      "learning_rate": 1.2551229334822412e-05,
      "loss": 0.0691,
      "step": 22010
    },
    {
      "epoch": 1.1175694197675008,
      "grad_norm": 0.427156001329422,
      "learning_rate": 1.2549537201549994e-05,
      "loss": 0.0603,
      "step": 22015
    },
    {
      "epoch": 1.1178232397583634,
      "grad_norm": 0.43225473165512085,
      "learning_rate": 1.2547845068277578e-05,
      "loss": 0.055,
      "step": 22020
    },
    {
      "epoch": 1.1180770597492258,
      "grad_norm": 0.38967064023017883,
      "learning_rate": 1.2546152935005163e-05,
      "loss": 0.0565,
      "step": 22025
    },
    {
      "epoch": 1.1183308797400884,
      "grad_norm": 1.0033137798309326,
      "learning_rate": 1.2544460801732745e-05,
      "loss": 0.055,
      "step": 22030
    },
    {
      "epoch": 1.1185846997309508,
      "grad_norm": 0.2806367874145508,
      "learning_rate": 1.254276866846033e-05,
      "loss": 0.0583,
      "step": 22035
    },
    {
      "epoch": 1.1188385197218134,
      "grad_norm": 0.33049389719963074,
      "learning_rate": 1.2541076535187912e-05,
      "loss": 0.0624,
      "step": 22040
    },
    {
      "epoch": 1.1190923397126757,
      "grad_norm": 0.2302655428647995,
      "learning_rate": 1.2539384401915495e-05,
      "loss": 0.0617,
      "step": 22045
    },
    {
      "epoch": 1.1193461597035383,
      "grad_norm": 0.35394585132598877,
      "learning_rate": 1.253769226864308e-05,
      "loss": 0.0618,
      "step": 22050
    },
    {
      "epoch": 1.1195999796944007,
      "grad_norm": 0.49913811683654785,
      "learning_rate": 1.2536000135370663e-05,
      "loss": 0.0574,
      "step": 22055
    },
    {
      "epoch": 1.1198537996852633,
      "grad_norm": 0.4379311501979828,
      "learning_rate": 1.2534308002098246e-05,
      "loss": 0.0618,
      "step": 22060
    },
    {
      "epoch": 1.1201076196761257,
      "grad_norm": 0.36873766779899597,
      "learning_rate": 1.253261586882583e-05,
      "loss": 0.0617,
      "step": 22065
    },
    {
      "epoch": 1.120361439666988,
      "grad_norm": 0.3143756091594696,
      "learning_rate": 1.2530923735553413e-05,
      "loss": 0.0616,
      "step": 22070
    },
    {
      "epoch": 1.1206152596578507,
      "grad_norm": 0.4979376494884491,
      "learning_rate": 1.2529231602280995e-05,
      "loss": 0.062,
      "step": 22075
    },
    {
      "epoch": 1.120869079648713,
      "grad_norm": 0.3125242292881012,
      "learning_rate": 1.252753946900858e-05,
      "loss": 0.0584,
      "step": 22080
    },
    {
      "epoch": 1.1211228996395757,
      "grad_norm": 0.3742777705192566,
      "learning_rate": 1.2525847335736164e-05,
      "loss": 0.0524,
      "step": 22085
    },
    {
      "epoch": 1.121376719630438,
      "grad_norm": 0.4746595621109009,
      "learning_rate": 1.2524155202463747e-05,
      "loss": 0.0571,
      "step": 22090
    },
    {
      "epoch": 1.1216305396213007,
      "grad_norm": 0.3313211500644684,
      "learning_rate": 1.2522463069191331e-05,
      "loss": 0.0469,
      "step": 22095
    },
    {
      "epoch": 1.121884359612163,
      "grad_norm": 0.31520578265190125,
      "learning_rate": 1.2520770935918913e-05,
      "loss": 0.0516,
      "step": 22100
    },
    {
      "epoch": 1.1221381796030254,
      "grad_norm": 0.31391963362693787,
      "learning_rate": 1.2519078802646498e-05,
      "loss": 0.0598,
      "step": 22105
    },
    {
      "epoch": 1.122391999593888,
      "grad_norm": 0.318359375,
      "learning_rate": 1.2517386669374082e-05,
      "loss": 0.0628,
      "step": 22110
    },
    {
      "epoch": 1.1226458195847504,
      "grad_norm": 0.40016159415245056,
      "learning_rate": 1.2515694536101663e-05,
      "loss": 0.0639,
      "step": 22115
    },
    {
      "epoch": 1.122899639575613,
      "grad_norm": 0.3543905019760132,
      "learning_rate": 1.2514002402829249e-05,
      "loss": 0.0572,
      "step": 22120
    },
    {
      "epoch": 1.1231534595664754,
      "grad_norm": 0.3604658544063568,
      "learning_rate": 1.251231026955683e-05,
      "loss": 0.052,
      "step": 22125
    },
    {
      "epoch": 1.123407279557338,
      "grad_norm": 0.3308972716331482,
      "learning_rate": 1.2510618136284414e-05,
      "loss": 0.0541,
      "step": 22130
    },
    {
      "epoch": 1.1236610995482004,
      "grad_norm": 0.3981793224811554,
      "learning_rate": 1.2508926003012e-05,
      "loss": 0.0618,
      "step": 22135
    },
    {
      "epoch": 1.123914919539063,
      "grad_norm": 0.2392720729112625,
      "learning_rate": 1.2507233869739581e-05,
      "loss": 0.0625,
      "step": 22140
    },
    {
      "epoch": 1.1241687395299254,
      "grad_norm": 0.35200855135917664,
      "learning_rate": 1.2505541736467166e-05,
      "loss": 0.054,
      "step": 22145
    },
    {
      "epoch": 1.124422559520788,
      "grad_norm": 0.3301455080509186,
      "learning_rate": 1.2503849603194748e-05,
      "loss": 0.0496,
      "step": 22150
    },
    {
      "epoch": 1.1246763795116503,
      "grad_norm": 0.32119038701057434,
      "learning_rate": 1.2502157469922332e-05,
      "loss": 0.062,
      "step": 22155
    },
    {
      "epoch": 1.1249301995025127,
      "grad_norm": 0.373477578163147,
      "learning_rate": 1.2500465336649917e-05,
      "loss": 0.0542,
      "step": 22160
    },
    {
      "epoch": 1.1251840194933753,
      "grad_norm": 0.2541143298149109,
      "learning_rate": 1.2498773203377499e-05,
      "loss": 0.0504,
      "step": 22165
    },
    {
      "epoch": 1.1254378394842377,
      "grad_norm": 0.3145160377025604,
      "learning_rate": 1.2497081070105082e-05,
      "loss": 0.0538,
      "step": 22170
    },
    {
      "epoch": 1.1256916594751003,
      "grad_norm": 1.1840285062789917,
      "learning_rate": 1.2495388936832666e-05,
      "loss": 0.0515,
      "step": 22175
    },
    {
      "epoch": 1.1259454794659627,
      "grad_norm": 0.36791080236434937,
      "learning_rate": 1.249369680356025e-05,
      "loss": 0.0567,
      "step": 22180
    },
    {
      "epoch": 1.1261992994568253,
      "grad_norm": 0.39223095774650574,
      "learning_rate": 1.2492004670287835e-05,
      "loss": 0.0577,
      "step": 22185
    },
    {
      "epoch": 1.1264531194476877,
      "grad_norm": 0.3730306625366211,
      "learning_rate": 1.2490312537015417e-05,
      "loss": 0.0533,
      "step": 22190
    },
    {
      "epoch": 1.1267069394385503,
      "grad_norm": 0.32315680384635925,
      "learning_rate": 1.2488620403742998e-05,
      "loss": 0.0573,
      "step": 22195
    },
    {
      "epoch": 1.1269607594294127,
      "grad_norm": 0.27425339818000793,
      "learning_rate": 1.2486928270470584e-05,
      "loss": 0.0487,
      "step": 22200
    },
    {
      "epoch": 1.1272145794202753,
      "grad_norm": 0.4643983244895935,
      "learning_rate": 1.2485236137198167e-05,
      "loss": 0.055,
      "step": 22205
    },
    {
      "epoch": 1.1274683994111376,
      "grad_norm": 0.4175454080104828,
      "learning_rate": 1.2483544003925749e-05,
      "loss": 0.0509,
      "step": 22210
    },
    {
      "epoch": 1.127722219402,
      "grad_norm": 0.5054285526275635,
      "learning_rate": 1.2481851870653334e-05,
      "loss": 0.0555,
      "step": 22215
    },
    {
      "epoch": 1.1279760393928626,
      "grad_norm": 0.44866254925727844,
      "learning_rate": 1.2480159737380916e-05,
      "loss": 0.0541,
      "step": 22220
    },
    {
      "epoch": 1.128229859383725,
      "grad_norm": 0.31655773520469666,
      "learning_rate": 1.24784676041085e-05,
      "loss": 0.0527,
      "step": 22225
    },
    {
      "epoch": 1.1284836793745876,
      "grad_norm": 0.3848085403442383,
      "learning_rate": 1.2476775470836085e-05,
      "loss": 0.0544,
      "step": 22230
    },
    {
      "epoch": 1.12873749936545,
      "grad_norm": 0.3327775001525879,
      "learning_rate": 1.2475083337563667e-05,
      "loss": 0.0494,
      "step": 22235
    },
    {
      "epoch": 1.1289913193563126,
      "grad_norm": 0.2371433675289154,
      "learning_rate": 1.2473391204291252e-05,
      "loss": 0.0548,
      "step": 22240
    },
    {
      "epoch": 1.129245139347175,
      "grad_norm": 0.3679490387439728,
      "learning_rate": 1.2471699071018834e-05,
      "loss": 0.0564,
      "step": 22245
    },
    {
      "epoch": 1.1294989593380373,
      "grad_norm": 0.31148838996887207,
      "learning_rate": 1.2470006937746417e-05,
      "loss": 0.0543,
      "step": 22250
    },
    {
      "epoch": 1.1297527793289,
      "grad_norm": 0.3589724004268646,
      "learning_rate": 1.2468314804474003e-05,
      "loss": 0.0596,
      "step": 22255
    },
    {
      "epoch": 1.1300065993197623,
      "grad_norm": 0.3857835531234741,
      "learning_rate": 1.2466622671201584e-05,
      "loss": 0.0639,
      "step": 22260
    },
    {
      "epoch": 1.130260419310625,
      "grad_norm": 0.4105840027332306,
      "learning_rate": 1.2464930537929168e-05,
      "loss": 0.0549,
      "step": 22265
    },
    {
      "epoch": 1.1305142393014873,
      "grad_norm": 0.3158581852912903,
      "learning_rate": 1.2463238404656752e-05,
      "loss": 0.0561,
      "step": 22270
    },
    {
      "epoch": 1.13076805929235,
      "grad_norm": 0.34058722853660583,
      "learning_rate": 1.2461546271384335e-05,
      "loss": 0.06,
      "step": 22275
    },
    {
      "epoch": 1.1310218792832123,
      "grad_norm": 0.6887267827987671,
      "learning_rate": 1.245985413811192e-05,
      "loss": 0.0546,
      "step": 22280
    },
    {
      "epoch": 1.131275699274075,
      "grad_norm": 0.35839903354644775,
      "learning_rate": 1.2458162004839502e-05,
      "loss": 0.0513,
      "step": 22285
    },
    {
      "epoch": 1.1315295192649373,
      "grad_norm": 0.37777021527290344,
      "learning_rate": 1.2456469871567086e-05,
      "loss": 0.0562,
      "step": 22290
    },
    {
      "epoch": 1.1317833392557999,
      "grad_norm": 0.3687744736671448,
      "learning_rate": 1.245477773829467e-05,
      "loss": 0.0577,
      "step": 22295
    },
    {
      "epoch": 1.1320371592466623,
      "grad_norm": 0.5017027854919434,
      "learning_rate": 1.2453085605022253e-05,
      "loss": 0.0603,
      "step": 22300
    },
    {
      "epoch": 1.1322909792375246,
      "grad_norm": 0.43287739157676697,
      "learning_rate": 1.2451393471749835e-05,
      "loss": 0.0681,
      "step": 22305
    },
    {
      "epoch": 1.1325447992283872,
      "grad_norm": 0.6787398457527161,
      "learning_rate": 1.244970133847742e-05,
      "loss": 0.049,
      "step": 22310
    },
    {
      "epoch": 1.1327986192192496,
      "grad_norm": 0.5929518938064575,
      "learning_rate": 1.2448009205205003e-05,
      "loss": 0.0557,
      "step": 22315
    },
    {
      "epoch": 1.1330524392101122,
      "grad_norm": 0.24244466423988342,
      "learning_rate": 1.2446317071932585e-05,
      "loss": 0.0502,
      "step": 22320
    },
    {
      "epoch": 1.1333062592009746,
      "grad_norm": 0.28957462310791016,
      "learning_rate": 1.244462493866017e-05,
      "loss": 0.0609,
      "step": 22325
    },
    {
      "epoch": 1.1335600791918372,
      "grad_norm": 0.3269180953502655,
      "learning_rate": 1.2442932805387752e-05,
      "loss": 0.0556,
      "step": 22330
    },
    {
      "epoch": 1.1338138991826996,
      "grad_norm": 0.41526785492897034,
      "learning_rate": 1.2441240672115338e-05,
      "loss": 0.0556,
      "step": 22335
    },
    {
      "epoch": 1.1340677191735622,
      "grad_norm": 0.3812713027000427,
      "learning_rate": 1.2439548538842921e-05,
      "loss": 0.0576,
      "step": 22340
    },
    {
      "epoch": 1.1343215391644246,
      "grad_norm": 0.3995012044906616,
      "learning_rate": 1.2437856405570503e-05,
      "loss": 0.0553,
      "step": 22345
    },
    {
      "epoch": 1.1345753591552872,
      "grad_norm": 0.6085227727890015,
      "learning_rate": 1.2436164272298088e-05,
      "loss": 0.0526,
      "step": 22350
    },
    {
      "epoch": 1.1348291791461496,
      "grad_norm": 0.3577788174152374,
      "learning_rate": 1.243447213902567e-05,
      "loss": 0.0568,
      "step": 22355
    },
    {
      "epoch": 1.135082999137012,
      "grad_norm": 0.3074178397655487,
      "learning_rate": 1.2432780005753254e-05,
      "loss": 0.0503,
      "step": 22360
    },
    {
      "epoch": 1.1353368191278745,
      "grad_norm": 0.30455219745635986,
      "learning_rate": 1.2431087872480839e-05,
      "loss": 0.0593,
      "step": 22365
    },
    {
      "epoch": 1.135590639118737,
      "grad_norm": 0.4841032028198242,
      "learning_rate": 1.242939573920842e-05,
      "loss": 0.0628,
      "step": 22370
    },
    {
      "epoch": 1.1358444591095995,
      "grad_norm": 0.24335934221744537,
      "learning_rate": 1.2427703605936003e-05,
      "loss": 0.0595,
      "step": 22375
    },
    {
      "epoch": 1.136098279100462,
      "grad_norm": 0.38525867462158203,
      "learning_rate": 1.2426011472663588e-05,
      "loss": 0.0683,
      "step": 22380
    },
    {
      "epoch": 1.1363520990913245,
      "grad_norm": 0.4575713276863098,
      "learning_rate": 1.2424319339391171e-05,
      "loss": 0.0688,
      "step": 22385
    },
    {
      "epoch": 1.1366059190821869,
      "grad_norm": 0.33208250999450684,
      "learning_rate": 1.2422627206118757e-05,
      "loss": 0.0542,
      "step": 22390
    },
    {
      "epoch": 1.1368597390730493,
      "grad_norm": 0.4368082880973816,
      "learning_rate": 1.2420935072846338e-05,
      "loss": 0.0597,
      "step": 22395
    },
    {
      "epoch": 1.1371135590639119,
      "grad_norm": 0.3381536900997162,
      "learning_rate": 1.241924293957392e-05,
      "loss": 0.0595,
      "step": 22400
    },
    {
      "epoch": 1.1373673790547745,
      "grad_norm": 0.40493860840797424,
      "learning_rate": 1.2417550806301506e-05,
      "loss": 0.0485,
      "step": 22405
    },
    {
      "epoch": 1.1376211990456369,
      "grad_norm": 0.3095938265323639,
      "learning_rate": 1.2415858673029089e-05,
      "loss": 0.0667,
      "step": 22410
    },
    {
      "epoch": 1.1378750190364992,
      "grad_norm": 0.35543668270111084,
      "learning_rate": 1.2414166539756671e-05,
      "loss": 0.0564,
      "step": 22415
    },
    {
      "epoch": 1.1381288390273618,
      "grad_norm": 0.35956689715385437,
      "learning_rate": 1.2412474406484256e-05,
      "loss": 0.0598,
      "step": 22420
    },
    {
      "epoch": 1.1383826590182242,
      "grad_norm": 0.33214613795280457,
      "learning_rate": 1.2410782273211838e-05,
      "loss": 0.0519,
      "step": 22425
    },
    {
      "epoch": 1.1386364790090868,
      "grad_norm": 0.43719860911369324,
      "learning_rate": 1.2409090139939423e-05,
      "loss": 0.0545,
      "step": 22430
    },
    {
      "epoch": 1.1388902989999492,
      "grad_norm": 0.37340477108955383,
      "learning_rate": 1.2407398006667007e-05,
      "loss": 0.0507,
      "step": 22435
    },
    {
      "epoch": 1.1391441189908118,
      "grad_norm": 0.2947007715702057,
      "learning_rate": 1.2405705873394589e-05,
      "loss": 0.0492,
      "step": 22440
    },
    {
      "epoch": 1.1393979389816742,
      "grad_norm": 0.3368941843509674,
      "learning_rate": 1.2404013740122174e-05,
      "loss": 0.0509,
      "step": 22445
    },
    {
      "epoch": 1.1396517589725366,
      "grad_norm": 0.36827924847602844,
      "learning_rate": 1.2402321606849756e-05,
      "loss": 0.066,
      "step": 22450
    },
    {
      "epoch": 1.1399055789633992,
      "grad_norm": 0.4480016827583313,
      "learning_rate": 1.240062947357734e-05,
      "loss": 0.0541,
      "step": 22455
    },
    {
      "epoch": 1.1401593989542615,
      "grad_norm": 0.39560964703559875,
      "learning_rate": 1.2398937340304925e-05,
      "loss": 0.0692,
      "step": 22460
    },
    {
      "epoch": 1.1404132189451242,
      "grad_norm": 0.59153813123703,
      "learning_rate": 1.2397245207032506e-05,
      "loss": 0.0477,
      "step": 22465
    },
    {
      "epoch": 1.1406670389359865,
      "grad_norm": 0.3255225419998169,
      "learning_rate": 1.239555307376009e-05,
      "loss": 0.0587,
      "step": 22470
    },
    {
      "epoch": 1.1409208589268491,
      "grad_norm": 0.33760660886764526,
      "learning_rate": 1.2393860940487674e-05,
      "loss": 0.0539,
      "step": 22475
    },
    {
      "epoch": 1.1411746789177115,
      "grad_norm": 1.3096075057983398,
      "learning_rate": 1.2392168807215257e-05,
      "loss": 0.0592,
      "step": 22480
    },
    {
      "epoch": 1.1414284989085741,
      "grad_norm": 0.37213486433029175,
      "learning_rate": 1.2390476673942842e-05,
      "loss": 0.0519,
      "step": 22485
    },
    {
      "epoch": 1.1416823188994365,
      "grad_norm": 0.3870268166065216,
      "learning_rate": 1.2388784540670424e-05,
      "loss": 0.0505,
      "step": 22490
    },
    {
      "epoch": 1.141936138890299,
      "grad_norm": 0.2687552869319916,
      "learning_rate": 1.2387092407398008e-05,
      "loss": 0.0592,
      "step": 22495
    },
    {
      "epoch": 1.1421899588811615,
      "grad_norm": 0.4848724603652954,
      "learning_rate": 1.2385400274125591e-05,
      "loss": 0.0663,
      "step": 22500
    },
    {
      "epoch": 1.1424437788720239,
      "grad_norm": 0.3139934539794922,
      "learning_rate": 1.2383708140853175e-05,
      "loss": 0.0502,
      "step": 22505
    },
    {
      "epoch": 1.1426975988628865,
      "grad_norm": 0.40067628026008606,
      "learning_rate": 1.2382016007580757e-05,
      "loss": 0.0549,
      "step": 22510
    },
    {
      "epoch": 1.1429514188537488,
      "grad_norm": 0.3154310882091522,
      "learning_rate": 1.2380323874308342e-05,
      "loss": 0.0561,
      "step": 22515
    },
    {
      "epoch": 1.1432052388446114,
      "grad_norm": 0.3540889322757721,
      "learning_rate": 1.2378631741035925e-05,
      "loss": 0.058,
      "step": 22520
    },
    {
      "epoch": 1.1434590588354738,
      "grad_norm": 0.36180123686790466,
      "learning_rate": 1.2376939607763509e-05,
      "loss": 0.0561,
      "step": 22525
    },
    {
      "epoch": 1.1437128788263364,
      "grad_norm": 0.3295062780380249,
      "learning_rate": 1.2375247474491093e-05,
      "loss": 0.0632,
      "step": 22530
    },
    {
      "epoch": 1.1439666988171988,
      "grad_norm": 1.8421072959899902,
      "learning_rate": 1.2373555341218674e-05,
      "loss": 0.0594,
      "step": 22535
    },
    {
      "epoch": 1.1442205188080614,
      "grad_norm": 0.5811465978622437,
      "learning_rate": 1.237186320794626e-05,
      "loss": 0.0654,
      "step": 22540
    },
    {
      "epoch": 1.1444743387989238,
      "grad_norm": 0.390346497297287,
      "learning_rate": 1.2370171074673843e-05,
      "loss": 0.0588,
      "step": 22545
    },
    {
      "epoch": 1.1447281587897864,
      "grad_norm": 0.43421465158462524,
      "learning_rate": 1.2368478941401425e-05,
      "loss": 0.05,
      "step": 22550
    },
    {
      "epoch": 1.1449819787806488,
      "grad_norm": 0.29329365491867065,
      "learning_rate": 1.236678680812901e-05,
      "loss": 0.0535,
      "step": 22555
    },
    {
      "epoch": 1.1452357987715112,
      "grad_norm": 0.3655465841293335,
      "learning_rate": 1.2365094674856592e-05,
      "loss": 0.0572,
      "step": 22560
    },
    {
      "epoch": 1.1454896187623738,
      "grad_norm": 0.44629284739494324,
      "learning_rate": 1.2363402541584176e-05,
      "loss": 0.0581,
      "step": 22565
    },
    {
      "epoch": 1.1457434387532361,
      "grad_norm": 0.5228719711303711,
      "learning_rate": 1.2361710408311761e-05,
      "loss": 0.0577,
      "step": 22570
    },
    {
      "epoch": 1.1459972587440987,
      "grad_norm": 0.4526336193084717,
      "learning_rate": 1.2360018275039343e-05,
      "loss": 0.0616,
      "step": 22575
    },
    {
      "epoch": 1.1462510787349611,
      "grad_norm": 1.1659830808639526,
      "learning_rate": 1.2358326141766928e-05,
      "loss": 0.0685,
      "step": 22580
    },
    {
      "epoch": 1.1465048987258237,
      "grad_norm": 0.3989523649215698,
      "learning_rate": 1.235663400849451e-05,
      "loss": 0.0617,
      "step": 22585
    },
    {
      "epoch": 1.146758718716686,
      "grad_norm": 0.5509122610092163,
      "learning_rate": 1.2354941875222093e-05,
      "loss": 0.05,
      "step": 22590
    },
    {
      "epoch": 1.1470125387075485,
      "grad_norm": 0.2802308201789856,
      "learning_rate": 1.2353249741949679e-05,
      "loss": 0.0541,
      "step": 22595
    },
    {
      "epoch": 1.147266358698411,
      "grad_norm": 0.40862369537353516,
      "learning_rate": 1.235155760867726e-05,
      "loss": 0.0579,
      "step": 22600
    },
    {
      "epoch": 1.1475201786892735,
      "grad_norm": 0.44067612290382385,
      "learning_rate": 1.2349865475404842e-05,
      "loss": 0.0558,
      "step": 22605
    },
    {
      "epoch": 1.147773998680136,
      "grad_norm": 0.3719422519207001,
      "learning_rate": 1.2348173342132428e-05,
      "loss": 0.0517,
      "step": 22610
    },
    {
      "epoch": 1.1480278186709985,
      "grad_norm": 0.33248206973075867,
      "learning_rate": 1.2346481208860011e-05,
      "loss": 0.0583,
      "step": 22615
    },
    {
      "epoch": 1.148281638661861,
      "grad_norm": 0.3742452561855316,
      "learning_rate": 1.2344789075587593e-05,
      "loss": 0.0538,
      "step": 22620
    },
    {
      "epoch": 1.1485354586527234,
      "grad_norm": 0.31650519371032715,
      "learning_rate": 1.2343096942315178e-05,
      "loss": 0.051,
      "step": 22625
    },
    {
      "epoch": 1.148789278643586,
      "grad_norm": 0.4304707944393158,
      "learning_rate": 1.234140480904276e-05,
      "loss": 0.0506,
      "step": 22630
    },
    {
      "epoch": 1.1490430986344484,
      "grad_norm": 0.36866042017936707,
      "learning_rate": 1.2339712675770345e-05,
      "loss": 0.0643,
      "step": 22635
    },
    {
      "epoch": 1.149296918625311,
      "grad_norm": 0.3071114718914032,
      "learning_rate": 1.2338020542497929e-05,
      "loss": 0.0504,
      "step": 22640
    },
    {
      "epoch": 1.1495507386161734,
      "grad_norm": 0.39008423686027527,
      "learning_rate": 1.233632840922551e-05,
      "loss": 0.0561,
      "step": 22645
    },
    {
      "epoch": 1.1498045586070358,
      "grad_norm": 0.3482741713523865,
      "learning_rate": 1.2334636275953096e-05,
      "loss": 0.0525,
      "step": 22650
    },
    {
      "epoch": 1.1500583785978984,
      "grad_norm": 0.3156396150588989,
      "learning_rate": 1.2332944142680678e-05,
      "loss": 0.0567,
      "step": 22655
    },
    {
      "epoch": 1.1503121985887608,
      "grad_norm": 0.3933767080307007,
      "learning_rate": 1.2331252009408261e-05,
      "loss": 0.0626,
      "step": 22660
    },
    {
      "epoch": 1.1505660185796234,
      "grad_norm": 0.38870787620544434,
      "learning_rate": 1.2329559876135847e-05,
      "loss": 0.061,
      "step": 22665
    },
    {
      "epoch": 1.1508198385704858,
      "grad_norm": 0.3350406289100647,
      "learning_rate": 1.2327867742863428e-05,
      "loss": 0.0553,
      "step": 22670
    },
    {
      "epoch": 1.1510736585613484,
      "grad_norm": 0.6869404315948486,
      "learning_rate": 1.2326175609591014e-05,
      "loss": 0.0477,
      "step": 22675
    },
    {
      "epoch": 1.1513274785522107,
      "grad_norm": 0.5427852869033813,
      "learning_rate": 1.2324483476318595e-05,
      "loss": 0.0537,
      "step": 22680
    },
    {
      "epoch": 1.1515812985430733,
      "grad_norm": 0.42787012457847595,
      "learning_rate": 1.2322791343046179e-05,
      "loss": 0.0551,
      "step": 22685
    },
    {
      "epoch": 1.1518351185339357,
      "grad_norm": 0.3151043951511383,
      "learning_rate": 1.2321099209773764e-05,
      "loss": 0.0548,
      "step": 22690
    },
    {
      "epoch": 1.1520889385247983,
      "grad_norm": 0.3836911618709564,
      "learning_rate": 1.2319407076501346e-05,
      "loss": 0.0511,
      "step": 22695
    },
    {
      "epoch": 1.1523427585156607,
      "grad_norm": 0.30847352743148804,
      "learning_rate": 1.231771494322893e-05,
      "loss": 0.0569,
      "step": 22700
    },
    {
      "epoch": 1.152596578506523,
      "grad_norm": 0.3887520432472229,
      "learning_rate": 1.2316022809956513e-05,
      "loss": 0.0585,
      "step": 22705
    },
    {
      "epoch": 1.1528503984973857,
      "grad_norm": 0.3231756091117859,
      "learning_rate": 1.2314330676684097e-05,
      "loss": 0.0547,
      "step": 22710
    },
    {
      "epoch": 1.153104218488248,
      "grad_norm": 0.4676765203475952,
      "learning_rate": 1.2312638543411679e-05,
      "loss": 0.0527,
      "step": 22715
    },
    {
      "epoch": 1.1533580384791107,
      "grad_norm": 0.3112950026988983,
      "learning_rate": 1.2310946410139264e-05,
      "loss": 0.0489,
      "step": 22720
    },
    {
      "epoch": 1.153611858469973,
      "grad_norm": 0.25811097025871277,
      "learning_rate": 1.2309254276866847e-05,
      "loss": 0.0496,
      "step": 22725
    },
    {
      "epoch": 1.1538656784608357,
      "grad_norm": 0.3294421136379242,
      "learning_rate": 1.2307562143594431e-05,
      "loss": 0.0508,
      "step": 22730
    },
    {
      "epoch": 1.154119498451698,
      "grad_norm": 0.4806308448314667,
      "learning_rate": 1.2305870010322014e-05,
      "loss": 0.0732,
      "step": 22735
    },
    {
      "epoch": 1.1543733184425604,
      "grad_norm": 0.3521295189857483,
      "learning_rate": 1.2304177877049596e-05,
      "loss": 0.0516,
      "step": 22740
    },
    {
      "epoch": 1.154627138433423,
      "grad_norm": 0.3540806472301483,
      "learning_rate": 1.2302485743777182e-05,
      "loss": 0.0489,
      "step": 22745
    },
    {
      "epoch": 1.1548809584242856,
      "grad_norm": 0.38654571771621704,
      "learning_rate": 1.2300793610504765e-05,
      "loss": 0.0492,
      "step": 22750
    },
    {
      "epoch": 1.155134778415148,
      "grad_norm": 0.36057183146476746,
      "learning_rate": 1.2299101477232347e-05,
      "loss": 0.0521,
      "step": 22755
    },
    {
      "epoch": 1.1553885984060104,
      "grad_norm": 0.3039451241493225,
      "learning_rate": 1.2297409343959932e-05,
      "loss": 0.0605,
      "step": 22760
    },
    {
      "epoch": 1.155642418396873,
      "grad_norm": 0.2759784758090973,
      "learning_rate": 1.2295717210687514e-05,
      "loss": 0.0529,
      "step": 22765
    },
    {
      "epoch": 1.1558962383877354,
      "grad_norm": 0.579740583896637,
      "learning_rate": 1.22940250774151e-05,
      "loss": 0.0673,
      "step": 22770
    },
    {
      "epoch": 1.156150058378598,
      "grad_norm": 0.40537041425704956,
      "learning_rate": 1.2292332944142683e-05,
      "loss": 0.0514,
      "step": 22775
    },
    {
      "epoch": 1.1564038783694603,
      "grad_norm": 0.3935674726963043,
      "learning_rate": 1.2290640810870265e-05,
      "loss": 0.0556,
      "step": 22780
    },
    {
      "epoch": 1.156657698360323,
      "grad_norm": 0.7717337608337402,
      "learning_rate": 1.228894867759785e-05,
      "loss": 0.0569,
      "step": 22785
    },
    {
      "epoch": 1.1569115183511853,
      "grad_norm": 0.28389760851860046,
      "learning_rate": 1.2287256544325432e-05,
      "loss": 0.059,
      "step": 22790
    },
    {
      "epoch": 1.1571653383420477,
      "grad_norm": 0.7284798622131348,
      "learning_rate": 1.2285564411053015e-05,
      "loss": 0.0533,
      "step": 22795
    },
    {
      "epoch": 1.1574191583329103,
      "grad_norm": 0.42667490243911743,
      "learning_rate": 1.2283872277780599e-05,
      "loss": 0.0586,
      "step": 22800
    },
    {
      "epoch": 1.1576729783237727,
      "grad_norm": 0.3872958719730377,
      "learning_rate": 1.2282180144508182e-05,
      "loss": 0.0446,
      "step": 22805
    },
    {
      "epoch": 1.1579267983146353,
      "grad_norm": 0.396384060382843,
      "learning_rate": 1.2280488011235764e-05,
      "loss": 0.0546,
      "step": 22810
    },
    {
      "epoch": 1.1581806183054977,
      "grad_norm": 0.4082648456096649,
      "learning_rate": 1.227879587796335e-05,
      "loss": 0.0594,
      "step": 22815
    },
    {
      "epoch": 1.1584344382963603,
      "grad_norm": 0.6174493432044983,
      "learning_rate": 1.2277103744690933e-05,
      "loss": 0.0501,
      "step": 22820
    },
    {
      "epoch": 1.1586882582872227,
      "grad_norm": 0.4318661391735077,
      "learning_rate": 1.2275411611418517e-05,
      "loss": 0.0598,
      "step": 22825
    },
    {
      "epoch": 1.1589420782780853,
      "grad_norm": 0.30461791157722473,
      "learning_rate": 1.22737194781461e-05,
      "loss": 0.0557,
      "step": 22830
    },
    {
      "epoch": 1.1591958982689476,
      "grad_norm": 0.4471897780895233,
      "learning_rate": 1.2272027344873682e-05,
      "loss": 0.0568,
      "step": 22835
    },
    {
      "epoch": 1.1594497182598102,
      "grad_norm": 0.6687557697296143,
      "learning_rate": 1.2270335211601267e-05,
      "loss": 0.0613,
      "step": 22840
    },
    {
      "epoch": 1.1597035382506726,
      "grad_norm": 0.457183301448822,
      "learning_rate": 1.226864307832885e-05,
      "loss": 0.0554,
      "step": 22845
    },
    {
      "epoch": 1.159957358241535,
      "grad_norm": 0.3360708951950073,
      "learning_rate": 1.2266950945056433e-05,
      "loss": 0.0591,
      "step": 22850
    },
    {
      "epoch": 1.1602111782323976,
      "grad_norm": 0.5214487314224243,
      "learning_rate": 1.2265258811784018e-05,
      "loss": 0.0533,
      "step": 22855
    },
    {
      "epoch": 1.16046499822326,
      "grad_norm": 0.3773740828037262,
      "learning_rate": 1.22635666785116e-05,
      "loss": 0.0551,
      "step": 22860
    },
    {
      "epoch": 1.1607188182141226,
      "grad_norm": 0.40249964594841003,
      "learning_rate": 1.2261874545239183e-05,
      "loss": 0.056,
      "step": 22865
    },
    {
      "epoch": 1.160972638204985,
      "grad_norm": 0.30552637577056885,
      "learning_rate": 1.2260182411966768e-05,
      "loss": 0.0502,
      "step": 22870
    },
    {
      "epoch": 1.1612264581958476,
      "grad_norm": 0.832575798034668,
      "learning_rate": 1.225849027869435e-05,
      "loss": 0.0637,
      "step": 22875
    },
    {
      "epoch": 1.16148027818671,
      "grad_norm": 0.8184303045272827,
      "learning_rate": 1.2256798145421936e-05,
      "loss": 0.0623,
      "step": 22880
    },
    {
      "epoch": 1.1617340981775726,
      "grad_norm": 0.45047980546951294,
      "learning_rate": 1.2255106012149517e-05,
      "loss": 0.0566,
      "step": 22885
    },
    {
      "epoch": 1.161987918168435,
      "grad_norm": 0.42531150579452515,
      "learning_rate": 1.2253413878877101e-05,
      "loss": 0.0478,
      "step": 22890
    },
    {
      "epoch": 1.1622417381592975,
      "grad_norm": 0.8212893605232239,
      "learning_rate": 1.2251721745604686e-05,
      "loss": 0.0766,
      "step": 22895
    },
    {
      "epoch": 1.16249555815016,
      "grad_norm": 0.7722528576850891,
      "learning_rate": 1.2250029612332268e-05,
      "loss": 0.0638,
      "step": 22900
    },
    {
      "epoch": 1.1627493781410223,
      "grad_norm": 0.7015485167503357,
      "learning_rate": 1.2248337479059852e-05,
      "loss": 0.0524,
      "step": 22905
    },
    {
      "epoch": 1.163003198131885,
      "grad_norm": 0.6041594743728638,
      "learning_rate": 1.2246645345787435e-05,
      "loss": 0.0542,
      "step": 22910
    },
    {
      "epoch": 1.1632570181227473,
      "grad_norm": 0.5705603957176208,
      "learning_rate": 1.2244953212515019e-05,
      "loss": 0.051,
      "step": 22915
    },
    {
      "epoch": 1.16351083811361,
      "grad_norm": 0.3686521351337433,
      "learning_rate": 1.2243261079242604e-05,
      "loss": 0.0517,
      "step": 22920
    },
    {
      "epoch": 1.1637646581044723,
      "grad_norm": 0.9701533317565918,
      "learning_rate": 1.2241568945970186e-05,
      "loss": 0.0582,
      "step": 22925
    },
    {
      "epoch": 1.1640184780953349,
      "grad_norm": 0.47844773530960083,
      "learning_rate": 1.223987681269777e-05,
      "loss": 0.0503,
      "step": 22930
    },
    {
      "epoch": 1.1642722980861973,
      "grad_norm": 0.5356249809265137,
      "learning_rate": 1.2238184679425353e-05,
      "loss": 0.0538,
      "step": 22935
    },
    {
      "epoch": 1.1645261180770596,
      "grad_norm": 0.4381619393825531,
      "learning_rate": 1.2236492546152936e-05,
      "loss": 0.0558,
      "step": 22940
    },
    {
      "epoch": 1.1647799380679222,
      "grad_norm": 0.4224344789981842,
      "learning_rate": 1.2234800412880518e-05,
      "loss": 0.0653,
      "step": 22945
    },
    {
      "epoch": 1.1650337580587846,
      "grad_norm": 0.48513519763946533,
      "learning_rate": 1.2233108279608103e-05,
      "loss": 0.0508,
      "step": 22950
    },
    {
      "epoch": 1.1652875780496472,
      "grad_norm": 0.31747058033943176,
      "learning_rate": 1.2231416146335687e-05,
      "loss": 0.0609,
      "step": 22955
    },
    {
      "epoch": 1.1655413980405096,
      "grad_norm": 0.9931601881980896,
      "learning_rate": 1.2229724013063269e-05,
      "loss": 0.0659,
      "step": 22960
    },
    {
      "epoch": 1.1657952180313722,
      "grad_norm": 0.30699819326400757,
      "learning_rate": 1.2228031879790854e-05,
      "loss": 0.0567,
      "step": 22965
    },
    {
      "epoch": 1.1660490380222346,
      "grad_norm": 0.5038007497787476,
      "learning_rate": 1.2226339746518436e-05,
      "loss": 0.0563,
      "step": 22970
    },
    {
      "epoch": 1.1663028580130972,
      "grad_norm": 0.8224351406097412,
      "learning_rate": 1.2224647613246021e-05,
      "loss": 0.0502,
      "step": 22975
    },
    {
      "epoch": 1.1665566780039596,
      "grad_norm": 0.3340928852558136,
      "learning_rate": 1.2222955479973603e-05,
      "loss": 0.0534,
      "step": 22980
    },
    {
      "epoch": 1.1668104979948222,
      "grad_norm": 0.2981889843940735,
      "learning_rate": 1.2221263346701187e-05,
      "loss": 0.0506,
      "step": 22985
    },
    {
      "epoch": 1.1670643179856846,
      "grad_norm": 0.46697402000427246,
      "learning_rate": 1.2219571213428772e-05,
      "loss": 0.055,
      "step": 22990
    },
    {
      "epoch": 1.167318137976547,
      "grad_norm": 0.3900391161441803,
      "learning_rate": 1.2217879080156354e-05,
      "loss": 0.0646,
      "step": 22995
    },
    {
      "epoch": 1.1675719579674095,
      "grad_norm": 0.25162240862846375,
      "learning_rate": 1.2216186946883937e-05,
      "loss": 0.0522,
      "step": 23000
    },
    {
      "epoch": 1.167825777958272,
      "grad_norm": 0.349566251039505,
      "learning_rate": 1.221449481361152e-05,
      "loss": 0.0526,
      "step": 23005
    },
    {
      "epoch": 1.1680795979491345,
      "grad_norm": 0.5660791993141174,
      "learning_rate": 1.2212802680339104e-05,
      "loss": 0.0635,
      "step": 23010
    },
    {
      "epoch": 1.168333417939997,
      "grad_norm": 0.42194125056266785,
      "learning_rate": 1.221111054706669e-05,
      "loss": 0.0556,
      "step": 23015
    },
    {
      "epoch": 1.1685872379308595,
      "grad_norm": 0.38500216603279114,
      "learning_rate": 1.2209418413794271e-05,
      "loss": 0.0589,
      "step": 23020
    },
    {
      "epoch": 1.1688410579217219,
      "grad_norm": 0.6102268695831299,
      "learning_rate": 1.2207726280521855e-05,
      "loss": 0.0497,
      "step": 23025
    },
    {
      "epoch": 1.1690948779125845,
      "grad_norm": 0.30943453311920166,
      "learning_rate": 1.2206034147249438e-05,
      "loss": 0.0577,
      "step": 23030
    },
    {
      "epoch": 1.1693486979034469,
      "grad_norm": 0.413257896900177,
      "learning_rate": 1.2204342013977022e-05,
      "loss": 0.0482,
      "step": 23035
    },
    {
      "epoch": 1.1696025178943095,
      "grad_norm": 0.5840480923652649,
      "learning_rate": 1.2202649880704604e-05,
      "loss": 0.0581,
      "step": 23040
    },
    {
      "epoch": 1.1698563378851718,
      "grad_norm": 0.3983316719532013,
      "learning_rate": 1.2200957747432189e-05,
      "loss": 0.0566,
      "step": 23045
    },
    {
      "epoch": 1.1701101578760342,
      "grad_norm": 0.6954087615013123,
      "learning_rate": 1.2199265614159773e-05,
      "loss": 0.0557,
      "step": 23050
    },
    {
      "epoch": 1.1703639778668968,
      "grad_norm": 0.36523228883743286,
      "learning_rate": 1.2197573480887355e-05,
      "loss": 0.0611,
      "step": 23055
    },
    {
      "epoch": 1.1706177978577592,
      "grad_norm": 0.4090610146522522,
      "learning_rate": 1.219588134761494e-05,
      "loss": 0.0591,
      "step": 23060
    },
    {
      "epoch": 1.1708716178486218,
      "grad_norm": 0.3347264528274536,
      "learning_rate": 1.2194189214342522e-05,
      "loss": 0.0537,
      "step": 23065
    },
    {
      "epoch": 1.1711254378394842,
      "grad_norm": 0.8110659718513489,
      "learning_rate": 1.2192497081070107e-05,
      "loss": 0.0716,
      "step": 23070
    },
    {
      "epoch": 1.1713792578303468,
      "grad_norm": 0.35596945881843567,
      "learning_rate": 1.219080494779769e-05,
      "loss": 0.0577,
      "step": 23075
    },
    {
      "epoch": 1.1716330778212092,
      "grad_norm": 0.39061930775642395,
      "learning_rate": 1.2189112814525272e-05,
      "loss": 0.0533,
      "step": 23080
    },
    {
      "epoch": 1.1718868978120716,
      "grad_norm": 0.4372428357601166,
      "learning_rate": 1.2187420681252857e-05,
      "loss": 0.0631,
      "step": 23085
    },
    {
      "epoch": 1.1721407178029342,
      "grad_norm": 0.35665372014045715,
      "learning_rate": 1.218572854798044e-05,
      "loss": 0.0549,
      "step": 23090
    },
    {
      "epoch": 1.1723945377937965,
      "grad_norm": 0.3845728933811188,
      "learning_rate": 1.2184036414708023e-05,
      "loss": 0.0632,
      "step": 23095
    },
    {
      "epoch": 1.1726483577846591,
      "grad_norm": 0.3194248080253601,
      "learning_rate": 1.2182344281435608e-05,
      "loss": 0.0512,
      "step": 23100
    },
    {
      "epoch": 1.1729021777755215,
      "grad_norm": 0.3907899260520935,
      "learning_rate": 1.218065214816319e-05,
      "loss": 0.0601,
      "step": 23105
    },
    {
      "epoch": 1.1731559977663841,
      "grad_norm": 0.37811875343322754,
      "learning_rate": 1.2178960014890774e-05,
      "loss": 0.0513,
      "step": 23110
    },
    {
      "epoch": 1.1734098177572465,
      "grad_norm": 0.2940334677696228,
      "learning_rate": 1.2177267881618357e-05,
      "loss": 0.0485,
      "step": 23115
    },
    {
      "epoch": 1.1736636377481091,
      "grad_norm": 0.3130854666233063,
      "learning_rate": 1.217557574834594e-05,
      "loss": 0.055,
      "step": 23120
    },
    {
      "epoch": 1.1739174577389715,
      "grad_norm": 0.5939212441444397,
      "learning_rate": 1.2173883615073526e-05,
      "loss": 0.0619,
      "step": 23125
    },
    {
      "epoch": 1.174171277729834,
      "grad_norm": 0.3176313042640686,
      "learning_rate": 1.2172191481801108e-05,
      "loss": 0.0541,
      "step": 23130
    },
    {
      "epoch": 1.1744250977206965,
      "grad_norm": 0.2850649654865265,
      "learning_rate": 1.2170499348528691e-05,
      "loss": 0.0575,
      "step": 23135
    },
    {
      "epoch": 1.1746789177115589,
      "grad_norm": 0.4402577877044678,
      "learning_rate": 1.2168807215256275e-05,
      "loss": 0.0668,
      "step": 23140
    },
    {
      "epoch": 1.1749327377024215,
      "grad_norm": 0.3713493347167969,
      "learning_rate": 1.2167115081983858e-05,
      "loss": 0.0547,
      "step": 23145
    },
    {
      "epoch": 1.1751865576932838,
      "grad_norm": 0.38666704297065735,
      "learning_rate": 1.216542294871144e-05,
      "loss": 0.0642,
      "step": 23150
    },
    {
      "epoch": 1.1754403776841464,
      "grad_norm": 0.4440547823905945,
      "learning_rate": 1.2163730815439025e-05,
      "loss": 0.0482,
      "step": 23155
    },
    {
      "epoch": 1.1756941976750088,
      "grad_norm": 0.22269631922245026,
      "learning_rate": 1.2162038682166607e-05,
      "loss": 0.048,
      "step": 23160
    },
    {
      "epoch": 1.1759480176658714,
      "grad_norm": 0.4494595527648926,
      "learning_rate": 1.2160346548894193e-05,
      "loss": 0.0526,
      "step": 23165
    },
    {
      "epoch": 1.1762018376567338,
      "grad_norm": 0.36091774702072144,
      "learning_rate": 1.2158654415621776e-05,
      "loss": 0.0502,
      "step": 23170
    },
    {
      "epoch": 1.1764556576475964,
      "grad_norm": 0.3348097503185272,
      "learning_rate": 1.2156962282349358e-05,
      "loss": 0.0491,
      "step": 23175
    },
    {
      "epoch": 1.1767094776384588,
      "grad_norm": 0.345774382352829,
      "learning_rate": 1.2155270149076943e-05,
      "loss": 0.0471,
      "step": 23180
    },
    {
      "epoch": 1.1769632976293214,
      "grad_norm": 0.3489903211593628,
      "learning_rate": 1.2153578015804525e-05,
      "loss": 0.0516,
      "step": 23185
    },
    {
      "epoch": 1.1772171176201838,
      "grad_norm": 0.2823465168476105,
      "learning_rate": 1.2151885882532109e-05,
      "loss": 0.0553,
      "step": 23190
    },
    {
      "epoch": 1.1774709376110462,
      "grad_norm": 0.553920328617096,
      "learning_rate": 1.2150193749259694e-05,
      "loss": 0.0514,
      "step": 23195
    },
    {
      "epoch": 1.1777247576019088,
      "grad_norm": 0.3592303991317749,
      "learning_rate": 1.2148501615987276e-05,
      "loss": 0.0626,
      "step": 23200
    },
    {
      "epoch": 1.1779785775927711,
      "grad_norm": 0.38532131910324097,
      "learning_rate": 1.214680948271486e-05,
      "loss": 0.0564,
      "step": 23205
    },
    {
      "epoch": 1.1782323975836337,
      "grad_norm": 0.27536770701408386,
      "learning_rate": 1.2145117349442443e-05,
      "loss": 0.0545,
      "step": 23210
    },
    {
      "epoch": 1.1784862175744961,
      "grad_norm": 0.44083353877067566,
      "learning_rate": 1.2143425216170026e-05,
      "loss": 0.058,
      "step": 23215
    },
    {
      "epoch": 1.1787400375653587,
      "grad_norm": 0.30215704441070557,
      "learning_rate": 1.2141733082897611e-05,
      "loss": 0.0515,
      "step": 23220
    },
    {
      "epoch": 1.178993857556221,
      "grad_norm": 0.2528264820575714,
      "learning_rate": 1.2140040949625193e-05,
      "loss": 0.0469,
      "step": 23225
    },
    {
      "epoch": 1.1792476775470835,
      "grad_norm": 0.3954659700393677,
      "learning_rate": 1.2138348816352777e-05,
      "loss": 0.0583,
      "step": 23230
    },
    {
      "epoch": 1.179501497537946,
      "grad_norm": 0.30110490322113037,
      "learning_rate": 1.213665668308036e-05,
      "loss": 0.0541,
      "step": 23235
    },
    {
      "epoch": 1.1797553175288087,
      "grad_norm": 0.47874072194099426,
      "learning_rate": 1.2134964549807944e-05,
      "loss": 0.0599,
      "step": 23240
    },
    {
      "epoch": 1.180009137519671,
      "grad_norm": 0.34137025475502014,
      "learning_rate": 1.2133272416535526e-05,
      "loss": 0.0576,
      "step": 23245
    },
    {
      "epoch": 1.1802629575105335,
      "grad_norm": 0.43007606267929077,
      "learning_rate": 1.2131580283263111e-05,
      "loss": 0.0609,
      "step": 23250
    },
    {
      "epoch": 1.180516777501396,
      "grad_norm": 0.35884323716163635,
      "learning_rate": 1.2129888149990695e-05,
      "loss": 0.0535,
      "step": 23255
    },
    {
      "epoch": 1.1807705974922584,
      "grad_norm": 0.3358921408653259,
      "learning_rate": 1.2128196016718276e-05,
      "loss": 0.0587,
      "step": 23260
    },
    {
      "epoch": 1.181024417483121,
      "grad_norm": 0.3821486830711365,
      "learning_rate": 1.2126503883445862e-05,
      "loss": 0.0604,
      "step": 23265
    },
    {
      "epoch": 1.1812782374739834,
      "grad_norm": 0.5798441767692566,
      "learning_rate": 1.2124811750173444e-05,
      "loss": 0.0598,
      "step": 23270
    },
    {
      "epoch": 1.181532057464846,
      "grad_norm": 0.344942569732666,
      "learning_rate": 1.2123119616901029e-05,
      "loss": 0.0614,
      "step": 23275
    },
    {
      "epoch": 1.1817858774557084,
      "grad_norm": 0.5960049629211426,
      "learning_rate": 1.2121427483628612e-05,
      "loss": 0.0595,
      "step": 23280
    },
    {
      "epoch": 1.1820396974465708,
      "grad_norm": 0.37386181950569153,
      "learning_rate": 1.2119735350356194e-05,
      "loss": 0.0579,
      "step": 23285
    },
    {
      "epoch": 1.1822935174374334,
      "grad_norm": 0.36428967118263245,
      "learning_rate": 1.211804321708378e-05,
      "loss": 0.0623,
      "step": 23290
    },
    {
      "epoch": 1.1825473374282958,
      "grad_norm": 0.39738011360168457,
      "learning_rate": 1.2116351083811361e-05,
      "loss": 0.0622,
      "step": 23295
    },
    {
      "epoch": 1.1828011574191584,
      "grad_norm": 0.5076942443847656,
      "learning_rate": 1.2114658950538945e-05,
      "loss": 0.0531,
      "step": 23300
    },
    {
      "epoch": 1.1830549774100207,
      "grad_norm": 0.3927176296710968,
      "learning_rate": 1.211296681726653e-05,
      "loss": 0.0598,
      "step": 23305
    },
    {
      "epoch": 1.1833087974008833,
      "grad_norm": 0.5115927457809448,
      "learning_rate": 1.2111274683994112e-05,
      "loss": 0.0533,
      "step": 23310
    },
    {
      "epoch": 1.1835626173917457,
      "grad_norm": 0.35300958156585693,
      "learning_rate": 1.2109582550721697e-05,
      "loss": 0.0498,
      "step": 23315
    },
    {
      "epoch": 1.1838164373826083,
      "grad_norm": 0.29630640149116516,
      "learning_rate": 1.2107890417449279e-05,
      "loss": 0.0604,
      "step": 23320
    },
    {
      "epoch": 1.1840702573734707,
      "grad_norm": 0.3651716411113739,
      "learning_rate": 1.2106198284176863e-05,
      "loss": 0.048,
      "step": 23325
    },
    {
      "epoch": 1.1843240773643333,
      "grad_norm": 0.2765333652496338,
      "learning_rate": 1.2104506150904448e-05,
      "loss": 0.0582,
      "step": 23330
    },
    {
      "epoch": 1.1845778973551957,
      "grad_norm": 0.33312156796455383,
      "learning_rate": 1.210281401763203e-05,
      "loss": 0.0541,
      "step": 23335
    },
    {
      "epoch": 1.184831717346058,
      "grad_norm": 0.42472630739212036,
      "learning_rate": 1.2101121884359613e-05,
      "loss": 0.06,
      "step": 23340
    },
    {
      "epoch": 1.1850855373369207,
      "grad_norm": 0.39112213253974915,
      "learning_rate": 1.2099429751087197e-05,
      "loss": 0.0551,
      "step": 23345
    },
    {
      "epoch": 1.185339357327783,
      "grad_norm": 0.4763796627521515,
      "learning_rate": 1.209773761781478e-05,
      "loss": 0.0501,
      "step": 23350
    },
    {
      "epoch": 1.1855931773186457,
      "grad_norm": 0.3270481824874878,
      "learning_rate": 1.2096045484542362e-05,
      "loss": 0.0492,
      "step": 23355
    },
    {
      "epoch": 1.185846997309508,
      "grad_norm": 0.32376089692115784,
      "learning_rate": 1.2094353351269947e-05,
      "loss": 0.0592,
      "step": 23360
    },
    {
      "epoch": 1.1861008173003706,
      "grad_norm": 0.5592198371887207,
      "learning_rate": 1.209266121799753e-05,
      "loss": 0.0584,
      "step": 23365
    },
    {
      "epoch": 1.186354637291233,
      "grad_norm": 0.34053146839141846,
      "learning_rate": 1.2090969084725114e-05,
      "loss": 0.0616,
      "step": 23370
    },
    {
      "epoch": 1.1866084572820956,
      "grad_norm": 0.4493182301521301,
      "learning_rate": 1.2089276951452698e-05,
      "loss": 0.0596,
      "step": 23375
    },
    {
      "epoch": 1.186862277272958,
      "grad_norm": 0.4678153693675995,
      "learning_rate": 1.208758481818028e-05,
      "loss": 0.0567,
      "step": 23380
    },
    {
      "epoch": 1.1871160972638206,
      "grad_norm": 0.4072043001651764,
      "learning_rate": 1.2085892684907865e-05,
      "loss": 0.0569,
      "step": 23385
    },
    {
      "epoch": 1.187369917254683,
      "grad_norm": 0.6094187498092651,
      "learning_rate": 1.2084200551635447e-05,
      "loss": 0.0656,
      "step": 23390
    },
    {
      "epoch": 1.1876237372455454,
      "grad_norm": 0.420998752117157,
      "learning_rate": 1.208250841836303e-05,
      "loss": 0.0612,
      "step": 23395
    },
    {
      "epoch": 1.187877557236408,
      "grad_norm": 0.5017419457435608,
      "learning_rate": 1.2080816285090616e-05,
      "loss": 0.0602,
      "step": 23400
    },
    {
      "epoch": 1.1881313772272704,
      "grad_norm": 0.2493000626564026,
      "learning_rate": 1.2079124151818198e-05,
      "loss": 0.0518,
      "step": 23405
    },
    {
      "epoch": 1.188385197218133,
      "grad_norm": 0.5939289331436157,
      "learning_rate": 1.2077432018545783e-05,
      "loss": 0.057,
      "step": 23410
    },
    {
      "epoch": 1.1886390172089953,
      "grad_norm": 0.34695038199424744,
      "learning_rate": 1.2075739885273365e-05,
      "loss": 0.0555,
      "step": 23415
    },
    {
      "epoch": 1.188892837199858,
      "grad_norm": 0.6310956478118896,
      "learning_rate": 1.2074047752000948e-05,
      "loss": 0.0516,
      "step": 23420
    },
    {
      "epoch": 1.1891466571907203,
      "grad_norm": 0.5171706080436707,
      "learning_rate": 1.2072355618728533e-05,
      "loss": 0.0543,
      "step": 23425
    },
    {
      "epoch": 1.1894004771815827,
      "grad_norm": 0.3515871465206146,
      "learning_rate": 1.2070663485456115e-05,
      "loss": 0.0583,
      "step": 23430
    },
    {
      "epoch": 1.1896542971724453,
      "grad_norm": 0.2978825271129608,
      "learning_rate": 1.2068971352183699e-05,
      "loss": 0.0456,
      "step": 23435
    },
    {
      "epoch": 1.1899081171633077,
      "grad_norm": 0.41832324862480164,
      "learning_rate": 1.2067279218911282e-05,
      "loss": 0.0531,
      "step": 23440
    },
    {
      "epoch": 1.1901619371541703,
      "grad_norm": 0.4018298387527466,
      "learning_rate": 1.2065587085638866e-05,
      "loss": 0.0523,
      "step": 23445
    },
    {
      "epoch": 1.1904157571450327,
      "grad_norm": 0.24528133869171143,
      "learning_rate": 1.2063894952366448e-05,
      "loss": 0.0498,
      "step": 23450
    },
    {
      "epoch": 1.1906695771358953,
      "grad_norm": 0.36855852603912354,
      "learning_rate": 1.2062202819094033e-05,
      "loss": 0.0533,
      "step": 23455
    },
    {
      "epoch": 1.1909233971267577,
      "grad_norm": 0.4531480073928833,
      "learning_rate": 1.2060510685821617e-05,
      "loss": 0.0552,
      "step": 23460
    },
    {
      "epoch": 1.1911772171176203,
      "grad_norm": 0.5693694949150085,
      "learning_rate": 1.20588185525492e-05,
      "loss": 0.0642,
      "step": 23465
    },
    {
      "epoch": 1.1914310371084826,
      "grad_norm": 0.3737713694572449,
      "learning_rate": 1.2057126419276784e-05,
      "loss": 0.0535,
      "step": 23470
    },
    {
      "epoch": 1.1916848570993452,
      "grad_norm": 0.40837082266807556,
      "learning_rate": 1.2055434286004365e-05,
      "loss": 0.0554,
      "step": 23475
    },
    {
      "epoch": 1.1919386770902076,
      "grad_norm": 0.2771040201187134,
      "learning_rate": 1.205374215273195e-05,
      "loss": 0.0503,
      "step": 23480
    },
    {
      "epoch": 1.19219249708107,
      "grad_norm": 0.2847016453742981,
      "learning_rate": 1.2052050019459534e-05,
      "loss": 0.0525,
      "step": 23485
    },
    {
      "epoch": 1.1924463170719326,
      "grad_norm": 0.2703809440135956,
      "learning_rate": 1.2050357886187116e-05,
      "loss": 0.0447,
      "step": 23490
    },
    {
      "epoch": 1.192700137062795,
      "grad_norm": 0.6293087005615234,
      "learning_rate": 1.2048665752914701e-05,
      "loss": 0.0614,
      "step": 23495
    },
    {
      "epoch": 1.1929539570536576,
      "grad_norm": 0.27921855449676514,
      "learning_rate": 1.2046973619642283e-05,
      "loss": 0.0458,
      "step": 23500
    },
    {
      "epoch": 1.19320777704452,
      "grad_norm": 0.45750892162323,
      "learning_rate": 1.2045281486369867e-05,
      "loss": 0.0526,
      "step": 23505
    },
    {
      "epoch": 1.1934615970353826,
      "grad_norm": 0.3219360113143921,
      "learning_rate": 1.2043589353097452e-05,
      "loss": 0.0477,
      "step": 23510
    },
    {
      "epoch": 1.193715417026245,
      "grad_norm": 0.5314846634864807,
      "learning_rate": 1.2041897219825034e-05,
      "loss": 0.0651,
      "step": 23515
    },
    {
      "epoch": 1.1939692370171076,
      "grad_norm": 0.410866916179657,
      "learning_rate": 1.2040205086552619e-05,
      "loss": 0.0584,
      "step": 23520
    },
    {
      "epoch": 1.19422305700797,
      "grad_norm": 0.2978534996509552,
      "learning_rate": 1.2038512953280201e-05,
      "loss": 0.0476,
      "step": 23525
    },
    {
      "epoch": 1.1944768769988325,
      "grad_norm": 0.2534315586090088,
      "learning_rate": 1.2036820820007784e-05,
      "loss": 0.0458,
      "step": 23530
    },
    {
      "epoch": 1.194730696989695,
      "grad_norm": 0.40118440985679626,
      "learning_rate": 1.203512868673537e-05,
      "loss": 0.0551,
      "step": 23535
    },
    {
      "epoch": 1.1949845169805573,
      "grad_norm": 0.44053852558135986,
      "learning_rate": 1.2033436553462952e-05,
      "loss": 0.0588,
      "step": 23540
    },
    {
      "epoch": 1.19523833697142,
      "grad_norm": 0.41670575737953186,
      "learning_rate": 1.2031744420190533e-05,
      "loss": 0.0581,
      "step": 23545
    },
    {
      "epoch": 1.1954921569622823,
      "grad_norm": 0.30073875188827515,
      "learning_rate": 1.2030052286918119e-05,
      "loss": 0.0507,
      "step": 23550
    },
    {
      "epoch": 1.1957459769531449,
      "grad_norm": 0.43326103687286377,
      "learning_rate": 1.2028360153645702e-05,
      "loss": 0.0632,
      "step": 23555
    },
    {
      "epoch": 1.1959997969440073,
      "grad_norm": 0.30326777696609497,
      "learning_rate": 1.2026668020373287e-05,
      "loss": 0.0525,
      "step": 23560
    },
    {
      "epoch": 1.1962536169348699,
      "grad_norm": 0.36257094144821167,
      "learning_rate": 1.202497588710087e-05,
      "loss": 0.0551,
      "step": 23565
    },
    {
      "epoch": 1.1965074369257322,
      "grad_norm": 0.32513904571533203,
      "learning_rate": 1.2023283753828451e-05,
      "loss": 0.0573,
      "step": 23570
    },
    {
      "epoch": 1.1967612569165946,
      "grad_norm": 0.3686450719833374,
      "learning_rate": 1.2021591620556036e-05,
      "loss": 0.0547,
      "step": 23575
    },
    {
      "epoch": 1.1970150769074572,
      "grad_norm": 0.3500473201274872,
      "learning_rate": 1.201989948728362e-05,
      "loss": 0.0478,
      "step": 23580
    },
    {
      "epoch": 1.1972688968983198,
      "grad_norm": 0.43649592995643616,
      "learning_rate": 1.2018207354011202e-05,
      "loss": 0.0501,
      "step": 23585
    },
    {
      "epoch": 1.1975227168891822,
      "grad_norm": 0.22626397013664246,
      "learning_rate": 1.2016515220738787e-05,
      "loss": 0.0556,
      "step": 23590
    },
    {
      "epoch": 1.1977765368800446,
      "grad_norm": 0.3905506730079651,
      "learning_rate": 1.2014823087466369e-05,
      "loss": 0.0703,
      "step": 23595
    },
    {
      "epoch": 1.1980303568709072,
      "grad_norm": 0.34414008259773254,
      "learning_rate": 1.2013130954193952e-05,
      "loss": 0.0523,
      "step": 23600
    },
    {
      "epoch": 1.1982841768617696,
      "grad_norm": 0.32847830653190613,
      "learning_rate": 1.2011438820921538e-05,
      "loss": 0.0649,
      "step": 23605
    },
    {
      "epoch": 1.1985379968526322,
      "grad_norm": 0.35023289918899536,
      "learning_rate": 1.200974668764912e-05,
      "loss": 0.0598,
      "step": 23610
    },
    {
      "epoch": 1.1987918168434946,
      "grad_norm": 0.29963234066963196,
      "learning_rate": 1.2008054554376705e-05,
      "loss": 0.0586,
      "step": 23615
    },
    {
      "epoch": 1.1990456368343572,
      "grad_norm": 0.3658560812473297,
      "learning_rate": 1.2006362421104287e-05,
      "loss": 0.0548,
      "step": 23620
    },
    {
      "epoch": 1.1992994568252195,
      "grad_norm": 0.6136611700057983,
      "learning_rate": 1.200467028783187e-05,
      "loss": 0.0572,
      "step": 23625
    },
    {
      "epoch": 1.199553276816082,
      "grad_norm": 0.39471113681793213,
      "learning_rate": 1.2002978154559455e-05,
      "loss": 0.0498,
      "step": 23630
    },
    {
      "epoch": 1.1998070968069445,
      "grad_norm": 0.592869758605957,
      "learning_rate": 1.2001286021287037e-05,
      "loss": 0.0608,
      "step": 23635
    },
    {
      "epoch": 1.200060916797807,
      "grad_norm": 0.41724860668182373,
      "learning_rate": 1.199959388801462e-05,
      "loss": 0.0513,
      "step": 23640
    },
    {
      "epoch": 1.2003147367886695,
      "grad_norm": 0.31232571601867676,
      "learning_rate": 1.1997901754742204e-05,
      "loss": 0.0612,
      "step": 23645
    },
    {
      "epoch": 1.200568556779532,
      "grad_norm": 0.33139002323150635,
      "learning_rate": 1.1996209621469788e-05,
      "loss": 0.052,
      "step": 23650
    },
    {
      "epoch": 1.2008223767703945,
      "grad_norm": 0.3042997717857361,
      "learning_rate": 1.1994517488197373e-05,
      "loss": 0.0504,
      "step": 23655
    },
    {
      "epoch": 1.2010761967612569,
      "grad_norm": 0.40804755687713623,
      "learning_rate": 1.1992825354924955e-05,
      "loss": 0.0535,
      "step": 23660
    },
    {
      "epoch": 1.2013300167521195,
      "grad_norm": 0.41323235630989075,
      "learning_rate": 1.1991133221652538e-05,
      "loss": 0.0618,
      "step": 23665
    },
    {
      "epoch": 1.2015838367429819,
      "grad_norm": 0.375251442193985,
      "learning_rate": 1.1989441088380122e-05,
      "loss": 0.0585,
      "step": 23670
    },
    {
      "epoch": 1.2018376567338445,
      "grad_norm": 0.4380832016468048,
      "learning_rate": 1.1987748955107706e-05,
      "loss": 0.0574,
      "step": 23675
    },
    {
      "epoch": 1.2020914767247068,
      "grad_norm": 0.5451785326004028,
      "learning_rate": 1.1986056821835287e-05,
      "loss": 0.0532,
      "step": 23680
    },
    {
      "epoch": 1.2023452967155692,
      "grad_norm": 0.3498691916465759,
      "learning_rate": 1.1984364688562873e-05,
      "loss": 0.0555,
      "step": 23685
    },
    {
      "epoch": 1.2025991167064318,
      "grad_norm": 0.31484562158584595,
      "learning_rate": 1.1982672555290456e-05,
      "loss": 0.054,
      "step": 23690
    },
    {
      "epoch": 1.2028529366972942,
      "grad_norm": 0.3103238642215729,
      "learning_rate": 1.1980980422018038e-05,
      "loss": 0.0557,
      "step": 23695
    },
    {
      "epoch": 1.2031067566881568,
      "grad_norm": 0.3442491590976715,
      "learning_rate": 1.1979288288745623e-05,
      "loss": 0.0488,
      "step": 23700
    },
    {
      "epoch": 1.2033605766790192,
      "grad_norm": 0.5479670166969299,
      "learning_rate": 1.1977596155473205e-05,
      "loss": 0.0545,
      "step": 23705
    },
    {
      "epoch": 1.2036143966698818,
      "grad_norm": 0.43110498785972595,
      "learning_rate": 1.197590402220079e-05,
      "loss": 0.0564,
      "step": 23710
    },
    {
      "epoch": 1.2038682166607442,
      "grad_norm": 0.3031986951828003,
      "learning_rate": 1.1974211888928374e-05,
      "loss": 0.0654,
      "step": 23715
    },
    {
      "epoch": 1.2041220366516068,
      "grad_norm": 0.4608815312385559,
      "learning_rate": 1.1972519755655956e-05,
      "loss": 0.0486,
      "step": 23720
    },
    {
      "epoch": 1.2043758566424692,
      "grad_norm": 0.31896111369132996,
      "learning_rate": 1.1970827622383541e-05,
      "loss": 0.0569,
      "step": 23725
    },
    {
      "epoch": 1.2046296766333318,
      "grad_norm": 0.3649381697177887,
      "learning_rate": 1.1969135489111123e-05,
      "loss": 0.0501,
      "step": 23730
    },
    {
      "epoch": 1.2048834966241941,
      "grad_norm": 0.45164427161216736,
      "learning_rate": 1.1967443355838706e-05,
      "loss": 0.0505,
      "step": 23735
    },
    {
      "epoch": 1.2051373166150565,
      "grad_norm": 0.3351602554321289,
      "learning_rate": 1.1965751222566292e-05,
      "loss": 0.0471,
      "step": 23740
    },
    {
      "epoch": 1.2053911366059191,
      "grad_norm": 0.4169917702674866,
      "learning_rate": 1.1964059089293874e-05,
      "loss": 0.0526,
      "step": 23745
    },
    {
      "epoch": 1.2056449565967815,
      "grad_norm": 0.5554206967353821,
      "learning_rate": 1.1962366956021455e-05,
      "loss": 0.0529,
      "step": 23750
    },
    {
      "epoch": 1.205898776587644,
      "grad_norm": 0.4624297320842743,
      "learning_rate": 1.196067482274904e-05,
      "loss": 0.0504,
      "step": 23755
    },
    {
      "epoch": 1.2061525965785065,
      "grad_norm": 0.2930656671524048,
      "learning_rate": 1.1958982689476624e-05,
      "loss": 0.0548,
      "step": 23760
    },
    {
      "epoch": 1.206406416569369,
      "grad_norm": 0.2735726833343506,
      "learning_rate": 1.195729055620421e-05,
      "loss": 0.0532,
      "step": 23765
    },
    {
      "epoch": 1.2066602365602315,
      "grad_norm": 0.3359985053539276,
      "learning_rate": 1.1955598422931791e-05,
      "loss": 0.0601,
      "step": 23770
    },
    {
      "epoch": 1.2069140565510939,
      "grad_norm": 0.31836122274398804,
      "learning_rate": 1.1953906289659373e-05,
      "loss": 0.0541,
      "step": 23775
    },
    {
      "epoch": 1.2071678765419565,
      "grad_norm": 0.5076533555984497,
      "learning_rate": 1.1952214156386958e-05,
      "loss": 0.0467,
      "step": 23780
    },
    {
      "epoch": 1.2074216965328188,
      "grad_norm": 0.4004075229167938,
      "learning_rate": 1.1950522023114542e-05,
      "loss": 0.06,
      "step": 23785
    },
    {
      "epoch": 1.2076755165236814,
      "grad_norm": 0.5112072229385376,
      "learning_rate": 1.1948829889842124e-05,
      "loss": 0.0576,
      "step": 23790
    },
    {
      "epoch": 1.2079293365145438,
      "grad_norm": 0.3895137310028076,
      "learning_rate": 1.1947137756569709e-05,
      "loss": 0.0458,
      "step": 23795
    },
    {
      "epoch": 1.2081831565054064,
      "grad_norm": 0.7249547839164734,
      "learning_rate": 1.194544562329729e-05,
      "loss": 0.0531,
      "step": 23800
    },
    {
      "epoch": 1.2084369764962688,
      "grad_norm": 0.7560728788375854,
      "learning_rate": 1.1943753490024876e-05,
      "loss": 0.0492,
      "step": 23805
    },
    {
      "epoch": 1.2086907964871314,
      "grad_norm": 0.3840545117855072,
      "learning_rate": 1.194206135675246e-05,
      "loss": 0.0474,
      "step": 23810
    },
    {
      "epoch": 1.2089446164779938,
      "grad_norm": 0.42460840940475464,
      "learning_rate": 1.1940369223480041e-05,
      "loss": 0.0537,
      "step": 23815
    },
    {
      "epoch": 1.2091984364688564,
      "grad_norm": 0.2921667695045471,
      "learning_rate": 1.1938677090207627e-05,
      "loss": 0.0526,
      "step": 23820
    },
    {
      "epoch": 1.2094522564597188,
      "grad_norm": 0.3440396189689636,
      "learning_rate": 1.1936984956935209e-05,
      "loss": 0.0454,
      "step": 23825
    },
    {
      "epoch": 1.2097060764505811,
      "grad_norm": 0.33783891797065735,
      "learning_rate": 1.1935292823662792e-05,
      "loss": 0.0466,
      "step": 23830
    },
    {
      "epoch": 1.2099598964414437,
      "grad_norm": 0.3296278417110443,
      "learning_rate": 1.1933600690390377e-05,
      "loss": 0.0621,
      "step": 23835
    },
    {
      "epoch": 1.2102137164323061,
      "grad_norm": 0.5119358897209167,
      "learning_rate": 1.193190855711796e-05,
      "loss": 0.0585,
      "step": 23840
    },
    {
      "epoch": 1.2104675364231687,
      "grad_norm": 0.738791823387146,
      "learning_rate": 1.1930216423845543e-05,
      "loss": 0.0524,
      "step": 23845
    },
    {
      "epoch": 1.2107213564140311,
      "grad_norm": 0.3069751262664795,
      "learning_rate": 1.1928524290573126e-05,
      "loss": 0.0441,
      "step": 23850
    },
    {
      "epoch": 1.2109751764048937,
      "grad_norm": 0.30451783537864685,
      "learning_rate": 1.192683215730071e-05,
      "loss": 0.0468,
      "step": 23855
    },
    {
      "epoch": 1.211228996395756,
      "grad_norm": 0.32865050435066223,
      "learning_rate": 1.1925140024028295e-05,
      "loss": 0.0534,
      "step": 23860
    },
    {
      "epoch": 1.2114828163866187,
      "grad_norm": 0.41793230175971985,
      "learning_rate": 1.1923447890755877e-05,
      "loss": 0.0504,
      "step": 23865
    },
    {
      "epoch": 1.211736636377481,
      "grad_norm": 0.6536849737167358,
      "learning_rate": 1.192175575748346e-05,
      "loss": 0.0527,
      "step": 23870
    },
    {
      "epoch": 1.2119904563683437,
      "grad_norm": 0.28495994210243225,
      "learning_rate": 1.1920063624211044e-05,
      "loss": 0.0505,
      "step": 23875
    },
    {
      "epoch": 1.212244276359206,
      "grad_norm": 1.1305482387542725,
      "learning_rate": 1.1918371490938628e-05,
      "loss": 0.0657,
      "step": 23880
    },
    {
      "epoch": 1.2124980963500684,
      "grad_norm": 0.4528692662715912,
      "learning_rate": 1.191667935766621e-05,
      "loss": 0.0519,
      "step": 23885
    },
    {
      "epoch": 1.212751916340931,
      "grad_norm": 0.49388232827186584,
      "learning_rate": 1.1914987224393795e-05,
      "loss": 0.0554,
      "step": 23890
    },
    {
      "epoch": 1.2130057363317934,
      "grad_norm": 0.3623510003089905,
      "learning_rate": 1.1913295091121378e-05,
      "loss": 0.0572,
      "step": 23895
    },
    {
      "epoch": 1.213259556322656,
      "grad_norm": 0.3501339852809906,
      "learning_rate": 1.1911602957848962e-05,
      "loss": 0.0584,
      "step": 23900
    },
    {
      "epoch": 1.2135133763135184,
      "grad_norm": 0.588069498538971,
      "learning_rate": 1.1909910824576545e-05,
      "loss": 0.053,
      "step": 23905
    },
    {
      "epoch": 1.213767196304381,
      "grad_norm": 0.5254976153373718,
      "learning_rate": 1.1908218691304127e-05,
      "loss": 0.053,
      "step": 23910
    },
    {
      "epoch": 1.2140210162952434,
      "grad_norm": 0.6018771529197693,
      "learning_rate": 1.1906526558031712e-05,
      "loss": 0.057,
      "step": 23915
    },
    {
      "epoch": 1.2142748362861058,
      "grad_norm": 0.3812793791294098,
      "learning_rate": 1.1904834424759296e-05,
      "loss": 0.0658,
      "step": 23920
    },
    {
      "epoch": 1.2145286562769684,
      "grad_norm": 0.36745819449424744,
      "learning_rate": 1.1903142291486878e-05,
      "loss": 0.0546,
      "step": 23925
    },
    {
      "epoch": 1.2147824762678308,
      "grad_norm": 0.28504976630210876,
      "learning_rate": 1.1901450158214463e-05,
      "loss": 0.0489,
      "step": 23930
    },
    {
      "epoch": 1.2150362962586934,
      "grad_norm": 0.2376314401626587,
      "learning_rate": 1.1899758024942045e-05,
      "loss": 0.0565,
      "step": 23935
    },
    {
      "epoch": 1.2152901162495557,
      "grad_norm": 0.36779358983039856,
      "learning_rate": 1.1898065891669628e-05,
      "loss": 0.0521,
      "step": 23940
    },
    {
      "epoch": 1.2155439362404183,
      "grad_norm": 0.2886993885040283,
      "learning_rate": 1.1896373758397214e-05,
      "loss": 0.0444,
      "step": 23945
    },
    {
      "epoch": 1.2157977562312807,
      "grad_norm": 0.42894473671913147,
      "learning_rate": 1.1894681625124795e-05,
      "loss": 0.0478,
      "step": 23950
    },
    {
      "epoch": 1.2160515762221433,
      "grad_norm": 0.4883272051811218,
      "learning_rate": 1.189298949185238e-05,
      "loss": 0.0555,
      "step": 23955
    },
    {
      "epoch": 1.2163053962130057,
      "grad_norm": 0.4135822057723999,
      "learning_rate": 1.1891297358579963e-05,
      "loss": 0.0607,
      "step": 23960
    },
    {
      "epoch": 1.2165592162038683,
      "grad_norm": 0.8400814533233643,
      "learning_rate": 1.1889605225307546e-05,
      "loss": 0.0594,
      "step": 23965
    },
    {
      "epoch": 1.2168130361947307,
      "grad_norm": 0.2872380316257477,
      "learning_rate": 1.188791309203513e-05,
      "loss": 0.0555,
      "step": 23970
    },
    {
      "epoch": 1.217066856185593,
      "grad_norm": 0.3974021375179291,
      "learning_rate": 1.1886220958762713e-05,
      "loss": 0.0583,
      "step": 23975
    },
    {
      "epoch": 1.2173206761764557,
      "grad_norm": 0.5222055912017822,
      "learning_rate": 1.1884528825490295e-05,
      "loss": 0.0612,
      "step": 23980
    },
    {
      "epoch": 1.217574496167318,
      "grad_norm": 0.37451913952827454,
      "learning_rate": 1.188283669221788e-05,
      "loss": 0.058,
      "step": 23985
    },
    {
      "epoch": 1.2178283161581807,
      "grad_norm": 0.38074713945388794,
      "learning_rate": 1.1881144558945464e-05,
      "loss": 0.0627,
      "step": 23990
    },
    {
      "epoch": 1.218082136149043,
      "grad_norm": 0.47878900170326233,
      "learning_rate": 1.1879452425673046e-05,
      "loss": 0.062,
      "step": 23995
    },
    {
      "epoch": 1.2183359561399056,
      "grad_norm": 0.5156375169754028,
      "learning_rate": 1.1877760292400631e-05,
      "loss": 0.0486,
      "step": 24000
    },
    {
      "epoch": 1.218589776130768,
      "grad_norm": 0.24383500218391418,
      "learning_rate": 1.1876068159128213e-05,
      "loss": 0.0474,
      "step": 24005
    },
    {
      "epoch": 1.2188435961216306,
      "grad_norm": 0.34059467911720276,
      "learning_rate": 1.1874376025855798e-05,
      "loss": 0.0466,
      "step": 24010
    },
    {
      "epoch": 1.219097416112493,
      "grad_norm": 0.31764811277389526,
      "learning_rate": 1.1872683892583382e-05,
      "loss": 0.0513,
      "step": 24015
    },
    {
      "epoch": 1.2193512361033556,
      "grad_norm": 0.43181148171424866,
      "learning_rate": 1.1870991759310963e-05,
      "loss": 0.0458,
      "step": 24020
    },
    {
      "epoch": 1.219605056094218,
      "grad_norm": 0.6314930319786072,
      "learning_rate": 1.1869299626038549e-05,
      "loss": 0.0542,
      "step": 24025
    },
    {
      "epoch": 1.2198588760850804,
      "grad_norm": 0.37427467107772827,
      "learning_rate": 1.186760749276613e-05,
      "loss": 0.0531,
      "step": 24030
    },
    {
      "epoch": 1.220112696075943,
      "grad_norm": 0.2935940623283386,
      "learning_rate": 1.1865915359493714e-05,
      "loss": 0.052,
      "step": 24035
    },
    {
      "epoch": 1.2203665160668054,
      "grad_norm": 0.48649024963378906,
      "learning_rate": 1.18642232262213e-05,
      "loss": 0.0579,
      "step": 24040
    },
    {
      "epoch": 1.220620336057668,
      "grad_norm": 0.5453752279281616,
      "learning_rate": 1.1862531092948881e-05,
      "loss": 0.0585,
      "step": 24045
    },
    {
      "epoch": 1.2208741560485303,
      "grad_norm": 0.24057643115520477,
      "learning_rate": 1.1860838959676466e-05,
      "loss": 0.0546,
      "step": 24050
    },
    {
      "epoch": 1.221127976039393,
      "grad_norm": 0.3876838684082031,
      "learning_rate": 1.1859146826404048e-05,
      "loss": 0.0548,
      "step": 24055
    },
    {
      "epoch": 1.2213817960302553,
      "grad_norm": 0.7951164245605469,
      "learning_rate": 1.1857454693131632e-05,
      "loss": 0.0601,
      "step": 24060
    },
    {
      "epoch": 1.2216356160211177,
      "grad_norm": 0.26889508962631226,
      "learning_rate": 1.1855762559859217e-05,
      "loss": 0.055,
      "step": 24065
    },
    {
      "epoch": 1.2218894360119803,
      "grad_norm": 0.48906293511390686,
      "learning_rate": 1.1854070426586799e-05,
      "loss": 0.0554,
      "step": 24070
    },
    {
      "epoch": 1.222143256002843,
      "grad_norm": 0.3633570373058319,
      "learning_rate": 1.1852378293314382e-05,
      "loss": 0.0529,
      "step": 24075
    },
    {
      "epoch": 1.2223970759937053,
      "grad_norm": 0.273607462644577,
      "learning_rate": 1.1850686160041966e-05,
      "loss": 0.0513,
      "step": 24080
    },
    {
      "epoch": 1.2226508959845677,
      "grad_norm": 0.6867349743843079,
      "learning_rate": 1.184899402676955e-05,
      "loss": 0.0531,
      "step": 24085
    },
    {
      "epoch": 1.2229047159754303,
      "grad_norm": 0.36821645498275757,
      "learning_rate": 1.1847301893497131e-05,
      "loss": 0.0494,
      "step": 24090
    },
    {
      "epoch": 1.2231585359662926,
      "grad_norm": 0.737259030342102,
      "learning_rate": 1.1845609760224717e-05,
      "loss": 0.061,
      "step": 24095
    },
    {
      "epoch": 1.2234123559571553,
      "grad_norm": 0.39399975538253784,
      "learning_rate": 1.18439176269523e-05,
      "loss": 0.0432,
      "step": 24100
    },
    {
      "epoch": 1.2236661759480176,
      "grad_norm": 0.34652629494667053,
      "learning_rate": 1.1842225493679884e-05,
      "loss": 0.06,
      "step": 24105
    },
    {
      "epoch": 1.2239199959388802,
      "grad_norm": 0.4035334289073944,
      "learning_rate": 1.1840533360407467e-05,
      "loss": 0.0552,
      "step": 24110
    },
    {
      "epoch": 1.2241738159297426,
      "grad_norm": 0.45802584290504456,
      "learning_rate": 1.1838841227135049e-05,
      "loss": 0.0525,
      "step": 24115
    },
    {
      "epoch": 1.224427635920605,
      "grad_norm": 0.26952633261680603,
      "learning_rate": 1.1837149093862634e-05,
      "loss": 0.0511,
      "step": 24120
    },
    {
      "epoch": 1.2246814559114676,
      "grad_norm": 0.5091894268989563,
      "learning_rate": 1.1835456960590218e-05,
      "loss": 0.0599,
      "step": 24125
    },
    {
      "epoch": 1.22493527590233,
      "grad_norm": 0.5348808169364929,
      "learning_rate": 1.18337648273178e-05,
      "loss": 0.0505,
      "step": 24130
    },
    {
      "epoch": 1.2251890958931926,
      "grad_norm": 0.41867509484291077,
      "learning_rate": 1.1832072694045385e-05,
      "loss": 0.0572,
      "step": 24135
    },
    {
      "epoch": 1.225442915884055,
      "grad_norm": 0.4618304967880249,
      "learning_rate": 1.1830380560772967e-05,
      "loss": 0.0479,
      "step": 24140
    },
    {
      "epoch": 1.2256967358749176,
      "grad_norm": 0.3024294078350067,
      "learning_rate": 1.1828688427500552e-05,
      "loss": 0.0496,
      "step": 24145
    },
    {
      "epoch": 1.22595055586578,
      "grad_norm": 0.3937217891216278,
      "learning_rate": 1.1826996294228134e-05,
      "loss": 0.0502,
      "step": 24150
    },
    {
      "epoch": 1.2262043758566425,
      "grad_norm": 0.44144943356513977,
      "learning_rate": 1.1825304160955717e-05,
      "loss": 0.0502,
      "step": 24155
    },
    {
      "epoch": 1.226458195847505,
      "grad_norm": 0.4977549910545349,
      "learning_rate": 1.1823612027683303e-05,
      "loss": 0.0539,
      "step": 24160
    },
    {
      "epoch": 1.2267120158383675,
      "grad_norm": 0.37051019072532654,
      "learning_rate": 1.1821919894410884e-05,
      "loss": 0.0527,
      "step": 24165
    },
    {
      "epoch": 1.22696583582923,
      "grad_norm": 0.5204314589500427,
      "learning_rate": 1.1820227761138468e-05,
      "loss": 0.0465,
      "step": 24170
    },
    {
      "epoch": 1.2272196558200923,
      "grad_norm": 1.4250450134277344,
      "learning_rate": 1.1818535627866052e-05,
      "loss": 0.0438,
      "step": 24175
    },
    {
      "epoch": 1.227473475810955,
      "grad_norm": 0.7377504110336304,
      "learning_rate": 1.1816843494593635e-05,
      "loss": 0.0584,
      "step": 24180
    },
    {
      "epoch": 1.2277272958018173,
      "grad_norm": 0.3235386908054352,
      "learning_rate": 1.1815151361321217e-05,
      "loss": 0.0511,
      "step": 24185
    },
    {
      "epoch": 1.2279811157926799,
      "grad_norm": 0.4034362733364105,
      "learning_rate": 1.1813459228048802e-05,
      "loss": 0.0467,
      "step": 24190
    },
    {
      "epoch": 1.2282349357835423,
      "grad_norm": 0.30886006355285645,
      "learning_rate": 1.1811767094776386e-05,
      "loss": 0.043,
      "step": 24195
    },
    {
      "epoch": 1.2284887557744049,
      "grad_norm": 0.38146716356277466,
      "learning_rate": 1.181007496150397e-05,
      "loss": 0.0594,
      "step": 24200
    },
    {
      "epoch": 1.2287425757652672,
      "grad_norm": 0.34178265929222107,
      "learning_rate": 1.1808382828231553e-05,
      "loss": 0.0486,
      "step": 24205
    },
    {
      "epoch": 1.2289963957561298,
      "grad_norm": 0.5366050601005554,
      "learning_rate": 1.1806690694959135e-05,
      "loss": 0.0621,
      "step": 24210
    },
    {
      "epoch": 1.2292502157469922,
      "grad_norm": 0.36923089623451233,
      "learning_rate": 1.180499856168672e-05,
      "loss": 0.0527,
      "step": 24215
    },
    {
      "epoch": 1.2295040357378548,
      "grad_norm": 0.39558228850364685,
      "learning_rate": 1.1803306428414303e-05,
      "loss": 0.0618,
      "step": 24220
    },
    {
      "epoch": 1.2297578557287172,
      "grad_norm": 0.3447304368019104,
      "learning_rate": 1.1801614295141885e-05,
      "loss": 0.0526,
      "step": 24225
    },
    {
      "epoch": 1.2300116757195796,
      "grad_norm": 0.5974411964416504,
      "learning_rate": 1.179992216186947e-05,
      "loss": 0.0522,
      "step": 24230
    },
    {
      "epoch": 1.2302654957104422,
      "grad_norm": 0.31374403834342957,
      "learning_rate": 1.1798230028597052e-05,
      "loss": 0.0497,
      "step": 24235
    },
    {
      "epoch": 1.2305193157013046,
      "grad_norm": 0.3632044792175293,
      "learning_rate": 1.1796537895324636e-05,
      "loss": 0.0487,
      "step": 24240
    },
    {
      "epoch": 1.2307731356921672,
      "grad_norm": 0.36959999799728394,
      "learning_rate": 1.1794845762052221e-05,
      "loss": 0.055,
      "step": 24245
    },
    {
      "epoch": 1.2310269556830296,
      "grad_norm": 0.43235883116722107,
      "learning_rate": 1.1793153628779803e-05,
      "loss": 0.0576,
      "step": 24250
    },
    {
      "epoch": 1.2312807756738922,
      "grad_norm": 0.3197312653064728,
      "learning_rate": 1.1791461495507388e-05,
      "loss": 0.0542,
      "step": 24255
    },
    {
      "epoch": 1.2315345956647545,
      "grad_norm": 0.4243973195552826,
      "learning_rate": 1.178976936223497e-05,
      "loss": 0.0492,
      "step": 24260
    },
    {
      "epoch": 1.231788415655617,
      "grad_norm": 0.3641511797904968,
      "learning_rate": 1.1788077228962554e-05,
      "loss": 0.0613,
      "step": 24265
    },
    {
      "epoch": 1.2320422356464795,
      "grad_norm": 0.343627393245697,
      "learning_rate": 1.1786385095690139e-05,
      "loss": 0.0487,
      "step": 24270
    },
    {
      "epoch": 1.232296055637342,
      "grad_norm": 0.633890688419342,
      "learning_rate": 1.178469296241772e-05,
      "loss": 0.0489,
      "step": 24275
    },
    {
      "epoch": 1.2325498756282045,
      "grad_norm": 0.4130115509033203,
      "learning_rate": 1.1783000829145304e-05,
      "loss": 0.0587,
      "step": 24280
    },
    {
      "epoch": 1.2328036956190669,
      "grad_norm": 0.37340712547302246,
      "learning_rate": 1.1781308695872888e-05,
      "loss": 0.0466,
      "step": 24285
    },
    {
      "epoch": 1.2330575156099295,
      "grad_norm": 1.31352961063385,
      "learning_rate": 1.1779616562600471e-05,
      "loss": 0.053,
      "step": 24290
    },
    {
      "epoch": 1.2333113356007919,
      "grad_norm": 0.25371623039245605,
      "learning_rate": 1.1777924429328057e-05,
      "loss": 0.0506,
      "step": 24295
    },
    {
      "epoch": 1.2335651555916545,
      "grad_norm": 0.5354921817779541,
      "learning_rate": 1.1776232296055638e-05,
      "loss": 0.0569,
      "step": 24300
    },
    {
      "epoch": 1.2338189755825169,
      "grad_norm": 0.3371390104293823,
      "learning_rate": 1.1774540162783222e-05,
      "loss": 0.051,
      "step": 24305
    },
    {
      "epoch": 1.2340727955733795,
      "grad_norm": 0.5659612417221069,
      "learning_rate": 1.1772848029510806e-05,
      "loss": 0.0454,
      "step": 24310
    },
    {
      "epoch": 1.2343266155642418,
      "grad_norm": 0.5271756052970886,
      "learning_rate": 1.1771155896238389e-05,
      "loss": 0.0574,
      "step": 24315
    },
    {
      "epoch": 1.2345804355551042,
      "grad_norm": 0.575239896774292,
      "learning_rate": 1.1769463762965971e-05,
      "loss": 0.0541,
      "step": 24320
    },
    {
      "epoch": 1.2348342555459668,
      "grad_norm": 0.3457287847995758,
      "learning_rate": 1.1767771629693556e-05,
      "loss": 0.0495,
      "step": 24325
    },
    {
      "epoch": 1.2350880755368292,
      "grad_norm": 0.3027961552143097,
      "learning_rate": 1.1766079496421138e-05,
      "loss": 0.0568,
      "step": 24330
    },
    {
      "epoch": 1.2353418955276918,
      "grad_norm": 0.48337095975875854,
      "learning_rate": 1.1764387363148722e-05,
      "loss": 0.0559,
      "step": 24335
    },
    {
      "epoch": 1.2355957155185542,
      "grad_norm": 0.31636735796928406,
      "learning_rate": 1.1762695229876307e-05,
      "loss": 0.0578,
      "step": 24340
    },
    {
      "epoch": 1.2358495355094168,
      "grad_norm": 0.3348773419857025,
      "learning_rate": 1.1761003096603889e-05,
      "loss": 0.0458,
      "step": 24345
    },
    {
      "epoch": 1.2361033555002792,
      "grad_norm": 0.3149411380290985,
      "learning_rate": 1.1759310963331474e-05,
      "loss": 0.0616,
      "step": 24350
    },
    {
      "epoch": 1.2363571754911418,
      "grad_norm": 0.2966812551021576,
      "learning_rate": 1.1757618830059056e-05,
      "loss": 0.0449,
      "step": 24355
    },
    {
      "epoch": 1.2366109954820041,
      "grad_norm": 0.8318333029747009,
      "learning_rate": 1.175592669678664e-05,
      "loss": 0.0491,
      "step": 24360
    },
    {
      "epoch": 1.2368648154728668,
      "grad_norm": 0.8147919178009033,
      "learning_rate": 1.1754234563514225e-05,
      "loss": 0.0512,
      "step": 24365
    },
    {
      "epoch": 1.2371186354637291,
      "grad_norm": 0.39781033992767334,
      "learning_rate": 1.1752542430241806e-05,
      "loss": 0.0461,
      "step": 24370
    },
    {
      "epoch": 1.2373724554545915,
      "grad_norm": 0.33265021443367004,
      "learning_rate": 1.175085029696939e-05,
      "loss": 0.0476,
      "step": 24375
    },
    {
      "epoch": 1.2376262754454541,
      "grad_norm": 0.3713069558143616,
      "learning_rate": 1.1749158163696973e-05,
      "loss": 0.0497,
      "step": 24380
    },
    {
      "epoch": 1.2378800954363165,
      "grad_norm": 0.3097081482410431,
      "learning_rate": 1.1747466030424557e-05,
      "loss": 0.0535,
      "step": 24385
    },
    {
      "epoch": 1.238133915427179,
      "grad_norm": 0.509508490562439,
      "learning_rate": 1.1745773897152139e-05,
      "loss": 0.0474,
      "step": 24390
    },
    {
      "epoch": 1.2383877354180415,
      "grad_norm": 0.2997609078884125,
      "learning_rate": 1.1744081763879724e-05,
      "loss": 0.0446,
      "step": 24395
    },
    {
      "epoch": 1.238641555408904,
      "grad_norm": 0.31849291920661926,
      "learning_rate": 1.1742389630607308e-05,
      "loss": 0.0623,
      "step": 24400
    },
    {
      "epoch": 1.2388953753997665,
      "grad_norm": 0.2741042673587799,
      "learning_rate": 1.1740697497334891e-05,
      "loss": 0.0506,
      "step": 24405
    },
    {
      "epoch": 1.2391491953906288,
      "grad_norm": 0.3714274764060974,
      "learning_rate": 1.1739005364062475e-05,
      "loss": 0.0531,
      "step": 24410
    },
    {
      "epoch": 1.2394030153814914,
      "grad_norm": 0.3533041179180145,
      "learning_rate": 1.1737313230790057e-05,
      "loss": 0.0527,
      "step": 24415
    },
    {
      "epoch": 1.239656835372354,
      "grad_norm": 0.35267361998558044,
      "learning_rate": 1.1735621097517642e-05,
      "loss": 0.0474,
      "step": 24420
    },
    {
      "epoch": 1.2399106553632164,
      "grad_norm": 0.2871090769767761,
      "learning_rate": 1.1733928964245225e-05,
      "loss": 0.0491,
      "step": 24425
    },
    {
      "epoch": 1.2401644753540788,
      "grad_norm": 0.34999963641166687,
      "learning_rate": 1.1732236830972807e-05,
      "loss": 0.049,
      "step": 24430
    },
    {
      "epoch": 1.2404182953449414,
      "grad_norm": 0.7294630408287048,
      "learning_rate": 1.1730544697700392e-05,
      "loss": 0.0574,
      "step": 24435
    },
    {
      "epoch": 1.2406721153358038,
      "grad_norm": 0.31107303500175476,
      "learning_rate": 1.1728852564427974e-05,
      "loss": 0.0428,
      "step": 24440
    },
    {
      "epoch": 1.2409259353266664,
      "grad_norm": 0.4596913158893585,
      "learning_rate": 1.172716043115556e-05,
      "loss": 0.0562,
      "step": 24445
    },
    {
      "epoch": 1.2411797553175288,
      "grad_norm": 0.28400036692619324,
      "learning_rate": 1.1725468297883143e-05,
      "loss": 0.0547,
      "step": 24450
    },
    {
      "epoch": 1.2414335753083914,
      "grad_norm": 0.3861754536628723,
      "learning_rate": 1.1723776164610725e-05,
      "loss": 0.0541,
      "step": 24455
    },
    {
      "epoch": 1.2416873952992538,
      "grad_norm": 0.46610140800476074,
      "learning_rate": 1.172208403133831e-05,
      "loss": 0.0574,
      "step": 24460
    },
    {
      "epoch": 1.2419412152901161,
      "grad_norm": 0.5145004987716675,
      "learning_rate": 1.1720391898065892e-05,
      "loss": 0.0496,
      "step": 24465
    },
    {
      "epoch": 1.2421950352809787,
      "grad_norm": 0.3357173502445221,
      "learning_rate": 1.1718699764793476e-05,
      "loss": 0.0434,
      "step": 24470
    },
    {
      "epoch": 1.2424488552718411,
      "grad_norm": 0.5073157548904419,
      "learning_rate": 1.171700763152106e-05,
      "loss": 0.0535,
      "step": 24475
    },
    {
      "epoch": 1.2427026752627037,
      "grad_norm": 0.5658380389213562,
      "learning_rate": 1.1715315498248643e-05,
      "loss": 0.0507,
      "step": 24480
    },
    {
      "epoch": 1.242956495253566,
      "grad_norm": 0.3627479374408722,
      "learning_rate": 1.1713623364976226e-05,
      "loss": 0.0525,
      "step": 24485
    },
    {
      "epoch": 1.2432103152444287,
      "grad_norm": 0.4254165589809418,
      "learning_rate": 1.171193123170381e-05,
      "loss": 0.0566,
      "step": 24490
    },
    {
      "epoch": 1.243464135235291,
      "grad_norm": 0.28777673840522766,
      "learning_rate": 1.1710239098431393e-05,
      "loss": 0.0485,
      "step": 24495
    },
    {
      "epoch": 1.2437179552261537,
      "grad_norm": 0.5588412284851074,
      "learning_rate": 1.1708546965158979e-05,
      "loss": 0.0498,
      "step": 24500
    },
    {
      "epoch": 1.243971775217016,
      "grad_norm": 0.38440296053886414,
      "learning_rate": 1.170685483188656e-05,
      "loss": 0.054,
      "step": 24505
    },
    {
      "epoch": 1.2442255952078787,
      "grad_norm": 0.434359610080719,
      "learning_rate": 1.1705162698614144e-05,
      "loss": 0.048,
      "step": 24510
    },
    {
      "epoch": 1.244479415198741,
      "grad_norm": 0.4097727835178375,
      "learning_rate": 1.1703470565341728e-05,
      "loss": 0.0579,
      "step": 24515
    },
    {
      "epoch": 1.2447332351896034,
      "grad_norm": 0.49199771881103516,
      "learning_rate": 1.1701778432069311e-05,
      "loss": 0.045,
      "step": 24520
    },
    {
      "epoch": 1.244987055180466,
      "grad_norm": 0.3446272015571594,
      "learning_rate": 1.1700086298796893e-05,
      "loss": 0.0554,
      "step": 24525
    },
    {
      "epoch": 1.2452408751713284,
      "grad_norm": 0.4560483694076538,
      "learning_rate": 1.1698394165524478e-05,
      "loss": 0.0521,
      "step": 24530
    },
    {
      "epoch": 1.245494695162191,
      "grad_norm": 0.30533355474472046,
      "learning_rate": 1.169670203225206e-05,
      "loss": 0.0606,
      "step": 24535
    },
    {
      "epoch": 1.2457485151530534,
      "grad_norm": 0.34249964356422424,
      "learning_rate": 1.1695009898979645e-05,
      "loss": 0.0526,
      "step": 24540
    },
    {
      "epoch": 1.246002335143916,
      "grad_norm": 0.6279357671737671,
      "learning_rate": 1.1693317765707229e-05,
      "loss": 0.0587,
      "step": 24545
    },
    {
      "epoch": 1.2462561551347784,
      "grad_norm": 0.24504680931568146,
      "learning_rate": 1.169162563243481e-05,
      "loss": 0.0506,
      "step": 24550
    },
    {
      "epoch": 1.246509975125641,
      "grad_norm": 0.622235894203186,
      "learning_rate": 1.1689933499162396e-05,
      "loss": 0.0611,
      "step": 24555
    },
    {
      "epoch": 1.2467637951165034,
      "grad_norm": 0.3807246685028076,
      "learning_rate": 1.1688241365889978e-05,
      "loss": 0.0588,
      "step": 24560
    },
    {
      "epoch": 1.247017615107366,
      "grad_norm": 0.35363635420799255,
      "learning_rate": 1.1686549232617561e-05,
      "loss": 0.0525,
      "step": 24565
    },
    {
      "epoch": 1.2472714350982284,
      "grad_norm": 0.415667861700058,
      "learning_rate": 1.1684857099345146e-05,
      "loss": 0.059,
      "step": 24570
    },
    {
      "epoch": 1.2475252550890907,
      "grad_norm": 0.40636613965034485,
      "learning_rate": 1.1683164966072728e-05,
      "loss": 0.0585,
      "step": 24575
    },
    {
      "epoch": 1.2477790750799533,
      "grad_norm": 0.33179885149002075,
      "learning_rate": 1.1681472832800312e-05,
      "loss": 0.0504,
      "step": 24580
    },
    {
      "epoch": 1.2480328950708157,
      "grad_norm": 0.41674908995628357,
      "learning_rate": 1.1679780699527895e-05,
      "loss": 0.054,
      "step": 24585
    },
    {
      "epoch": 1.2482867150616783,
      "grad_norm": 0.4602307081222534,
      "learning_rate": 1.1678088566255479e-05,
      "loss": 0.0541,
      "step": 24590
    },
    {
      "epoch": 1.2485405350525407,
      "grad_norm": 0.3662072718143463,
      "learning_rate": 1.1676396432983064e-05,
      "loss": 0.0487,
      "step": 24595
    },
    {
      "epoch": 1.2487943550434033,
      "grad_norm": 1.340538501739502,
      "learning_rate": 1.1674704299710646e-05,
      "loss": 0.0649,
      "step": 24600
    },
    {
      "epoch": 1.2490481750342657,
      "grad_norm": 0.242966428399086,
      "learning_rate": 1.167301216643823e-05,
      "loss": 0.0593,
      "step": 24605
    },
    {
      "epoch": 1.249301995025128,
      "grad_norm": 0.32781925797462463,
      "learning_rate": 1.1671320033165813e-05,
      "loss": 0.0484,
      "step": 24610
    },
    {
      "epoch": 1.2495558150159907,
      "grad_norm": 0.29135435819625854,
      "learning_rate": 1.1669627899893397e-05,
      "loss": 0.0393,
      "step": 24615
    },
    {
      "epoch": 1.249809635006853,
      "grad_norm": 0.44302451610565186,
      "learning_rate": 1.1667935766620979e-05,
      "loss": 0.0551,
      "step": 24620
    },
    {
      "epoch": 1.2500634549977157,
      "grad_norm": 0.29195067286491394,
      "learning_rate": 1.1666243633348564e-05,
      "loss": 0.0443,
      "step": 24625
    },
    {
      "epoch": 1.250317274988578,
      "grad_norm": 0.5356406569480896,
      "learning_rate": 1.1664551500076147e-05,
      "loss": 0.0483,
      "step": 24630
    },
    {
      "epoch": 1.2505710949794406,
      "grad_norm": 0.3515762984752655,
      "learning_rate": 1.166285936680373e-05,
      "loss": 0.0472,
      "step": 24635
    },
    {
      "epoch": 1.250824914970303,
      "grad_norm": 0.40354248881340027,
      "learning_rate": 1.1661167233531314e-05,
      "loss": 0.0555,
      "step": 24640
    },
    {
      "epoch": 1.2510787349611656,
      "grad_norm": 0.4865587651729584,
      "learning_rate": 1.1659475100258896e-05,
      "loss": 0.0484,
      "step": 24645
    },
    {
      "epoch": 1.251332554952028,
      "grad_norm": 0.43588995933532715,
      "learning_rate": 1.1657782966986482e-05,
      "loss": 0.0626,
      "step": 24650
    },
    {
      "epoch": 1.2515863749428906,
      "grad_norm": 0.5347614884376526,
      "learning_rate": 1.1656090833714065e-05,
      "loss": 0.0614,
      "step": 24655
    },
    {
      "epoch": 1.251840194933753,
      "grad_norm": 0.40371572971343994,
      "learning_rate": 1.1654398700441647e-05,
      "loss": 0.0654,
      "step": 24660
    },
    {
      "epoch": 1.2520940149246154,
      "grad_norm": 0.3678436875343323,
      "learning_rate": 1.1652706567169232e-05,
      "loss": 0.05,
      "step": 24665
    },
    {
      "epoch": 1.252347834915478,
      "grad_norm": 0.4183593988418579,
      "learning_rate": 1.1651014433896814e-05,
      "loss": 0.06,
      "step": 24670
    },
    {
      "epoch": 1.2526016549063403,
      "grad_norm": 0.29606175422668457,
      "learning_rate": 1.1649322300624398e-05,
      "loss": 0.0531,
      "step": 24675
    },
    {
      "epoch": 1.252855474897203,
      "grad_norm": 0.3883073031902313,
      "learning_rate": 1.1647630167351983e-05,
      "loss": 0.0563,
      "step": 24680
    },
    {
      "epoch": 1.2531092948880653,
      "grad_norm": 0.326962947845459,
      "learning_rate": 1.1645938034079565e-05,
      "loss": 0.0499,
      "step": 24685
    },
    {
      "epoch": 1.253363114878928,
      "grad_norm": 0.3533216416835785,
      "learning_rate": 1.164424590080715e-05,
      "loss": 0.0468,
      "step": 24690
    },
    {
      "epoch": 1.2536169348697903,
      "grad_norm": 0.43244466185569763,
      "learning_rate": 1.1642553767534732e-05,
      "loss": 0.0548,
      "step": 24695
    },
    {
      "epoch": 1.2538707548606527,
      "grad_norm": 0.4531016945838928,
      "learning_rate": 1.1640861634262315e-05,
      "loss": 0.0518,
      "step": 24700
    },
    {
      "epoch": 1.2541245748515153,
      "grad_norm": 0.27352747321128845,
      "learning_rate": 1.16391695009899e-05,
      "loss": 0.056,
      "step": 24705
    },
    {
      "epoch": 1.254378394842378,
      "grad_norm": 0.29257744550704956,
      "learning_rate": 1.1637477367717482e-05,
      "loss": 0.0498,
      "step": 24710
    },
    {
      "epoch": 1.2546322148332403,
      "grad_norm": 0.4877859950065613,
      "learning_rate": 1.1635785234445064e-05,
      "loss": 0.0525,
      "step": 24715
    },
    {
      "epoch": 1.2548860348241027,
      "grad_norm": 0.3827455937862396,
      "learning_rate": 1.163409310117265e-05,
      "loss": 0.0478,
      "step": 24720
    },
    {
      "epoch": 1.2551398548149653,
      "grad_norm": 0.6896519660949707,
      "learning_rate": 1.1632400967900233e-05,
      "loss": 0.0474,
      "step": 24725
    },
    {
      "epoch": 1.2553936748058276,
      "grad_norm": 0.4330749809741974,
      "learning_rate": 1.1630708834627815e-05,
      "loss": 0.051,
      "step": 24730
    },
    {
      "epoch": 1.2556474947966902,
      "grad_norm": 0.41606810688972473,
      "learning_rate": 1.16290167013554e-05,
      "loss": 0.0599,
      "step": 24735
    },
    {
      "epoch": 1.2559013147875526,
      "grad_norm": 0.37970244884490967,
      "learning_rate": 1.1627324568082982e-05,
      "loss": 0.0424,
      "step": 24740
    },
    {
      "epoch": 1.2561551347784152,
      "grad_norm": 0.29445701837539673,
      "learning_rate": 1.1625632434810567e-05,
      "loss": 0.0461,
      "step": 24745
    },
    {
      "epoch": 1.2564089547692776,
      "grad_norm": 0.24456225335597992,
      "learning_rate": 1.162394030153815e-05,
      "loss": 0.0462,
      "step": 24750
    },
    {
      "epoch": 1.25666277476014,
      "grad_norm": 0.2942696213722229,
      "learning_rate": 1.1622248168265733e-05,
      "loss": 0.0548,
      "step": 24755
    },
    {
      "epoch": 1.2569165947510026,
      "grad_norm": 0.28085386753082275,
      "learning_rate": 1.1620556034993318e-05,
      "loss": 0.0497,
      "step": 24760
    },
    {
      "epoch": 1.2571704147418652,
      "grad_norm": 0.37461450695991516,
      "learning_rate": 1.16188639017209e-05,
      "loss": 0.0485,
      "step": 24765
    },
    {
      "epoch": 1.2574242347327276,
      "grad_norm": 0.42366278171539307,
      "learning_rate": 1.1617171768448483e-05,
      "loss": 0.0472,
      "step": 24770
    },
    {
      "epoch": 1.25767805472359,
      "grad_norm": 0.3100433945655823,
      "learning_rate": 1.1615479635176068e-05,
      "loss": 0.0529,
      "step": 24775
    },
    {
      "epoch": 1.2579318747144526,
      "grad_norm": 0.32511138916015625,
      "learning_rate": 1.161378750190365e-05,
      "loss": 0.0549,
      "step": 24780
    },
    {
      "epoch": 1.258185694705315,
      "grad_norm": 0.4048891067504883,
      "learning_rate": 1.1612095368631236e-05,
      "loss": 0.0555,
      "step": 24785
    },
    {
      "epoch": 1.2584395146961775,
      "grad_norm": 0.49252721667289734,
      "learning_rate": 1.1610403235358817e-05,
      "loss": 0.0568,
      "step": 24790
    },
    {
      "epoch": 1.25869333468704,
      "grad_norm": 0.3089674413204193,
      "learning_rate": 1.1608711102086401e-05,
      "loss": 0.0545,
      "step": 24795
    },
    {
      "epoch": 1.2589471546779025,
      "grad_norm": 0.4839104115962982,
      "learning_rate": 1.1607018968813986e-05,
      "loss": 0.0586,
      "step": 24800
    },
    {
      "epoch": 1.259200974668765,
      "grad_norm": 0.4469437003135681,
      "learning_rate": 1.1605326835541568e-05,
      "loss": 0.0451,
      "step": 24805
    },
    {
      "epoch": 1.2594547946596273,
      "grad_norm": 0.30239444971084595,
      "learning_rate": 1.1603634702269152e-05,
      "loss": 0.0422,
      "step": 24810
    },
    {
      "epoch": 1.2597086146504899,
      "grad_norm": 0.8179628252983093,
      "learning_rate": 1.1601942568996735e-05,
      "loss": 0.0497,
      "step": 24815
    },
    {
      "epoch": 1.2599624346413525,
      "grad_norm": 0.5173925161361694,
      "learning_rate": 1.1600250435724319e-05,
      "loss": 0.0557,
      "step": 24820
    },
    {
      "epoch": 1.2602162546322149,
      "grad_norm": 0.35866352915763855,
      "learning_rate": 1.15985583024519e-05,
      "loss": 0.051,
      "step": 24825
    },
    {
      "epoch": 1.2604700746230773,
      "grad_norm": 0.4166609048843384,
      "learning_rate": 1.1596866169179486e-05,
      "loss": 0.0497,
      "step": 24830
    },
    {
      "epoch": 1.2607238946139399,
      "grad_norm": 0.31274521350860596,
      "learning_rate": 1.159517403590707e-05,
      "loss": 0.0548,
      "step": 24835
    },
    {
      "epoch": 1.2609777146048022,
      "grad_norm": 0.39399391412734985,
      "learning_rate": 1.1593481902634653e-05,
      "loss": 0.0495,
      "step": 24840
    },
    {
      "epoch": 1.2612315345956646,
      "grad_norm": 0.2745342552661896,
      "learning_rate": 1.1591789769362236e-05,
      "loss": 0.0474,
      "step": 24845
    },
    {
      "epoch": 1.2614853545865272,
      "grad_norm": 0.4054584801197052,
      "learning_rate": 1.1590097636089818e-05,
      "loss": 0.0537,
      "step": 24850
    },
    {
      "epoch": 1.2617391745773898,
      "grad_norm": 0.332953542470932,
      "learning_rate": 1.1588405502817403e-05,
      "loss": 0.0501,
      "step": 24855
    },
    {
      "epoch": 1.2619929945682522,
      "grad_norm": 0.4233452379703522,
      "learning_rate": 1.1586713369544987e-05,
      "loss": 0.0566,
      "step": 24860
    },
    {
      "epoch": 1.2622468145591146,
      "grad_norm": 0.2981891334056854,
      "learning_rate": 1.1585021236272569e-05,
      "loss": 0.0534,
      "step": 24865
    },
    {
      "epoch": 1.2625006345499772,
      "grad_norm": 0.2727615535259247,
      "learning_rate": 1.1583329103000154e-05,
      "loss": 0.0458,
      "step": 24870
    },
    {
      "epoch": 1.2627544545408396,
      "grad_norm": 1.0050932168960571,
      "learning_rate": 1.1581636969727736e-05,
      "loss": 0.0501,
      "step": 24875
    },
    {
      "epoch": 1.2630082745317022,
      "grad_norm": 0.5648157000541687,
      "learning_rate": 1.157994483645532e-05,
      "loss": 0.051,
      "step": 24880
    },
    {
      "epoch": 1.2632620945225645,
      "grad_norm": 0.35903069376945496,
      "learning_rate": 1.1578252703182905e-05,
      "loss": 0.0504,
      "step": 24885
    },
    {
      "epoch": 1.2635159145134272,
      "grad_norm": 0.42427051067352295,
      "learning_rate": 1.1576560569910487e-05,
      "loss": 0.0519,
      "step": 24890
    },
    {
      "epoch": 1.2637697345042895,
      "grad_norm": 0.34407487511634827,
      "learning_rate": 1.1574868436638072e-05,
      "loss": 0.0614,
      "step": 24895
    },
    {
      "epoch": 1.264023554495152,
      "grad_norm": 0.34046033024787903,
      "learning_rate": 1.1573176303365654e-05,
      "loss": 0.0437,
      "step": 24900
    },
    {
      "epoch": 1.2642773744860145,
      "grad_norm": 0.3334759771823883,
      "learning_rate": 1.1571484170093237e-05,
      "loss": 0.0474,
      "step": 24905
    },
    {
      "epoch": 1.2645311944768771,
      "grad_norm": 0.2973962128162384,
      "learning_rate": 1.1569792036820822e-05,
      "loss": 0.0466,
      "step": 24910
    },
    {
      "epoch": 1.2647850144677395,
      "grad_norm": 0.4161064624786377,
      "learning_rate": 1.1568099903548404e-05,
      "loss": 0.0495,
      "step": 24915
    },
    {
      "epoch": 1.2650388344586019,
      "grad_norm": 0.4790429174900055,
      "learning_rate": 1.1566407770275986e-05,
      "loss": 0.0569,
      "step": 24920
    },
    {
      "epoch": 1.2652926544494645,
      "grad_norm": 0.340096116065979,
      "learning_rate": 1.1564715637003571e-05,
      "loss": 0.056,
      "step": 24925
    },
    {
      "epoch": 1.2655464744403269,
      "grad_norm": 0.3699130713939667,
      "learning_rate": 1.1563023503731155e-05,
      "loss": 0.0506,
      "step": 24930
    },
    {
      "epoch": 1.2658002944311895,
      "grad_norm": 0.33714622259140015,
      "learning_rate": 1.156133137045874e-05,
      "loss": 0.05,
      "step": 24935
    },
    {
      "epoch": 1.2660541144220518,
      "grad_norm": 0.35506245493888855,
      "learning_rate": 1.1559639237186322e-05,
      "loss": 0.0555,
      "step": 24940
    },
    {
      "epoch": 1.2663079344129144,
      "grad_norm": 0.2794026732444763,
      "learning_rate": 1.1557947103913904e-05,
      "loss": 0.0554,
      "step": 24945
    },
    {
      "epoch": 1.2665617544037768,
      "grad_norm": 0.28179222345352173,
      "learning_rate": 1.1556254970641489e-05,
      "loss": 0.0469,
      "step": 24950
    },
    {
      "epoch": 1.2668155743946392,
      "grad_norm": 0.31685301661491394,
      "learning_rate": 1.1554562837369073e-05,
      "loss": 0.0568,
      "step": 24955
    },
    {
      "epoch": 1.2670693943855018,
      "grad_norm": 0.38221171498298645,
      "learning_rate": 1.1552870704096654e-05,
      "loss": 0.0509,
      "step": 24960
    },
    {
      "epoch": 1.2673232143763644,
      "grad_norm": 0.38538140058517456,
      "learning_rate": 1.155117857082424e-05,
      "loss": 0.059,
      "step": 24965
    },
    {
      "epoch": 1.2675770343672268,
      "grad_norm": 0.5618222951889038,
      "learning_rate": 1.1549486437551822e-05,
      "loss": 0.0545,
      "step": 24970
    },
    {
      "epoch": 1.2678308543580892,
      "grad_norm": 0.3329094648361206,
      "learning_rate": 1.1547794304279405e-05,
      "loss": 0.0567,
      "step": 24975
    },
    {
      "epoch": 1.2680846743489518,
      "grad_norm": 0.3512543737888336,
      "learning_rate": 1.154610217100699e-05,
      "loss": 0.0565,
      "step": 24980
    },
    {
      "epoch": 1.2683384943398142,
      "grad_norm": 0.38060635328292847,
      "learning_rate": 1.1544410037734572e-05,
      "loss": 0.0436,
      "step": 24985
    },
    {
      "epoch": 1.2685923143306768,
      "grad_norm": 0.3356670141220093,
      "learning_rate": 1.1542717904462157e-05,
      "loss": 0.0486,
      "step": 24990
    },
    {
      "epoch": 1.2688461343215391,
      "grad_norm": 0.3373483717441559,
      "learning_rate": 1.154102577118974e-05,
      "loss": 0.0518,
      "step": 24995
    },
    {
      "epoch": 1.2690999543124017,
      "grad_norm": 0.351968377828598,
      "learning_rate": 1.1539333637917323e-05,
      "loss": 0.0479,
      "step": 25000
    },
    {
      "epoch": 1.2693537743032641,
      "grad_norm": 0.4267289936542511,
      "learning_rate": 1.1537641504644908e-05,
      "loss": 0.0542,
      "step": 25005
    },
    {
      "epoch": 1.2696075942941265,
      "grad_norm": 1.2102017402648926,
      "learning_rate": 1.153594937137249e-05,
      "loss": 0.0677,
      "step": 25010
    },
    {
      "epoch": 1.269861414284989,
      "grad_norm": 0.6700267195701599,
      "learning_rate": 1.1534257238100073e-05,
      "loss": 0.055,
      "step": 25015
    },
    {
      "epoch": 1.2701152342758515,
      "grad_norm": 0.2946222126483917,
      "learning_rate": 1.1532565104827657e-05,
      "loss": 0.0531,
      "step": 25020
    },
    {
      "epoch": 1.270369054266714,
      "grad_norm": 0.25673529505729675,
      "learning_rate": 1.153087297155524e-05,
      "loss": 0.0568,
      "step": 25025
    },
    {
      "epoch": 1.2706228742575765,
      "grad_norm": 0.4724583923816681,
      "learning_rate": 1.1529180838282826e-05,
      "loss": 0.0508,
      "step": 25030
    },
    {
      "epoch": 1.270876694248439,
      "grad_norm": 0.36126697063446045,
      "learning_rate": 1.1527488705010408e-05,
      "loss": 0.0493,
      "step": 25035
    },
    {
      "epoch": 1.2711305142393015,
      "grad_norm": 0.43014219403266907,
      "learning_rate": 1.1525796571737991e-05,
      "loss": 0.0444,
      "step": 25040
    },
    {
      "epoch": 1.2713843342301638,
      "grad_norm": 0.4349735677242279,
      "learning_rate": 1.1524104438465575e-05,
      "loss": 0.053,
      "step": 25045
    },
    {
      "epoch": 1.2716381542210264,
      "grad_norm": 0.42500919103622437,
      "learning_rate": 1.1522412305193158e-05,
      "loss": 0.0484,
      "step": 25050
    },
    {
      "epoch": 1.271891974211889,
      "grad_norm": 0.38369154930114746,
      "learning_rate": 1.152072017192074e-05,
      "loss": 0.0522,
      "step": 25055
    },
    {
      "epoch": 1.2721457942027514,
      "grad_norm": 0.4491569697856903,
      "learning_rate": 1.1519028038648325e-05,
      "loss": 0.0453,
      "step": 25060
    },
    {
      "epoch": 1.2723996141936138,
      "grad_norm": 0.3386807441711426,
      "learning_rate": 1.1517335905375909e-05,
      "loss": 0.0487,
      "step": 25065
    },
    {
      "epoch": 1.2726534341844764,
      "grad_norm": 0.26542624831199646,
      "learning_rate": 1.151564377210349e-05,
      "loss": 0.0479,
      "step": 25070
    },
    {
      "epoch": 1.2729072541753388,
      "grad_norm": 0.33871063590049744,
      "learning_rate": 1.1513951638831076e-05,
      "loss": 0.0567,
      "step": 25075
    },
    {
      "epoch": 1.2731610741662014,
      "grad_norm": 0.46228867769241333,
      "learning_rate": 1.1512259505558658e-05,
      "loss": 0.0491,
      "step": 25080
    },
    {
      "epoch": 1.2734148941570638,
      "grad_norm": 0.6303410530090332,
      "learning_rate": 1.1510567372286243e-05,
      "loss": 0.05,
      "step": 25085
    },
    {
      "epoch": 1.2736687141479264,
      "grad_norm": 0.31501638889312744,
      "learning_rate": 1.1508875239013827e-05,
      "loss": 0.0489,
      "step": 25090
    },
    {
      "epoch": 1.2739225341387888,
      "grad_norm": 0.36004316806793213,
      "learning_rate": 1.1507183105741409e-05,
      "loss": 0.0551,
      "step": 25095
    },
    {
      "epoch": 1.2741763541296511,
      "grad_norm": 0.8616158366203308,
      "learning_rate": 1.1505490972468994e-05,
      "loss": 0.0569,
      "step": 25100
    },
    {
      "epoch": 1.2744301741205137,
      "grad_norm": 0.33611050248146057,
      "learning_rate": 1.1503798839196576e-05,
      "loss": 0.0469,
      "step": 25105
    },
    {
      "epoch": 1.2746839941113763,
      "grad_norm": 0.41049808263778687,
      "learning_rate": 1.1502106705924159e-05,
      "loss": 0.0562,
      "step": 25110
    },
    {
      "epoch": 1.2749378141022387,
      "grad_norm": 0.2955979108810425,
      "learning_rate": 1.1500414572651744e-05,
      "loss": 0.0497,
      "step": 25115
    },
    {
      "epoch": 1.275191634093101,
      "grad_norm": 0.44358739256858826,
      "learning_rate": 1.1498722439379326e-05,
      "loss": 0.0515,
      "step": 25120
    },
    {
      "epoch": 1.2754454540839637,
      "grad_norm": 0.374898761510849,
      "learning_rate": 1.1497030306106908e-05,
      "loss": 0.0522,
      "step": 25125
    },
    {
      "epoch": 1.275699274074826,
      "grad_norm": 0.30691927671432495,
      "learning_rate": 1.1495338172834493e-05,
      "loss": 0.0479,
      "step": 25130
    },
    {
      "epoch": 1.2759530940656887,
      "grad_norm": 0.34827449917793274,
      "learning_rate": 1.1493646039562077e-05,
      "loss": 0.0551,
      "step": 25135
    },
    {
      "epoch": 1.276206914056551,
      "grad_norm": 0.37255775928497314,
      "learning_rate": 1.149195390628966e-05,
      "loss": 0.0509,
      "step": 25140
    },
    {
      "epoch": 1.2764607340474137,
      "grad_norm": 0.5028783082962036,
      "learning_rate": 1.1490261773017244e-05,
      "loss": 0.0527,
      "step": 25145
    },
    {
      "epoch": 1.276714554038276,
      "grad_norm": 0.44755929708480835,
      "learning_rate": 1.1488569639744826e-05,
      "loss": 0.0497,
      "step": 25150
    },
    {
      "epoch": 1.2769683740291384,
      "grad_norm": 0.3543397784233093,
      "learning_rate": 1.1486877506472411e-05,
      "loss": 0.0502,
      "step": 25155
    },
    {
      "epoch": 1.277222194020001,
      "grad_norm": 0.4144035577774048,
      "learning_rate": 1.1485185373199995e-05,
      "loss": 0.0516,
      "step": 25160
    },
    {
      "epoch": 1.2774760140108634,
      "grad_norm": 0.3628363311290741,
      "learning_rate": 1.1483493239927576e-05,
      "loss": 0.0466,
      "step": 25165
    },
    {
      "epoch": 1.277729834001726,
      "grad_norm": 0.3269154727458954,
      "learning_rate": 1.1481801106655162e-05,
      "loss": 0.0546,
      "step": 25170
    },
    {
      "epoch": 1.2779836539925884,
      "grad_norm": 0.4602469205856323,
      "learning_rate": 1.1480108973382744e-05,
      "loss": 0.0483,
      "step": 25175
    },
    {
      "epoch": 1.278237473983451,
      "grad_norm": 0.8606670498847961,
      "learning_rate": 1.1478416840110329e-05,
      "loss": 0.0661,
      "step": 25180
    },
    {
      "epoch": 1.2784912939743134,
      "grad_norm": 0.6975868940353394,
      "learning_rate": 1.1476724706837912e-05,
      "loss": 0.044,
      "step": 25185
    },
    {
      "epoch": 1.2787451139651758,
      "grad_norm": 0.2676074802875519,
      "learning_rate": 1.1475032573565494e-05,
      "loss": 0.0466,
      "step": 25190
    },
    {
      "epoch": 1.2789989339560384,
      "grad_norm": 0.22543856501579285,
      "learning_rate": 1.147334044029308e-05,
      "loss": 0.0542,
      "step": 25195
    },
    {
      "epoch": 1.279252753946901,
      "grad_norm": 0.33862537145614624,
      "learning_rate": 1.1471648307020661e-05,
      "loss": 0.0604,
      "step": 25200
    },
    {
      "epoch": 1.2795065739377633,
      "grad_norm": 0.3779035210609436,
      "learning_rate": 1.1469956173748245e-05,
      "loss": 0.0522,
      "step": 25205
    },
    {
      "epoch": 1.2797603939286257,
      "grad_norm": 0.3730185329914093,
      "learning_rate": 1.146826404047583e-05,
      "loss": 0.0507,
      "step": 25210
    },
    {
      "epoch": 1.2800142139194883,
      "grad_norm": 0.404440313577652,
      "learning_rate": 1.1466571907203412e-05,
      "loss": 0.0565,
      "step": 25215
    },
    {
      "epoch": 1.2802680339103507,
      "grad_norm": 0.39377304911613464,
      "learning_rate": 1.1464879773930995e-05,
      "loss": 0.0566,
      "step": 25220
    },
    {
      "epoch": 1.2805218539012133,
      "grad_norm": 0.3067817986011505,
      "learning_rate": 1.1463187640658579e-05,
      "loss": 0.051,
      "step": 25225
    },
    {
      "epoch": 1.2807756738920757,
      "grad_norm": 0.42959004640579224,
      "learning_rate": 1.1461495507386163e-05,
      "loss": 0.0592,
      "step": 25230
    },
    {
      "epoch": 1.2810294938829383,
      "grad_norm": 0.3348968029022217,
      "learning_rate": 1.1459803374113748e-05,
      "loss": 0.0459,
      "step": 25235
    },
    {
      "epoch": 1.2812833138738007,
      "grad_norm": 0.5492569804191589,
      "learning_rate": 1.145811124084133e-05,
      "loss": 0.051,
      "step": 25240
    },
    {
      "epoch": 1.281537133864663,
      "grad_norm": 0.5650330781936646,
      "learning_rate": 1.1456419107568913e-05,
      "loss": 0.053,
      "step": 25245
    },
    {
      "epoch": 1.2817909538555257,
      "grad_norm": 0.33115166425704956,
      "learning_rate": 1.1454726974296497e-05,
      "loss": 0.0472,
      "step": 25250
    },
    {
      "epoch": 1.2820447738463883,
      "grad_norm": 0.6106663346290588,
      "learning_rate": 1.145303484102408e-05,
      "loss": 0.0512,
      "step": 25255
    },
    {
      "epoch": 1.2822985938372506,
      "grad_norm": 0.6498463749885559,
      "learning_rate": 1.1451342707751662e-05,
      "loss": 0.0539,
      "step": 25260
    },
    {
      "epoch": 1.282552413828113,
      "grad_norm": 0.25299450755119324,
      "learning_rate": 1.1449650574479247e-05,
      "loss": 0.0481,
      "step": 25265
    },
    {
      "epoch": 1.2828062338189756,
      "grad_norm": 0.26644986867904663,
      "learning_rate": 1.1447958441206831e-05,
      "loss": 0.0586,
      "step": 25270
    },
    {
      "epoch": 1.283060053809838,
      "grad_norm": 0.3053162097930908,
      "learning_rate": 1.1446266307934414e-05,
      "loss": 0.0538,
      "step": 25275
    },
    {
      "epoch": 1.2833138738007006,
      "grad_norm": 0.28157272934913635,
      "learning_rate": 1.1444574174661998e-05,
      "loss": 0.0511,
      "step": 25280
    },
    {
      "epoch": 1.283567693791563,
      "grad_norm": 0.4689750671386719,
      "learning_rate": 1.144288204138958e-05,
      "loss": 0.0464,
      "step": 25285
    },
    {
      "epoch": 1.2838215137824256,
      "grad_norm": 0.8079620599746704,
      "learning_rate": 1.1441189908117165e-05,
      "loss": 0.052,
      "step": 25290
    },
    {
      "epoch": 1.284075333773288,
      "grad_norm": 0.263162761926651,
      "learning_rate": 1.1439497774844749e-05,
      "loss": 0.0522,
      "step": 25295
    },
    {
      "epoch": 1.2843291537641504,
      "grad_norm": 0.43071693181991577,
      "learning_rate": 1.143780564157233e-05,
      "loss": 0.0523,
      "step": 25300
    },
    {
      "epoch": 1.284582973755013,
      "grad_norm": 0.31495729088783264,
      "learning_rate": 1.1436113508299916e-05,
      "loss": 0.0613,
      "step": 25305
    },
    {
      "epoch": 1.2848367937458756,
      "grad_norm": 0.38278722763061523,
      "learning_rate": 1.1434421375027498e-05,
      "loss": 0.0524,
      "step": 25310
    },
    {
      "epoch": 1.285090613736738,
      "grad_norm": 0.4674856960773468,
      "learning_rate": 1.1432729241755081e-05,
      "loss": 0.0469,
      "step": 25315
    },
    {
      "epoch": 1.2853444337276003,
      "grad_norm": 0.3894411623477936,
      "learning_rate": 1.1431037108482665e-05,
      "loss": 0.0493,
      "step": 25320
    },
    {
      "epoch": 1.285598253718463,
      "grad_norm": 0.3096882104873657,
      "learning_rate": 1.1429344975210248e-05,
      "loss": 0.0484,
      "step": 25325
    },
    {
      "epoch": 1.2858520737093253,
      "grad_norm": 0.40916988253593445,
      "learning_rate": 1.1427652841937833e-05,
      "loss": 0.0501,
      "step": 25330
    },
    {
      "epoch": 1.286105893700188,
      "grad_norm": 0.4035155177116394,
      "learning_rate": 1.1425960708665415e-05,
      "loss": 0.0509,
      "step": 25335
    },
    {
      "epoch": 1.2863597136910503,
      "grad_norm": 0.2631663382053375,
      "learning_rate": 1.1424268575392999e-05,
      "loss": 0.0508,
      "step": 25340
    },
    {
      "epoch": 1.286613533681913,
      "grad_norm": 1.0878431797027588,
      "learning_rate": 1.1422576442120582e-05,
      "loss": 0.0365,
      "step": 25345
    },
    {
      "epoch": 1.2868673536727753,
      "grad_norm": 0.4171302318572998,
      "learning_rate": 1.1420884308848166e-05,
      "loss": 0.0441,
      "step": 25350
    },
    {
      "epoch": 1.2871211736636377,
      "grad_norm": 0.3274853229522705,
      "learning_rate": 1.1419192175575748e-05,
      "loss": 0.0445,
      "step": 25355
    },
    {
      "epoch": 1.2873749936545003,
      "grad_norm": 0.5124243497848511,
      "learning_rate": 1.1417500042303333e-05,
      "loss": 0.0465,
      "step": 25360
    },
    {
      "epoch": 1.2876288136453626,
      "grad_norm": 0.6019737720489502,
      "learning_rate": 1.1415807909030917e-05,
      "loss": 0.0537,
      "step": 25365
    },
    {
      "epoch": 1.2878826336362252,
      "grad_norm": 0.3629150092601776,
      "learning_rate": 1.1414115775758498e-05,
      "loss": 0.0481,
      "step": 25370
    },
    {
      "epoch": 1.2881364536270876,
      "grad_norm": 0.40465471148490906,
      "learning_rate": 1.1412423642486084e-05,
      "loss": 0.0542,
      "step": 25375
    },
    {
      "epoch": 1.2883902736179502,
      "grad_norm": 0.5901461839675903,
      "learning_rate": 1.1410731509213665e-05,
      "loss": 0.0501,
      "step": 25380
    },
    {
      "epoch": 1.2886440936088126,
      "grad_norm": 0.3407045006752014,
      "learning_rate": 1.140903937594125e-05,
      "loss": 0.0559,
      "step": 25385
    },
    {
      "epoch": 1.288897913599675,
      "grad_norm": 0.41215410828590393,
      "learning_rate": 1.1407347242668834e-05,
      "loss": 0.0538,
      "step": 25390
    },
    {
      "epoch": 1.2891517335905376,
      "grad_norm": 0.3115251064300537,
      "learning_rate": 1.1405655109396416e-05,
      "loss": 0.0394,
      "step": 25395
    },
    {
      "epoch": 1.2894055535814002,
      "grad_norm": 0.2772251069545746,
      "learning_rate": 1.1403962976124001e-05,
      "loss": 0.0421,
      "step": 25400
    },
    {
      "epoch": 1.2896593735722626,
      "grad_norm": 0.3756926953792572,
      "learning_rate": 1.1402270842851583e-05,
      "loss": 0.053,
      "step": 25405
    },
    {
      "epoch": 1.289913193563125,
      "grad_norm": 0.5077455639839172,
      "learning_rate": 1.1400578709579167e-05,
      "loss": 0.0505,
      "step": 25410
    },
    {
      "epoch": 1.2901670135539876,
      "grad_norm": 0.2544682025909424,
      "learning_rate": 1.1398886576306752e-05,
      "loss": 0.0451,
      "step": 25415
    },
    {
      "epoch": 1.29042083354485,
      "grad_norm": 0.39733994007110596,
      "learning_rate": 1.1397194443034334e-05,
      "loss": 0.0488,
      "step": 25420
    },
    {
      "epoch": 1.2906746535357125,
      "grad_norm": 0.3462377190589905,
      "learning_rate": 1.1395502309761919e-05,
      "loss": 0.0463,
      "step": 25425
    },
    {
      "epoch": 1.290928473526575,
      "grad_norm": 0.3922407031059265,
      "learning_rate": 1.1393810176489501e-05,
      "loss": 0.0542,
      "step": 25430
    },
    {
      "epoch": 1.2911822935174375,
      "grad_norm": 0.3106277883052826,
      "learning_rate": 1.1392118043217084e-05,
      "loss": 0.0574,
      "step": 25435
    },
    {
      "epoch": 1.2914361135083,
      "grad_norm": 0.40110254287719727,
      "learning_rate": 1.139042590994467e-05,
      "loss": 0.045,
      "step": 25440
    },
    {
      "epoch": 1.2916899334991623,
      "grad_norm": 0.44177547097206116,
      "learning_rate": 1.1388733776672252e-05,
      "loss": 0.061,
      "step": 25445
    },
    {
      "epoch": 1.2919437534900249,
      "grad_norm": 0.44935235381126404,
      "learning_rate": 1.1387041643399835e-05,
      "loss": 0.0483,
      "step": 25450
    },
    {
      "epoch": 1.2921975734808875,
      "grad_norm": 0.32007691264152527,
      "learning_rate": 1.1385349510127419e-05,
      "loss": 0.0477,
      "step": 25455
    },
    {
      "epoch": 1.2924513934717499,
      "grad_norm": 0.4596714675426483,
      "learning_rate": 1.1383657376855002e-05,
      "loss": 0.0447,
      "step": 25460
    },
    {
      "epoch": 1.2927052134626122,
      "grad_norm": 0.3188910484313965,
      "learning_rate": 1.1381965243582584e-05,
      "loss": 0.0502,
      "step": 25465
    },
    {
      "epoch": 1.2929590334534748,
      "grad_norm": 0.43237102031707764,
      "learning_rate": 1.138027311031017e-05,
      "loss": 0.0472,
      "step": 25470
    },
    {
      "epoch": 1.2932128534443372,
      "grad_norm": 0.2928224205970764,
      "learning_rate": 1.1378580977037753e-05,
      "loss": 0.0484,
      "step": 25475
    },
    {
      "epoch": 1.2934666734351998,
      "grad_norm": 0.2517949044704437,
      "learning_rate": 1.1376888843765336e-05,
      "loss": 0.0438,
      "step": 25480
    },
    {
      "epoch": 1.2937204934260622,
      "grad_norm": 0.3528852164745331,
      "learning_rate": 1.137519671049292e-05,
      "loss": 0.0468,
      "step": 25485
    },
    {
      "epoch": 1.2939743134169248,
      "grad_norm": 0.2391764372587204,
      "learning_rate": 1.1373504577220502e-05,
      "loss": 0.0477,
      "step": 25490
    },
    {
      "epoch": 1.2942281334077872,
      "grad_norm": 0.2708483040332794,
      "learning_rate": 1.1371812443948087e-05,
      "loss": 0.0438,
      "step": 25495
    },
    {
      "epoch": 1.2944819533986496,
      "grad_norm": 0.38952237367630005,
      "learning_rate": 1.1370120310675669e-05,
      "loss": 0.0619,
      "step": 25500
    },
    {
      "epoch": 1.2947357733895122,
      "grad_norm": 0.682826817035675,
      "learning_rate": 1.1368428177403252e-05,
      "loss": 0.0586,
      "step": 25505
    },
    {
      "epoch": 1.2949895933803746,
      "grad_norm": 0.2654576897621155,
      "learning_rate": 1.1366736044130838e-05,
      "loss": 0.0527,
      "step": 25510
    },
    {
      "epoch": 1.2952434133712372,
      "grad_norm": 0.5474327802658081,
      "learning_rate": 1.136504391085842e-05,
      "loss": 0.0615,
      "step": 25515
    },
    {
      "epoch": 1.2954972333620995,
      "grad_norm": 0.263549268245697,
      "learning_rate": 1.1363351777586003e-05,
      "loss": 0.0462,
      "step": 25520
    },
    {
      "epoch": 1.2957510533529621,
      "grad_norm": 0.6336255073547363,
      "learning_rate": 1.1361659644313587e-05,
      "loss": 0.053,
      "step": 25525
    },
    {
      "epoch": 1.2960048733438245,
      "grad_norm": 0.26589730381965637,
      "learning_rate": 1.135996751104117e-05,
      "loss": 0.0509,
      "step": 25530
    },
    {
      "epoch": 1.296258693334687,
      "grad_norm": 0.3139997124671936,
      "learning_rate": 1.1358275377768755e-05,
      "loss": 0.057,
      "step": 25535
    },
    {
      "epoch": 1.2965125133255495,
      "grad_norm": 0.40500542521476746,
      "learning_rate": 1.1356583244496337e-05,
      "loss": 0.0496,
      "step": 25540
    },
    {
      "epoch": 1.2967663333164121,
      "grad_norm": 0.3901219367980957,
      "learning_rate": 1.135489111122392e-05,
      "loss": 0.0513,
      "step": 25545
    },
    {
      "epoch": 1.2970201533072745,
      "grad_norm": 0.22926051914691925,
      "learning_rate": 1.1353198977951504e-05,
      "loss": 0.0445,
      "step": 25550
    },
    {
      "epoch": 1.2972739732981369,
      "grad_norm": 0.3791753351688385,
      "learning_rate": 1.1351506844679088e-05,
      "loss": 0.0644,
      "step": 25555
    },
    {
      "epoch": 1.2975277932889995,
      "grad_norm": 0.7541919350624084,
      "learning_rate": 1.134981471140667e-05,
      "loss": 0.0392,
      "step": 25560
    },
    {
      "epoch": 1.2977816132798619,
      "grad_norm": 0.4487682580947876,
      "learning_rate": 1.1348122578134255e-05,
      "loss": 0.0476,
      "step": 25565
    },
    {
      "epoch": 1.2980354332707245,
      "grad_norm": 0.4413999021053314,
      "learning_rate": 1.1346430444861838e-05,
      "loss": 0.0533,
      "step": 25570
    },
    {
      "epoch": 1.2982892532615868,
      "grad_norm": 0.29243093729019165,
      "learning_rate": 1.1344738311589422e-05,
      "loss": 0.052,
      "step": 25575
    },
    {
      "epoch": 1.2985430732524494,
      "grad_norm": 0.36786124110221863,
      "learning_rate": 1.1343046178317006e-05,
      "loss": 0.0542,
      "step": 25580
    },
    {
      "epoch": 1.2987968932433118,
      "grad_norm": 0.6370408535003662,
      "learning_rate": 1.1341354045044587e-05,
      "loss": 0.0532,
      "step": 25585
    },
    {
      "epoch": 1.2990507132341742,
      "grad_norm": 0.6985453367233276,
      "learning_rate": 1.1339661911772173e-05,
      "loss": 0.0422,
      "step": 25590
    },
    {
      "epoch": 1.2993045332250368,
      "grad_norm": 0.9178361296653748,
      "learning_rate": 1.1337969778499756e-05,
      "loss": 0.0451,
      "step": 25595
    },
    {
      "epoch": 1.2995583532158994,
      "grad_norm": 0.5521905422210693,
      "learning_rate": 1.1336277645227338e-05,
      "loss": 0.0528,
      "step": 25600
    },
    {
      "epoch": 1.2998121732067618,
      "grad_norm": 0.33282098174095154,
      "learning_rate": 1.1334585511954923e-05,
      "loss": 0.0559,
      "step": 25605
    },
    {
      "epoch": 1.3000659931976242,
      "grad_norm": 0.4263562262058258,
      "learning_rate": 1.1332893378682505e-05,
      "loss": 0.0561,
      "step": 25610
    },
    {
      "epoch": 1.3003198131884868,
      "grad_norm": 0.5796880722045898,
      "learning_rate": 1.1331201245410089e-05,
      "loss": 0.049,
      "step": 25615
    },
    {
      "epoch": 1.3005736331793492,
      "grad_norm": 0.2884933650493622,
      "learning_rate": 1.1329509112137674e-05,
      "loss": 0.0465,
      "step": 25620
    },
    {
      "epoch": 1.3008274531702118,
      "grad_norm": 0.3866657018661499,
      "learning_rate": 1.1327816978865256e-05,
      "loss": 0.0595,
      "step": 25625
    },
    {
      "epoch": 1.3010812731610741,
      "grad_norm": 0.5192033052444458,
      "learning_rate": 1.1326124845592841e-05,
      "loss": 0.0481,
      "step": 25630
    },
    {
      "epoch": 1.3013350931519367,
      "grad_norm": 0.7006744146347046,
      "learning_rate": 1.1324432712320423e-05,
      "loss": 0.0502,
      "step": 25635
    },
    {
      "epoch": 1.3015889131427991,
      "grad_norm": 0.3354194760322571,
      "learning_rate": 1.1322740579048006e-05,
      "loss": 0.0528,
      "step": 25640
    },
    {
      "epoch": 1.3018427331336615,
      "grad_norm": 0.3940652310848236,
      "learning_rate": 1.1321048445775592e-05,
      "loss": 0.0429,
      "step": 25645
    },
    {
      "epoch": 1.302096553124524,
      "grad_norm": 0.3817504942417145,
      "learning_rate": 1.1319356312503173e-05,
      "loss": 0.0487,
      "step": 25650
    },
    {
      "epoch": 1.3023503731153867,
      "grad_norm": 0.35483697056770325,
      "learning_rate": 1.1317664179230757e-05,
      "loss": 0.0546,
      "step": 25655
    },
    {
      "epoch": 1.302604193106249,
      "grad_norm": 0.48677176237106323,
      "learning_rate": 1.131597204595834e-05,
      "loss": 0.0511,
      "step": 25660
    },
    {
      "epoch": 1.3028580130971115,
      "grad_norm": 0.2799428701400757,
      "learning_rate": 1.1314279912685924e-05,
      "loss": 0.0508,
      "step": 25665
    },
    {
      "epoch": 1.303111833087974,
      "grad_norm": 0.3958439528942108,
      "learning_rate": 1.131258777941351e-05,
      "loss": 0.0571,
      "step": 25670
    },
    {
      "epoch": 1.3033656530788365,
      "grad_norm": 0.3991915285587311,
      "learning_rate": 1.1310895646141091e-05,
      "loss": 0.0592,
      "step": 25675
    },
    {
      "epoch": 1.3036194730696988,
      "grad_norm": 0.541405200958252,
      "learning_rate": 1.1309203512868675e-05,
      "loss": 0.0509,
      "step": 25680
    },
    {
      "epoch": 1.3038732930605614,
      "grad_norm": 0.4154348075389862,
      "learning_rate": 1.1307511379596258e-05,
      "loss": 0.0466,
      "step": 25685
    },
    {
      "epoch": 1.304127113051424,
      "grad_norm": 0.37647801637649536,
      "learning_rate": 1.1305819246323842e-05,
      "loss": 0.0512,
      "step": 25690
    },
    {
      "epoch": 1.3043809330422864,
      "grad_norm": 0.3786545991897583,
      "learning_rate": 1.1304127113051424e-05,
      "loss": 0.045,
      "step": 25695
    },
    {
      "epoch": 1.3046347530331488,
      "grad_norm": 0.3503737151622772,
      "learning_rate": 1.1302434979779009e-05,
      "loss": 0.0596,
      "step": 25700
    },
    {
      "epoch": 1.3048885730240114,
      "grad_norm": 0.3044377565383911,
      "learning_rate": 1.130074284650659e-05,
      "loss": 0.0542,
      "step": 25705
    },
    {
      "epoch": 1.3051423930148738,
      "grad_norm": 0.37505218386650085,
      "learning_rate": 1.1299050713234174e-05,
      "loss": 0.0511,
      "step": 25710
    },
    {
      "epoch": 1.3053962130057364,
      "grad_norm": 0.5570057034492493,
      "learning_rate": 1.129735857996176e-05,
      "loss": 0.0441,
      "step": 25715
    },
    {
      "epoch": 1.3056500329965988,
      "grad_norm": 0.3466768264770508,
      "learning_rate": 1.1295666446689341e-05,
      "loss": 0.0513,
      "step": 25720
    },
    {
      "epoch": 1.3059038529874614,
      "grad_norm": 0.44723430275917053,
      "learning_rate": 1.1293974313416927e-05,
      "loss": 0.0512,
      "step": 25725
    },
    {
      "epoch": 1.3061576729783237,
      "grad_norm": 0.34250175952911377,
      "learning_rate": 1.1292282180144508e-05,
      "loss": 0.0449,
      "step": 25730
    },
    {
      "epoch": 1.3064114929691861,
      "grad_norm": 0.30145788192749023,
      "learning_rate": 1.1290590046872092e-05,
      "loss": 0.0477,
      "step": 25735
    },
    {
      "epoch": 1.3066653129600487,
      "grad_norm": 0.2720540463924408,
      "learning_rate": 1.1288897913599677e-05,
      "loss": 0.0452,
      "step": 25740
    },
    {
      "epoch": 1.3069191329509113,
      "grad_norm": 0.4038149416446686,
      "learning_rate": 1.1287205780327259e-05,
      "loss": 0.0462,
      "step": 25745
    },
    {
      "epoch": 1.3071729529417737,
      "grad_norm": 0.3033551573753357,
      "learning_rate": 1.1285513647054843e-05,
      "loss": 0.046,
      "step": 25750
    },
    {
      "epoch": 1.307426772932636,
      "grad_norm": 0.48980191349983215,
      "learning_rate": 1.1283821513782426e-05,
      "loss": 0.0501,
      "step": 25755
    },
    {
      "epoch": 1.3076805929234987,
      "grad_norm": 0.382783442735672,
      "learning_rate": 1.128212938051001e-05,
      "loss": 0.0501,
      "step": 25760
    },
    {
      "epoch": 1.307934412914361,
      "grad_norm": 0.3193140923976898,
      "learning_rate": 1.1280437247237592e-05,
      "loss": 0.0469,
      "step": 25765
    },
    {
      "epoch": 1.3081882329052237,
      "grad_norm": 0.4315343499183655,
      "learning_rate": 1.1278745113965177e-05,
      "loss": 0.0522,
      "step": 25770
    },
    {
      "epoch": 1.308442052896086,
      "grad_norm": 0.402089387178421,
      "learning_rate": 1.127705298069276e-05,
      "loss": 0.0533,
      "step": 25775
    },
    {
      "epoch": 1.3086958728869487,
      "grad_norm": 0.3406699299812317,
      "learning_rate": 1.1275360847420344e-05,
      "loss": 0.0467,
      "step": 25780
    },
    {
      "epoch": 1.308949692877811,
      "grad_norm": 0.48056527972221375,
      "learning_rate": 1.1273668714147927e-05,
      "loss": 0.059,
      "step": 25785
    },
    {
      "epoch": 1.3092035128686734,
      "grad_norm": 0.390354186296463,
      "learning_rate": 1.127197658087551e-05,
      "loss": 0.0484,
      "step": 25790
    },
    {
      "epoch": 1.309457332859536,
      "grad_norm": 0.4609234631061554,
      "learning_rate": 1.1270284447603095e-05,
      "loss": 0.0498,
      "step": 25795
    },
    {
      "epoch": 1.3097111528503986,
      "grad_norm": 0.19950836896896362,
      "learning_rate": 1.1268592314330678e-05,
      "loss": 0.0433,
      "step": 25800
    },
    {
      "epoch": 1.309964972841261,
      "grad_norm": 0.23946751654148102,
      "learning_rate": 1.126690018105826e-05,
      "loss": 0.0502,
      "step": 25805
    },
    {
      "epoch": 1.3102187928321234,
      "grad_norm": 0.37170344591140747,
      "learning_rate": 1.1265208047785845e-05,
      "loss": 0.0499,
      "step": 25810
    },
    {
      "epoch": 1.310472612822986,
      "grad_norm": 0.35320132970809937,
      "learning_rate": 1.1263515914513427e-05,
      "loss": 0.0483,
      "step": 25815
    },
    {
      "epoch": 1.3107264328138484,
      "grad_norm": 0.2944350838661194,
      "learning_rate": 1.1261823781241012e-05,
      "loss": 0.05,
      "step": 25820
    },
    {
      "epoch": 1.310980252804711,
      "grad_norm": 0.6497879028320312,
      "learning_rate": 1.1260131647968596e-05,
      "loss": 0.0475,
      "step": 25825
    },
    {
      "epoch": 1.3112340727955734,
      "grad_norm": 0.42111438512802124,
      "learning_rate": 1.1258439514696178e-05,
      "loss": 0.047,
      "step": 25830
    },
    {
      "epoch": 1.311487892786436,
      "grad_norm": 0.2944899797439575,
      "learning_rate": 1.1256747381423763e-05,
      "loss": 0.052,
      "step": 25835
    },
    {
      "epoch": 1.3117417127772983,
      "grad_norm": 0.3662308156490326,
      "learning_rate": 1.1255055248151345e-05,
      "loss": 0.0492,
      "step": 25840
    },
    {
      "epoch": 1.3119955327681607,
      "grad_norm": 0.40842756628990173,
      "learning_rate": 1.1253363114878928e-05,
      "loss": 0.053,
      "step": 25845
    },
    {
      "epoch": 1.3122493527590233,
      "grad_norm": 0.4699631929397583,
      "learning_rate": 1.1251670981606514e-05,
      "loss": 0.0571,
      "step": 25850
    },
    {
      "epoch": 1.3125031727498857,
      "grad_norm": 0.3852125108242035,
      "learning_rate": 1.1249978848334095e-05,
      "loss": 0.0473,
      "step": 25855
    },
    {
      "epoch": 1.3127569927407483,
      "grad_norm": 0.3017500042915344,
      "learning_rate": 1.1248286715061679e-05,
      "loss": 0.0454,
      "step": 25860
    },
    {
      "epoch": 1.3130108127316107,
      "grad_norm": 0.2818055748939514,
      "learning_rate": 1.1246594581789263e-05,
      "loss": 0.0413,
      "step": 25865
    },
    {
      "epoch": 1.3132646327224733,
      "grad_norm": 0.34530168771743774,
      "learning_rate": 1.1244902448516846e-05,
      "loss": 0.0443,
      "step": 25870
    },
    {
      "epoch": 1.3135184527133357,
      "grad_norm": 0.41481852531433105,
      "learning_rate": 1.1243210315244431e-05,
      "loss": 0.0523,
      "step": 25875
    },
    {
      "epoch": 1.313772272704198,
      "grad_norm": 0.293000727891922,
      "learning_rate": 1.1241518181972013e-05,
      "loss": 0.052,
      "step": 25880
    },
    {
      "epoch": 1.3140260926950607,
      "grad_norm": 0.4565964341163635,
      "learning_rate": 1.1239826048699595e-05,
      "loss": 0.0477,
      "step": 25885
    },
    {
      "epoch": 1.3142799126859233,
      "grad_norm": 0.5154722332954407,
      "learning_rate": 1.123813391542718e-05,
      "loss": 0.0455,
      "step": 25890
    },
    {
      "epoch": 1.3145337326767856,
      "grad_norm": 0.33028629422187805,
      "learning_rate": 1.1236441782154764e-05,
      "loss": 0.0489,
      "step": 25895
    },
    {
      "epoch": 1.314787552667648,
      "grad_norm": 0.3494911789894104,
      "learning_rate": 1.1234749648882346e-05,
      "loss": 0.0531,
      "step": 25900
    },
    {
      "epoch": 1.3150413726585106,
      "grad_norm": 0.46097883582115173,
      "learning_rate": 1.1233057515609931e-05,
      "loss": 0.0521,
      "step": 25905
    },
    {
      "epoch": 1.315295192649373,
      "grad_norm": 0.3439639210700989,
      "learning_rate": 1.1231365382337513e-05,
      "loss": 0.04,
      "step": 25910
    },
    {
      "epoch": 1.3155490126402356,
      "grad_norm": 0.28974398970603943,
      "learning_rate": 1.1229673249065098e-05,
      "loss": 0.0522,
      "step": 25915
    },
    {
      "epoch": 1.315802832631098,
      "grad_norm": 0.35030993819236755,
      "learning_rate": 1.1227981115792681e-05,
      "loss": 0.0477,
      "step": 25920
    },
    {
      "epoch": 1.3160566526219606,
      "grad_norm": 0.3437199294567108,
      "learning_rate": 1.1226288982520263e-05,
      "loss": 0.0418,
      "step": 25925
    },
    {
      "epoch": 1.316310472612823,
      "grad_norm": 0.32538580894470215,
      "learning_rate": 1.1224596849247849e-05,
      "loss": 0.0499,
      "step": 25930
    },
    {
      "epoch": 1.3165642926036853,
      "grad_norm": 1.0620239973068237,
      "learning_rate": 1.122290471597543e-05,
      "loss": 0.0524,
      "step": 25935
    },
    {
      "epoch": 1.316818112594548,
      "grad_norm": 0.3068779706954956,
      "learning_rate": 1.1221212582703014e-05,
      "loss": 0.0544,
      "step": 25940
    },
    {
      "epoch": 1.3170719325854106,
      "grad_norm": 0.39146485924720764,
      "learning_rate": 1.12195204494306e-05,
      "loss": 0.0482,
      "step": 25945
    },
    {
      "epoch": 1.317325752576273,
      "grad_norm": 0.7082789540290833,
      "learning_rate": 1.1217828316158181e-05,
      "loss": 0.049,
      "step": 25950
    },
    {
      "epoch": 1.3175795725671353,
      "grad_norm": 0.27648693323135376,
      "learning_rate": 1.1216136182885765e-05,
      "loss": 0.0517,
      "step": 25955
    },
    {
      "epoch": 1.317833392557998,
      "grad_norm": 0.3340929448604584,
      "learning_rate": 1.1214444049613348e-05,
      "loss": 0.0419,
      "step": 25960
    },
    {
      "epoch": 1.3180872125488603,
      "grad_norm": 0.28633975982666016,
      "learning_rate": 1.1212751916340932e-05,
      "loss": 0.0463,
      "step": 25965
    },
    {
      "epoch": 1.318341032539723,
      "grad_norm": 0.3893069624900818,
      "learning_rate": 1.1211059783068517e-05,
      "loss": 0.0547,
      "step": 25970
    },
    {
      "epoch": 1.3185948525305853,
      "grad_norm": 0.39877814054489136,
      "learning_rate": 1.1209367649796099e-05,
      "loss": 0.0544,
      "step": 25975
    },
    {
      "epoch": 1.3188486725214479,
      "grad_norm": 0.46034538745880127,
      "learning_rate": 1.1207675516523682e-05,
      "loss": 0.0521,
      "step": 25980
    },
    {
      "epoch": 1.3191024925123103,
      "grad_norm": 0.4608680009841919,
      "learning_rate": 1.1205983383251266e-05,
      "loss": 0.0545,
      "step": 25985
    },
    {
      "epoch": 1.3193563125031726,
      "grad_norm": 0.4240970313549042,
      "learning_rate": 1.120429124997885e-05,
      "loss": 0.0551,
      "step": 25990
    },
    {
      "epoch": 1.3196101324940352,
      "grad_norm": 0.2486318200826645,
      "learning_rate": 1.1202599116706431e-05,
      "loss": 0.0487,
      "step": 25995
    },
    {
      "epoch": 1.3198639524848976,
      "grad_norm": 0.3128660321235657,
      "learning_rate": 1.1200906983434017e-05,
      "loss": 0.0485,
      "step": 26000
    },
    {
      "epoch": 1.3201177724757602,
      "grad_norm": 0.36881980299949646,
      "learning_rate": 1.11992148501616e-05,
      "loss": 0.0485,
      "step": 26005
    },
    {
      "epoch": 1.3203715924666226,
      "grad_norm": 0.49103283882141113,
      "learning_rate": 1.1197522716889182e-05,
      "loss": 0.0563,
      "step": 26010
    },
    {
      "epoch": 1.3206254124574852,
      "grad_norm": 0.2944199740886688,
      "learning_rate": 1.1195830583616767e-05,
      "loss": 0.0502,
      "step": 26015
    },
    {
      "epoch": 1.3208792324483476,
      "grad_norm": 0.29078707098960876,
      "learning_rate": 1.1194138450344349e-05,
      "loss": 0.0485,
      "step": 26020
    },
    {
      "epoch": 1.32113305243921,
      "grad_norm": 0.5979177355766296,
      "learning_rate": 1.1192446317071934e-05,
      "loss": 0.0465,
      "step": 26025
    },
    {
      "epoch": 1.3213868724300726,
      "grad_norm": 0.6528639197349548,
      "learning_rate": 1.1190754183799518e-05,
      "loss": 0.0519,
      "step": 26030
    },
    {
      "epoch": 1.3216406924209352,
      "grad_norm": 0.44926077127456665,
      "learning_rate": 1.11890620505271e-05,
      "loss": 0.0522,
      "step": 26035
    },
    {
      "epoch": 1.3218945124117976,
      "grad_norm": 0.4386170506477356,
      "learning_rate": 1.1187369917254685e-05,
      "loss": 0.0482,
      "step": 26040
    },
    {
      "epoch": 1.32214833240266,
      "grad_norm": 0.3527389168739319,
      "learning_rate": 1.1185677783982267e-05,
      "loss": 0.0573,
      "step": 26045
    },
    {
      "epoch": 1.3224021523935225,
      "grad_norm": 0.24854600429534912,
      "learning_rate": 1.118398565070985e-05,
      "loss": 0.0373,
      "step": 26050
    },
    {
      "epoch": 1.322655972384385,
      "grad_norm": 0.45639580488204956,
      "learning_rate": 1.1182293517437436e-05,
      "loss": 0.0534,
      "step": 26055
    },
    {
      "epoch": 1.3229097923752475,
      "grad_norm": 1.2171883583068848,
      "learning_rate": 1.1180601384165017e-05,
      "loss": 0.0497,
      "step": 26060
    },
    {
      "epoch": 1.32316361236611,
      "grad_norm": 0.4451811909675598,
      "learning_rate": 1.1178909250892603e-05,
      "loss": 0.0467,
      "step": 26065
    },
    {
      "epoch": 1.3234174323569725,
      "grad_norm": 0.2924448251724243,
      "learning_rate": 1.1177217117620184e-05,
      "loss": 0.062,
      "step": 26070
    },
    {
      "epoch": 1.323671252347835,
      "grad_norm": 0.2893722355365753,
      "learning_rate": 1.1175524984347768e-05,
      "loss": 0.046,
      "step": 26075
    },
    {
      "epoch": 1.3239250723386973,
      "grad_norm": 0.21968765556812286,
      "learning_rate": 1.1173832851075353e-05,
      "loss": 0.0409,
      "step": 26080
    },
    {
      "epoch": 1.3241788923295599,
      "grad_norm": 0.4578390419483185,
      "learning_rate": 1.1172140717802935e-05,
      "loss": 0.049,
      "step": 26085
    },
    {
      "epoch": 1.3244327123204225,
      "grad_norm": 0.41233834624290466,
      "learning_rate": 1.1170448584530517e-05,
      "loss": 0.0499,
      "step": 26090
    },
    {
      "epoch": 1.3246865323112849,
      "grad_norm": 0.5314094424247742,
      "learning_rate": 1.1168756451258102e-05,
      "loss": 0.0514,
      "step": 26095
    },
    {
      "epoch": 1.3249403523021472,
      "grad_norm": 0.3177417814731598,
      "learning_rate": 1.1167064317985686e-05,
      "loss": 0.0518,
      "step": 26100
    },
    {
      "epoch": 1.3251941722930098,
      "grad_norm": 0.321321040391922,
      "learning_rate": 1.1165372184713268e-05,
      "loss": 0.0467,
      "step": 26105
    },
    {
      "epoch": 1.3254479922838722,
      "grad_norm": 0.5315613150596619,
      "learning_rate": 1.1163680051440853e-05,
      "loss": 0.0553,
      "step": 26110
    },
    {
      "epoch": 1.3257018122747348,
      "grad_norm": 0.4163055717945099,
      "learning_rate": 1.1161987918168435e-05,
      "loss": 0.0518,
      "step": 26115
    },
    {
      "epoch": 1.3259556322655972,
      "grad_norm": 0.5081636905670166,
      "learning_rate": 1.116029578489602e-05,
      "loss": 0.0479,
      "step": 26120
    },
    {
      "epoch": 1.3262094522564598,
      "grad_norm": 0.37250816822052,
      "learning_rate": 1.1158603651623603e-05,
      "loss": 0.052,
      "step": 26125
    },
    {
      "epoch": 1.3264632722473222,
      "grad_norm": 0.4670708477497101,
      "learning_rate": 1.1156911518351185e-05,
      "loss": 0.059,
      "step": 26130
    },
    {
      "epoch": 1.3267170922381846,
      "grad_norm": 0.28547587990760803,
      "learning_rate": 1.115521938507877e-05,
      "loss": 0.0557,
      "step": 26135
    },
    {
      "epoch": 1.3269709122290472,
      "grad_norm": 0.3063032627105713,
      "learning_rate": 1.1153527251806352e-05,
      "loss": 0.0508,
      "step": 26140
    },
    {
      "epoch": 1.3272247322199098,
      "grad_norm": 0.2864580452442169,
      "learning_rate": 1.1151835118533936e-05,
      "loss": 0.0552,
      "step": 26145
    },
    {
      "epoch": 1.3274785522107722,
      "grad_norm": 0.32648247480392456,
      "learning_rate": 1.1150142985261521e-05,
      "loss": 0.0478,
      "step": 26150
    },
    {
      "epoch": 1.3277323722016345,
      "grad_norm": 0.482799768447876,
      "learning_rate": 1.1148450851989103e-05,
      "loss": 0.0559,
      "step": 26155
    },
    {
      "epoch": 1.3279861921924971,
      "grad_norm": 0.3012193739414215,
      "learning_rate": 1.1146758718716688e-05,
      "loss": 0.0579,
      "step": 26160
    },
    {
      "epoch": 1.3282400121833595,
      "grad_norm": 0.332803338766098,
      "learning_rate": 1.114506658544427e-05,
      "loss": 0.0492,
      "step": 26165
    },
    {
      "epoch": 1.3284938321742221,
      "grad_norm": 0.46283581852912903,
      "learning_rate": 1.1143374452171854e-05,
      "loss": 0.0457,
      "step": 26170
    },
    {
      "epoch": 1.3287476521650845,
      "grad_norm": 0.39792996644973755,
      "learning_rate": 1.1141682318899439e-05,
      "loss": 0.0524,
      "step": 26175
    },
    {
      "epoch": 1.329001472155947,
      "grad_norm": 0.4443744122982025,
      "learning_rate": 1.113999018562702e-05,
      "loss": 0.0535,
      "step": 26180
    },
    {
      "epoch": 1.3292552921468095,
      "grad_norm": 0.3356640934944153,
      "learning_rate": 1.1138298052354604e-05,
      "loss": 0.0464,
      "step": 26185
    },
    {
      "epoch": 1.3295091121376719,
      "grad_norm": 0.4023672640323639,
      "learning_rate": 1.1136605919082188e-05,
      "loss": 0.0567,
      "step": 26190
    },
    {
      "epoch": 1.3297629321285345,
      "grad_norm": 0.43528980016708374,
      "learning_rate": 1.1134913785809771e-05,
      "loss": 0.0532,
      "step": 26195
    },
    {
      "epoch": 1.3300167521193969,
      "grad_norm": 0.7247572541236877,
      "learning_rate": 1.1133221652537353e-05,
      "loss": 0.052,
      "step": 26200
    },
    {
      "epoch": 1.3302705721102595,
      "grad_norm": 0.48162537813186646,
      "learning_rate": 1.1131529519264938e-05,
      "loss": 0.0607,
      "step": 26205
    },
    {
      "epoch": 1.3305243921011218,
      "grad_norm": 0.42504459619522095,
      "learning_rate": 1.1129837385992522e-05,
      "loss": 0.0472,
      "step": 26210
    },
    {
      "epoch": 1.3307782120919844,
      "grad_norm": 0.3819892108440399,
      "learning_rate": 1.1128145252720106e-05,
      "loss": 0.0466,
      "step": 26215
    },
    {
      "epoch": 1.3310320320828468,
      "grad_norm": 0.505711019039154,
      "learning_rate": 1.1126453119447689e-05,
      "loss": 0.0595,
      "step": 26220
    },
    {
      "epoch": 1.3312858520737092,
      "grad_norm": 0.44940781593322754,
      "learning_rate": 1.1124760986175271e-05,
      "loss": 0.0543,
      "step": 26225
    },
    {
      "epoch": 1.3315396720645718,
      "grad_norm": 0.8618845343589783,
      "learning_rate": 1.1123068852902856e-05,
      "loss": 0.0493,
      "step": 26230
    },
    {
      "epoch": 1.3317934920554344,
      "grad_norm": 0.5297338962554932,
      "learning_rate": 1.112137671963044e-05,
      "loss": 0.0579,
      "step": 26235
    },
    {
      "epoch": 1.3320473120462968,
      "grad_norm": 0.362981379032135,
      "learning_rate": 1.1119684586358022e-05,
      "loss": 0.0471,
      "step": 26240
    },
    {
      "epoch": 1.3323011320371592,
      "grad_norm": 0.3708239495754242,
      "learning_rate": 1.1117992453085607e-05,
      "loss": 0.051,
      "step": 26245
    },
    {
      "epoch": 1.3325549520280218,
      "grad_norm": 0.6170402765274048,
      "learning_rate": 1.1116300319813189e-05,
      "loss": 0.0552,
      "step": 26250
    },
    {
      "epoch": 1.3328087720188841,
      "grad_norm": 0.3187443017959595,
      "learning_rate": 1.1114608186540772e-05,
      "loss": 0.0465,
      "step": 26255
    },
    {
      "epoch": 1.3330625920097467,
      "grad_norm": 0.42820754647254944,
      "learning_rate": 1.1112916053268357e-05,
      "loss": 0.0607,
      "step": 26260
    },
    {
      "epoch": 1.3333164120006091,
      "grad_norm": 0.3025449812412262,
      "learning_rate": 1.111122391999594e-05,
      "loss": 0.0456,
      "step": 26265
    },
    {
      "epoch": 1.3335702319914717,
      "grad_norm": 0.40699583292007446,
      "learning_rate": 1.1109531786723525e-05,
      "loss": 0.0461,
      "step": 26270
    },
    {
      "epoch": 1.3338240519823341,
      "grad_norm": 0.3597828447818756,
      "learning_rate": 1.1107839653451106e-05,
      "loss": 0.0546,
      "step": 26275
    },
    {
      "epoch": 1.3340778719731965,
      "grad_norm": 0.2918253540992737,
      "learning_rate": 1.110614752017869e-05,
      "loss": 0.0478,
      "step": 26280
    },
    {
      "epoch": 1.334331691964059,
      "grad_norm": 0.36164140701293945,
      "learning_rate": 1.1104455386906275e-05,
      "loss": 0.0475,
      "step": 26285
    },
    {
      "epoch": 1.3345855119549217,
      "grad_norm": 0.3688763976097107,
      "learning_rate": 1.1102763253633857e-05,
      "loss": 0.0578,
      "step": 26290
    },
    {
      "epoch": 1.334839331945784,
      "grad_norm": 0.37201929092407227,
      "learning_rate": 1.1101071120361439e-05,
      "loss": 0.0551,
      "step": 26295
    },
    {
      "epoch": 1.3350931519366465,
      "grad_norm": 0.2655457854270935,
      "learning_rate": 1.1099378987089024e-05,
      "loss": 0.0497,
      "step": 26300
    },
    {
      "epoch": 1.335346971927509,
      "grad_norm": 0.37188592553138733,
      "learning_rate": 1.1097686853816608e-05,
      "loss": 0.0532,
      "step": 26305
    },
    {
      "epoch": 1.3356007919183714,
      "grad_norm": 0.3306806683540344,
      "learning_rate": 1.1095994720544191e-05,
      "loss": 0.0599,
      "step": 26310
    },
    {
      "epoch": 1.335854611909234,
      "grad_norm": 0.3516598045825958,
      "learning_rate": 1.1094302587271775e-05,
      "loss": 0.0461,
      "step": 26315
    },
    {
      "epoch": 1.3361084319000964,
      "grad_norm": 0.2789323031902313,
      "learning_rate": 1.1092610453999357e-05,
      "loss": 0.0518,
      "step": 26320
    },
    {
      "epoch": 1.336362251890959,
      "grad_norm": 0.3232646584510803,
      "learning_rate": 1.1090918320726942e-05,
      "loss": 0.0489,
      "step": 26325
    },
    {
      "epoch": 1.3366160718818214,
      "grad_norm": 1.0127755403518677,
      "learning_rate": 1.1089226187454525e-05,
      "loss": 0.0518,
      "step": 26330
    },
    {
      "epoch": 1.3368698918726838,
      "grad_norm": 0.2616908550262451,
      "learning_rate": 1.1087534054182107e-05,
      "loss": 0.0481,
      "step": 26335
    },
    {
      "epoch": 1.3371237118635464,
      "grad_norm": 0.46473437547683716,
      "learning_rate": 1.1085841920909692e-05,
      "loss": 0.053,
      "step": 26340
    },
    {
      "epoch": 1.3373775318544088,
      "grad_norm": 0.34728309512138367,
      "learning_rate": 1.1084149787637274e-05,
      "loss": 0.0489,
      "step": 26345
    },
    {
      "epoch": 1.3376313518452714,
      "grad_norm": 0.3350659906864166,
      "learning_rate": 1.1082457654364858e-05,
      "loss": 0.0474,
      "step": 26350
    },
    {
      "epoch": 1.3378851718361338,
      "grad_norm": 0.43599680066108704,
      "learning_rate": 1.1080765521092443e-05,
      "loss": 0.0467,
      "step": 26355
    },
    {
      "epoch": 1.3381389918269964,
      "grad_norm": 0.42427605390548706,
      "learning_rate": 1.1079073387820025e-05,
      "loss": 0.0498,
      "step": 26360
    },
    {
      "epoch": 1.3383928118178587,
      "grad_norm": 0.4834648668766022,
      "learning_rate": 1.107738125454761e-05,
      "loss": 0.0596,
      "step": 26365
    },
    {
      "epoch": 1.3386466318087211,
      "grad_norm": 0.2929410934448242,
      "learning_rate": 1.1075689121275192e-05,
      "loss": 0.0538,
      "step": 26370
    },
    {
      "epoch": 1.3389004517995837,
      "grad_norm": 0.578696072101593,
      "learning_rate": 1.1073996988002776e-05,
      "loss": 0.0525,
      "step": 26375
    },
    {
      "epoch": 1.3391542717904463,
      "grad_norm": 0.4364106059074402,
      "learning_rate": 1.107230485473036e-05,
      "loss": 0.0528,
      "step": 26380
    },
    {
      "epoch": 1.3394080917813087,
      "grad_norm": 0.406345397233963,
      "learning_rate": 1.1070612721457943e-05,
      "loss": 0.0466,
      "step": 26385
    },
    {
      "epoch": 1.339661911772171,
      "grad_norm": 0.4139547049999237,
      "learning_rate": 1.1068920588185526e-05,
      "loss": 0.0534,
      "step": 26390
    },
    {
      "epoch": 1.3399157317630337,
      "grad_norm": 0.28089866042137146,
      "learning_rate": 1.106722845491311e-05,
      "loss": 0.0519,
      "step": 26395
    },
    {
      "epoch": 1.340169551753896,
      "grad_norm": 0.5075027346611023,
      "learning_rate": 1.1065536321640693e-05,
      "loss": 0.0598,
      "step": 26400
    },
    {
      "epoch": 1.3404233717447587,
      "grad_norm": 0.3053726255893707,
      "learning_rate": 1.1063844188368279e-05,
      "loss": 0.0403,
      "step": 26405
    },
    {
      "epoch": 1.340677191735621,
      "grad_norm": 0.48526138067245483,
      "learning_rate": 1.106215205509586e-05,
      "loss": 0.0561,
      "step": 26410
    },
    {
      "epoch": 1.3409310117264837,
      "grad_norm": 0.4186187982559204,
      "learning_rate": 1.1060459921823444e-05,
      "loss": 0.0574,
      "step": 26415
    },
    {
      "epoch": 1.341184831717346,
      "grad_norm": 0.6404673457145691,
      "learning_rate": 1.1058767788551027e-05,
      "loss": 0.0542,
      "step": 26420
    },
    {
      "epoch": 1.3414386517082084,
      "grad_norm": 0.28784051537513733,
      "learning_rate": 1.1057075655278611e-05,
      "loss": 0.0543,
      "step": 26425
    },
    {
      "epoch": 1.341692471699071,
      "grad_norm": 0.48208898305892944,
      "learning_rate": 1.1055383522006193e-05,
      "loss": 0.0492,
      "step": 26430
    },
    {
      "epoch": 1.3419462916899336,
      "grad_norm": 0.405626505613327,
      "learning_rate": 1.1053691388733778e-05,
      "loss": 0.0535,
      "step": 26435
    },
    {
      "epoch": 1.342200111680796,
      "grad_norm": 0.39484503865242004,
      "learning_rate": 1.1051999255461362e-05,
      "loss": 0.0552,
      "step": 26440
    },
    {
      "epoch": 1.3424539316716584,
      "grad_norm": 0.28524714708328247,
      "learning_rate": 1.1050307122188944e-05,
      "loss": 0.0544,
      "step": 26445
    },
    {
      "epoch": 1.342707751662521,
      "grad_norm": 0.27638357877731323,
      "learning_rate": 1.1048614988916529e-05,
      "loss": 0.0517,
      "step": 26450
    },
    {
      "epoch": 1.3429615716533834,
      "grad_norm": 0.6675527095794678,
      "learning_rate": 1.104692285564411e-05,
      "loss": 0.0635,
      "step": 26455
    },
    {
      "epoch": 1.343215391644246,
      "grad_norm": 0.24780261516571045,
      "learning_rate": 1.1045230722371696e-05,
      "loss": 0.0421,
      "step": 26460
    },
    {
      "epoch": 1.3434692116351084,
      "grad_norm": 0.3185769021511078,
      "learning_rate": 1.104353858909928e-05,
      "loss": 0.0428,
      "step": 26465
    },
    {
      "epoch": 1.343723031625971,
      "grad_norm": 0.38865941762924194,
      "learning_rate": 1.1041846455826861e-05,
      "loss": 0.0616,
      "step": 26470
    },
    {
      "epoch": 1.3439768516168333,
      "grad_norm": 0.44646382331848145,
      "learning_rate": 1.1040154322554446e-05,
      "loss": 0.0631,
      "step": 26475
    },
    {
      "epoch": 1.3442306716076957,
      "grad_norm": 0.25871074199676514,
      "learning_rate": 1.1038462189282028e-05,
      "loss": 0.0424,
      "step": 26480
    },
    {
      "epoch": 1.3444844915985583,
      "grad_norm": 0.35550186038017273,
      "learning_rate": 1.1036770056009612e-05,
      "loss": 0.0449,
      "step": 26485
    },
    {
      "epoch": 1.344738311589421,
      "grad_norm": 0.3502696752548218,
      "learning_rate": 1.1035077922737195e-05,
      "loss": 0.0565,
      "step": 26490
    },
    {
      "epoch": 1.3449921315802833,
      "grad_norm": 0.49729862809181213,
      "learning_rate": 1.1033385789464779e-05,
      "loss": 0.0475,
      "step": 26495
    },
    {
      "epoch": 1.3452459515711457,
      "grad_norm": 0.35838475823402405,
      "learning_rate": 1.103169365619236e-05,
      "loss": 0.0508,
      "step": 26500
    },
    {
      "epoch": 1.3454997715620083,
      "grad_norm": 0.34287241101264954,
      "learning_rate": 1.1030001522919946e-05,
      "loss": 0.0574,
      "step": 26505
    },
    {
      "epoch": 1.3457535915528707,
      "grad_norm": 0.26396411657333374,
      "learning_rate": 1.102830938964753e-05,
      "loss": 0.0536,
      "step": 26510
    },
    {
      "epoch": 1.3460074115437333,
      "grad_norm": 0.31700873374938965,
      "learning_rate": 1.1026617256375113e-05,
      "loss": 0.0564,
      "step": 26515
    },
    {
      "epoch": 1.3462612315345956,
      "grad_norm": 0.3169967234134674,
      "learning_rate": 1.1024925123102697e-05,
      "loss": 0.0393,
      "step": 26520
    },
    {
      "epoch": 1.3465150515254583,
      "grad_norm": 0.39078614115715027,
      "learning_rate": 1.1023232989830279e-05,
      "loss": 0.0432,
      "step": 26525
    },
    {
      "epoch": 1.3467688715163206,
      "grad_norm": 0.3944978713989258,
      "learning_rate": 1.1021540856557864e-05,
      "loss": 0.0499,
      "step": 26530
    },
    {
      "epoch": 1.347022691507183,
      "grad_norm": 0.5801070332527161,
      "learning_rate": 1.1019848723285447e-05,
      "loss": 0.051,
      "step": 26535
    },
    {
      "epoch": 1.3472765114980456,
      "grad_norm": 0.5324280261993408,
      "learning_rate": 1.101815659001303e-05,
      "loss": 0.0494,
      "step": 26540
    },
    {
      "epoch": 1.347530331488908,
      "grad_norm": 0.5855595469474792,
      "learning_rate": 1.1016464456740614e-05,
      "loss": 0.0562,
      "step": 26545
    },
    {
      "epoch": 1.3477841514797706,
      "grad_norm": 0.3513774275779724,
      "learning_rate": 1.1014772323468196e-05,
      "loss": 0.0463,
      "step": 26550
    },
    {
      "epoch": 1.348037971470633,
      "grad_norm": 0.4091087877750397,
      "learning_rate": 1.1013080190195781e-05,
      "loss": 0.0414,
      "step": 26555
    },
    {
      "epoch": 1.3482917914614956,
      "grad_norm": 0.27667826414108276,
      "learning_rate": 1.1011388056923365e-05,
      "loss": 0.0423,
      "step": 26560
    },
    {
      "epoch": 1.348545611452358,
      "grad_norm": 0.2812269628047943,
      "learning_rate": 1.1009695923650947e-05,
      "loss": 0.0389,
      "step": 26565
    },
    {
      "epoch": 1.3487994314432203,
      "grad_norm": 0.31701719760894775,
      "learning_rate": 1.1008003790378532e-05,
      "loss": 0.052,
      "step": 26570
    },
    {
      "epoch": 1.349053251434083,
      "grad_norm": 0.24721567332744598,
      "learning_rate": 1.1006311657106114e-05,
      "loss": 0.042,
      "step": 26575
    },
    {
      "epoch": 1.3493070714249455,
      "grad_norm": 0.32528063654899597,
      "learning_rate": 1.1004619523833698e-05,
      "loss": 0.0449,
      "step": 26580
    },
    {
      "epoch": 1.349560891415808,
      "grad_norm": 0.35122808814048767,
      "learning_rate": 1.1002927390561283e-05,
      "loss": 0.0546,
      "step": 26585
    },
    {
      "epoch": 1.3498147114066703,
      "grad_norm": 0.39924734830856323,
      "learning_rate": 1.1001235257288865e-05,
      "loss": 0.052,
      "step": 26590
    },
    {
      "epoch": 1.350068531397533,
      "grad_norm": 0.38985615968704224,
      "learning_rate": 1.0999543124016448e-05,
      "loss": 0.0497,
      "step": 26595
    },
    {
      "epoch": 1.3503223513883953,
      "grad_norm": 0.4624733328819275,
      "learning_rate": 1.0997850990744032e-05,
      "loss": 0.041,
      "step": 26600
    },
    {
      "epoch": 1.350576171379258,
      "grad_norm": 0.2167605310678482,
      "learning_rate": 1.0996158857471615e-05,
      "loss": 0.0487,
      "step": 26605
    },
    {
      "epoch": 1.3508299913701203,
      "grad_norm": 0.47157183289527893,
      "learning_rate": 1.09944667241992e-05,
      "loss": 0.0501,
      "step": 26610
    },
    {
      "epoch": 1.3510838113609829,
      "grad_norm": 0.2828313410282135,
      "learning_rate": 1.0992774590926782e-05,
      "loss": 0.0461,
      "step": 26615
    },
    {
      "epoch": 1.3513376313518453,
      "grad_norm": 0.3092103898525238,
      "learning_rate": 1.0991082457654366e-05,
      "loss": 0.0462,
      "step": 26620
    },
    {
      "epoch": 1.3515914513427076,
      "grad_norm": 1.7850135564804077,
      "learning_rate": 1.098939032438195e-05,
      "loss": 0.0522,
      "step": 26625
    },
    {
      "epoch": 1.3518452713335702,
      "grad_norm": 0.596710205078125,
      "learning_rate": 1.0987698191109533e-05,
      "loss": 0.0459,
      "step": 26630
    },
    {
      "epoch": 1.3520990913244328,
      "grad_norm": 0.5141996145248413,
      "learning_rate": 1.0986006057837115e-05,
      "loss": 0.0423,
      "step": 26635
    },
    {
      "epoch": 1.3523529113152952,
      "grad_norm": 0.24726712703704834,
      "learning_rate": 1.09843139245647e-05,
      "loss": 0.0492,
      "step": 26640
    },
    {
      "epoch": 1.3526067313061576,
      "grad_norm": 0.4391964077949524,
      "learning_rate": 1.0982621791292284e-05,
      "loss": 0.0477,
      "step": 26645
    },
    {
      "epoch": 1.3528605512970202,
      "grad_norm": 0.7240193486213684,
      "learning_rate": 1.0980929658019865e-05,
      "loss": 0.0449,
      "step": 26650
    },
    {
      "epoch": 1.3531143712878826,
      "grad_norm": 0.3740382492542267,
      "learning_rate": 1.097923752474745e-05,
      "loss": 0.0519,
      "step": 26655
    },
    {
      "epoch": 1.3533681912787452,
      "grad_norm": 0.3421389162540436,
      "learning_rate": 1.0977545391475033e-05,
      "loss": 0.0401,
      "step": 26660
    },
    {
      "epoch": 1.3536220112696076,
      "grad_norm": 0.34667646884918213,
      "learning_rate": 1.0975853258202618e-05,
      "loss": 0.0411,
      "step": 26665
    },
    {
      "epoch": 1.3538758312604702,
      "grad_norm": 0.28433099389076233,
      "learning_rate": 1.09741611249302e-05,
      "loss": 0.0476,
      "step": 26670
    },
    {
      "epoch": 1.3541296512513326,
      "grad_norm": 0.41038647294044495,
      "learning_rate": 1.0972468991657783e-05,
      "loss": 0.0419,
      "step": 26675
    },
    {
      "epoch": 1.354383471242195,
      "grad_norm": 0.3607024550437927,
      "learning_rate": 1.0970776858385368e-05,
      "loss": 0.0516,
      "step": 26680
    },
    {
      "epoch": 1.3546372912330575,
      "grad_norm": 0.2682955265045166,
      "learning_rate": 1.096908472511295e-05,
      "loss": 0.05,
      "step": 26685
    },
    {
      "epoch": 1.35489111122392,
      "grad_norm": 0.3467102646827698,
      "learning_rate": 1.0967392591840534e-05,
      "loss": 0.0475,
      "step": 26690
    },
    {
      "epoch": 1.3551449312147825,
      "grad_norm": 0.3220283091068268,
      "learning_rate": 1.0965700458568117e-05,
      "loss": 0.0461,
      "step": 26695
    },
    {
      "epoch": 1.355398751205645,
      "grad_norm": 0.35406333208084106,
      "learning_rate": 1.0964008325295701e-05,
      "loss": 0.0526,
      "step": 26700
    },
    {
      "epoch": 1.3556525711965075,
      "grad_norm": 0.3669520616531372,
      "learning_rate": 1.0962316192023286e-05,
      "loss": 0.0453,
      "step": 26705
    },
    {
      "epoch": 1.3559063911873699,
      "grad_norm": 0.5051359534263611,
      "learning_rate": 1.0960624058750868e-05,
      "loss": 0.0543,
      "step": 26710
    },
    {
      "epoch": 1.3561602111782323,
      "grad_norm": 0.33588844537734985,
      "learning_rate": 1.0958931925478452e-05,
      "loss": 0.0479,
      "step": 26715
    },
    {
      "epoch": 1.3564140311690949,
      "grad_norm": 0.29593461751937866,
      "learning_rate": 1.0957239792206035e-05,
      "loss": 0.0385,
      "step": 26720
    },
    {
      "epoch": 1.3566678511599575,
      "grad_norm": 0.3803567886352539,
      "learning_rate": 1.0955547658933619e-05,
      "loss": 0.0549,
      "step": 26725
    },
    {
      "epoch": 1.3569216711508199,
      "grad_norm": 0.6764376163482666,
      "learning_rate": 1.09538555256612e-05,
      "loss": 0.0641,
      "step": 26730
    },
    {
      "epoch": 1.3571754911416822,
      "grad_norm": 0.3458918631076813,
      "learning_rate": 1.0952163392388786e-05,
      "loss": 0.045,
      "step": 26735
    },
    {
      "epoch": 1.3574293111325448,
      "grad_norm": 0.45418867468833923,
      "learning_rate": 1.095047125911637e-05,
      "loss": 0.0469,
      "step": 26740
    },
    {
      "epoch": 1.3576831311234072,
      "grad_norm": 0.3401700258255005,
      "learning_rate": 1.0948779125843951e-05,
      "loss": 0.0508,
      "step": 26745
    },
    {
      "epoch": 1.3579369511142698,
      "grad_norm": 0.39936044812202454,
      "learning_rate": 1.0947086992571536e-05,
      "loss": 0.0469,
      "step": 26750
    },
    {
      "epoch": 1.3581907711051322,
      "grad_norm": 0.45466601848602295,
      "learning_rate": 1.0945394859299118e-05,
      "loss": 0.0469,
      "step": 26755
    },
    {
      "epoch": 1.3584445910959948,
      "grad_norm": 0.3149279057979584,
      "learning_rate": 1.0943702726026703e-05,
      "loss": 0.0537,
      "step": 26760
    },
    {
      "epoch": 1.3586984110868572,
      "grad_norm": 0.3702044188976288,
      "learning_rate": 1.0942010592754287e-05,
      "loss": 0.0425,
      "step": 26765
    },
    {
      "epoch": 1.3589522310777196,
      "grad_norm": 0.33377188444137573,
      "learning_rate": 1.0940318459481869e-05,
      "loss": 0.0511,
      "step": 26770
    },
    {
      "epoch": 1.3592060510685822,
      "grad_norm": 0.2742324769496918,
      "learning_rate": 1.0938626326209454e-05,
      "loss": 0.042,
      "step": 26775
    },
    {
      "epoch": 1.3594598710594448,
      "grad_norm": 0.4314764142036438,
      "learning_rate": 1.0936934192937036e-05,
      "loss": 0.046,
      "step": 26780
    },
    {
      "epoch": 1.3597136910503071,
      "grad_norm": 0.44701406359672546,
      "learning_rate": 1.093524205966462e-05,
      "loss": 0.0511,
      "step": 26785
    },
    {
      "epoch": 1.3599675110411695,
      "grad_norm": 0.34280073642730713,
      "learning_rate": 1.0933549926392205e-05,
      "loss": 0.0533,
      "step": 26790
    },
    {
      "epoch": 1.3602213310320321,
      "grad_norm": 0.4505353271961212,
      "learning_rate": 1.0931857793119787e-05,
      "loss": 0.0584,
      "step": 26795
    },
    {
      "epoch": 1.3604751510228945,
      "grad_norm": 0.3516581356525421,
      "learning_rate": 1.0930165659847372e-05,
      "loss": 0.0501,
      "step": 26800
    },
    {
      "epoch": 1.3607289710137571,
      "grad_norm": 0.38294366002082825,
      "learning_rate": 1.0928473526574954e-05,
      "loss": 0.0412,
      "step": 26805
    },
    {
      "epoch": 1.3609827910046195,
      "grad_norm": 0.2905791401863098,
      "learning_rate": 1.0926781393302537e-05,
      "loss": 0.0421,
      "step": 26810
    },
    {
      "epoch": 1.361236610995482,
      "grad_norm": 0.45999136567115784,
      "learning_rate": 1.0925089260030122e-05,
      "loss": 0.0564,
      "step": 26815
    },
    {
      "epoch": 1.3614904309863445,
      "grad_norm": 0.29272863268852234,
      "learning_rate": 1.0923397126757704e-05,
      "loss": 0.042,
      "step": 26820
    },
    {
      "epoch": 1.3617442509772069,
      "grad_norm": 0.42476195096969604,
      "learning_rate": 1.0921704993485288e-05,
      "loss": 0.0474,
      "step": 26825
    },
    {
      "epoch": 1.3619980709680695,
      "grad_norm": 0.3225775361061096,
      "learning_rate": 1.0920012860212871e-05,
      "loss": 0.0546,
      "step": 26830
    },
    {
      "epoch": 1.3622518909589318,
      "grad_norm": 0.6205003261566162,
      "learning_rate": 1.0918320726940455e-05,
      "loss": 0.0485,
      "step": 26835
    },
    {
      "epoch": 1.3625057109497944,
      "grad_norm": 0.2589777112007141,
      "learning_rate": 1.0916628593668037e-05,
      "loss": 0.0493,
      "step": 26840
    },
    {
      "epoch": 1.3627595309406568,
      "grad_norm": 0.44953614473342896,
      "learning_rate": 1.0914936460395622e-05,
      "loss": 0.0607,
      "step": 26845
    },
    {
      "epoch": 1.3630133509315194,
      "grad_norm": 0.3794015944004059,
      "learning_rate": 1.0913244327123206e-05,
      "loss": 0.0462,
      "step": 26850
    },
    {
      "epoch": 1.3632671709223818,
      "grad_norm": 0.35074856877326965,
      "learning_rate": 1.0911552193850789e-05,
      "loss": 0.0481,
      "step": 26855
    },
    {
      "epoch": 1.3635209909132442,
      "grad_norm": 0.9279066920280457,
      "learning_rate": 1.0909860060578373e-05,
      "loss": 0.0451,
      "step": 26860
    },
    {
      "epoch": 1.3637748109041068,
      "grad_norm": 0.5168582797050476,
      "learning_rate": 1.0908167927305954e-05,
      "loss": 0.049,
      "step": 26865
    },
    {
      "epoch": 1.3640286308949694,
      "grad_norm": 0.3608698844909668,
      "learning_rate": 1.090647579403354e-05,
      "loss": 0.0502,
      "step": 26870
    },
    {
      "epoch": 1.3642824508858318,
      "grad_norm": 0.5553390383720398,
      "learning_rate": 1.0904783660761122e-05,
      "loss": 0.0471,
      "step": 26875
    },
    {
      "epoch": 1.3645362708766942,
      "grad_norm": 0.28950929641723633,
      "learning_rate": 1.0903091527488705e-05,
      "loss": 0.0463,
      "step": 26880
    },
    {
      "epoch": 1.3647900908675568,
      "grad_norm": 0.4183814525604248,
      "learning_rate": 1.090139939421629e-05,
      "loss": 0.0508,
      "step": 26885
    },
    {
      "epoch": 1.3650439108584191,
      "grad_norm": 0.3526676893234253,
      "learning_rate": 1.0899707260943872e-05,
      "loss": 0.0442,
      "step": 26890
    },
    {
      "epoch": 1.3652977308492817,
      "grad_norm": 0.2606235146522522,
      "learning_rate": 1.0898015127671456e-05,
      "loss": 0.0496,
      "step": 26895
    },
    {
      "epoch": 1.3655515508401441,
      "grad_norm": 0.3592001795768738,
      "learning_rate": 1.089632299439904e-05,
      "loss": 0.0474,
      "step": 26900
    },
    {
      "epoch": 1.3658053708310067,
      "grad_norm": 0.36163032054901123,
      "learning_rate": 1.0894630861126623e-05,
      "loss": 0.0582,
      "step": 26905
    },
    {
      "epoch": 1.366059190821869,
      "grad_norm": 0.3937353789806366,
      "learning_rate": 1.0892938727854208e-05,
      "loss": 0.0497,
      "step": 26910
    },
    {
      "epoch": 1.3663130108127315,
      "grad_norm": 0.37750929594039917,
      "learning_rate": 1.089124659458179e-05,
      "loss": 0.0506,
      "step": 26915
    },
    {
      "epoch": 1.366566830803594,
      "grad_norm": 0.33422794938087463,
      "learning_rate": 1.0889554461309373e-05,
      "loss": 0.0427,
      "step": 26920
    },
    {
      "epoch": 1.3668206507944567,
      "grad_norm": 0.36222967505455017,
      "learning_rate": 1.0887862328036957e-05,
      "loss": 0.0571,
      "step": 26925
    },
    {
      "epoch": 1.367074470785319,
      "grad_norm": 0.33056381344795227,
      "learning_rate": 1.088617019476454e-05,
      "loss": 0.0397,
      "step": 26930
    },
    {
      "epoch": 1.3673282907761815,
      "grad_norm": 0.3641473054885864,
      "learning_rate": 1.0884478061492122e-05,
      "loss": 0.0448,
      "step": 26935
    },
    {
      "epoch": 1.367582110767044,
      "grad_norm": 0.4339211583137512,
      "learning_rate": 1.0882785928219708e-05,
      "loss": 0.0483,
      "step": 26940
    },
    {
      "epoch": 1.3678359307579064,
      "grad_norm": 0.31715840101242065,
      "learning_rate": 1.0881093794947291e-05,
      "loss": 0.0501,
      "step": 26945
    },
    {
      "epoch": 1.368089750748769,
      "grad_norm": 0.32084155082702637,
      "learning_rate": 1.0879401661674875e-05,
      "loss": 0.0495,
      "step": 26950
    },
    {
      "epoch": 1.3683435707396314,
      "grad_norm": 0.40496891736984253,
      "learning_rate": 1.0877709528402458e-05,
      "loss": 0.0462,
      "step": 26955
    },
    {
      "epoch": 1.368597390730494,
      "grad_norm": 0.289095014333725,
      "learning_rate": 1.087601739513004e-05,
      "loss": 0.0506,
      "step": 26960
    },
    {
      "epoch": 1.3688512107213564,
      "grad_norm": 0.5172923803329468,
      "learning_rate": 1.0874325261857625e-05,
      "loss": 0.0424,
      "step": 26965
    },
    {
      "epoch": 1.3691050307122188,
      "grad_norm": 0.30263659358024597,
      "learning_rate": 1.0872633128585209e-05,
      "loss": 0.0437,
      "step": 26970
    },
    {
      "epoch": 1.3693588507030814,
      "grad_norm": 0.3625158369541168,
      "learning_rate": 1.087094099531279e-05,
      "loss": 0.0478,
      "step": 26975
    },
    {
      "epoch": 1.369612670693944,
      "grad_norm": 0.5003167390823364,
      "learning_rate": 1.0869248862040376e-05,
      "loss": 0.0494,
      "step": 26980
    },
    {
      "epoch": 1.3698664906848064,
      "grad_norm": 0.42639175057411194,
      "learning_rate": 1.0867556728767958e-05,
      "loss": 0.0513,
      "step": 26985
    },
    {
      "epoch": 1.3701203106756688,
      "grad_norm": 0.623604416847229,
      "learning_rate": 1.0865864595495541e-05,
      "loss": 0.0526,
      "step": 26990
    },
    {
      "epoch": 1.3703741306665314,
      "grad_norm": 0.4742110073566437,
      "learning_rate": 1.0864172462223127e-05,
      "loss": 0.051,
      "step": 26995
    },
    {
      "epoch": 1.3706279506573937,
      "grad_norm": 0.623345673084259,
      "learning_rate": 1.0862480328950708e-05,
      "loss": 0.0437,
      "step": 27000
    },
    {
      "epoch": 1.3708817706482563,
      "grad_norm": 0.3437178134918213,
      "learning_rate": 1.0860788195678294e-05,
      "loss": 0.0458,
      "step": 27005
    },
    {
      "epoch": 1.3711355906391187,
      "grad_norm": 0.288751482963562,
      "learning_rate": 1.0859096062405876e-05,
      "loss": 0.0515,
      "step": 27010
    },
    {
      "epoch": 1.3713894106299813,
      "grad_norm": 0.2961622476577759,
      "learning_rate": 1.0857403929133459e-05,
      "loss": 0.0535,
      "step": 27015
    },
    {
      "epoch": 1.3716432306208437,
      "grad_norm": 0.2858904302120209,
      "learning_rate": 1.0855711795861044e-05,
      "loss": 0.0408,
      "step": 27020
    },
    {
      "epoch": 1.371897050611706,
      "grad_norm": 0.25875383615493774,
      "learning_rate": 1.0854019662588626e-05,
      "loss": 0.0423,
      "step": 27025
    },
    {
      "epoch": 1.3721508706025687,
      "grad_norm": 0.9569272398948669,
      "learning_rate": 1.085232752931621e-05,
      "loss": 0.044,
      "step": 27030
    },
    {
      "epoch": 1.372404690593431,
      "grad_norm": 0.4269765615463257,
      "learning_rate": 1.0850635396043793e-05,
      "loss": 0.0434,
      "step": 27035
    },
    {
      "epoch": 1.3726585105842937,
      "grad_norm": 0.5560934543609619,
      "learning_rate": 1.0848943262771377e-05,
      "loss": 0.0572,
      "step": 27040
    },
    {
      "epoch": 1.372912330575156,
      "grad_norm": 0.42453646659851074,
      "learning_rate": 1.0847251129498962e-05,
      "loss": 0.0521,
      "step": 27045
    },
    {
      "epoch": 1.3731661505660187,
      "grad_norm": 0.36190876364707947,
      "learning_rate": 1.0845558996226544e-05,
      "loss": 0.0531,
      "step": 27050
    },
    {
      "epoch": 1.373419970556881,
      "grad_norm": 0.6442621946334839,
      "learning_rate": 1.0843866862954126e-05,
      "loss": 0.0482,
      "step": 27055
    },
    {
      "epoch": 1.3736737905477434,
      "grad_norm": 0.34046563506126404,
      "learning_rate": 1.0842174729681711e-05,
      "loss": 0.0421,
      "step": 27060
    },
    {
      "epoch": 1.373927610538606,
      "grad_norm": 0.30943602323532104,
      "learning_rate": 1.0840482596409295e-05,
      "loss": 0.0537,
      "step": 27065
    },
    {
      "epoch": 1.3741814305294686,
      "grad_norm": 0.4879555106163025,
      "learning_rate": 1.0838790463136876e-05,
      "loss": 0.0492,
      "step": 27070
    },
    {
      "epoch": 1.374435250520331,
      "grad_norm": 0.30128493905067444,
      "learning_rate": 1.0837098329864462e-05,
      "loss": 0.0496,
      "step": 27075
    },
    {
      "epoch": 1.3746890705111934,
      "grad_norm": 0.37397828698158264,
      "learning_rate": 1.0835406196592044e-05,
      "loss": 0.0532,
      "step": 27080
    },
    {
      "epoch": 1.374942890502056,
      "grad_norm": 0.2845868468284607,
      "learning_rate": 1.0833714063319627e-05,
      "loss": 0.045,
      "step": 27085
    },
    {
      "epoch": 1.3751967104929184,
      "grad_norm": 0.42873308062553406,
      "learning_rate": 1.0832021930047212e-05,
      "loss": 0.0649,
      "step": 27090
    },
    {
      "epoch": 1.375450530483781,
      "grad_norm": 0.4360988736152649,
      "learning_rate": 1.0830329796774794e-05,
      "loss": 0.0467,
      "step": 27095
    },
    {
      "epoch": 1.3757043504746433,
      "grad_norm": 0.33429446816444397,
      "learning_rate": 1.082863766350238e-05,
      "loss": 0.0443,
      "step": 27100
    },
    {
      "epoch": 1.375958170465506,
      "grad_norm": 0.41631633043289185,
      "learning_rate": 1.0826945530229961e-05,
      "loss": 0.0546,
      "step": 27105
    },
    {
      "epoch": 1.3762119904563683,
      "grad_norm": 0.3739117980003357,
      "learning_rate": 1.0825253396957545e-05,
      "loss": 0.0518,
      "step": 27110
    },
    {
      "epoch": 1.3764658104472307,
      "grad_norm": 0.25905776023864746,
      "learning_rate": 1.082356126368513e-05,
      "loss": 0.0444,
      "step": 27115
    },
    {
      "epoch": 1.3767196304380933,
      "grad_norm": 0.3633892238140106,
      "learning_rate": 1.0821869130412712e-05,
      "loss": 0.0469,
      "step": 27120
    },
    {
      "epoch": 1.376973450428956,
      "grad_norm": 0.31577759981155396,
      "learning_rate": 1.0820176997140295e-05,
      "loss": 0.0525,
      "step": 27125
    },
    {
      "epoch": 1.3772272704198183,
      "grad_norm": 0.5098549723625183,
      "learning_rate": 1.0818484863867879e-05,
      "loss": 0.0447,
      "step": 27130
    },
    {
      "epoch": 1.3774810904106807,
      "grad_norm": 0.2873249650001526,
      "learning_rate": 1.0816792730595462e-05,
      "loss": 0.0482,
      "step": 27135
    },
    {
      "epoch": 1.3777349104015433,
      "grad_norm": 0.4410748779773712,
      "learning_rate": 1.0815100597323044e-05,
      "loss": 0.0538,
      "step": 27140
    },
    {
      "epoch": 1.3779887303924057,
      "grad_norm": 0.4886588454246521,
      "learning_rate": 1.081340846405063e-05,
      "loss": 0.0503,
      "step": 27145
    },
    {
      "epoch": 1.3782425503832683,
      "grad_norm": 0.376172810792923,
      "learning_rate": 1.0811716330778213e-05,
      "loss": 0.0482,
      "step": 27150
    },
    {
      "epoch": 1.3784963703741306,
      "grad_norm": 0.3205304443836212,
      "learning_rate": 1.0810024197505797e-05,
      "loss": 0.0458,
      "step": 27155
    },
    {
      "epoch": 1.3787501903649932,
      "grad_norm": 0.3526761829853058,
      "learning_rate": 1.080833206423338e-05,
      "loss": 0.0482,
      "step": 27160
    },
    {
      "epoch": 1.3790040103558556,
      "grad_norm": 0.3599817454814911,
      "learning_rate": 1.0806639930960962e-05,
      "loss": 0.0511,
      "step": 27165
    },
    {
      "epoch": 1.379257830346718,
      "grad_norm": 0.29613739252090454,
      "learning_rate": 1.0804947797688547e-05,
      "loss": 0.0451,
      "step": 27170
    },
    {
      "epoch": 1.3795116503375806,
      "grad_norm": 0.3637460470199585,
      "learning_rate": 1.0803255664416131e-05,
      "loss": 0.0536,
      "step": 27175
    },
    {
      "epoch": 1.379765470328443,
      "grad_norm": 0.3151790499687195,
      "learning_rate": 1.0801563531143713e-05,
      "loss": 0.0455,
      "step": 27180
    },
    {
      "epoch": 1.3800192903193056,
      "grad_norm": 0.643429160118103,
      "learning_rate": 1.0799871397871298e-05,
      "loss": 0.0518,
      "step": 27185
    },
    {
      "epoch": 1.380273110310168,
      "grad_norm": 0.26551660895347595,
      "learning_rate": 1.079817926459888e-05,
      "loss": 0.0418,
      "step": 27190
    },
    {
      "epoch": 1.3805269303010306,
      "grad_norm": 0.30446264147758484,
      "learning_rate": 1.0796487131326465e-05,
      "loss": 0.0431,
      "step": 27195
    },
    {
      "epoch": 1.380780750291893,
      "grad_norm": 0.5254416465759277,
      "learning_rate": 1.0794794998054049e-05,
      "loss": 0.0519,
      "step": 27200
    },
    {
      "epoch": 1.3810345702827553,
      "grad_norm": 0.33535945415496826,
      "learning_rate": 1.079310286478163e-05,
      "loss": 0.0352,
      "step": 27205
    },
    {
      "epoch": 1.381288390273618,
      "grad_norm": 0.5450144410133362,
      "learning_rate": 1.0791410731509216e-05,
      "loss": 0.0412,
      "step": 27210
    },
    {
      "epoch": 1.3815422102644805,
      "grad_norm": 0.40926575660705566,
      "learning_rate": 1.0789718598236798e-05,
      "loss": 0.0526,
      "step": 27215
    },
    {
      "epoch": 1.381796030255343,
      "grad_norm": 0.3639059066772461,
      "learning_rate": 1.0788026464964381e-05,
      "loss": 0.0552,
      "step": 27220
    },
    {
      "epoch": 1.3820498502462053,
      "grad_norm": 0.5474151968955994,
      "learning_rate": 1.0786334331691966e-05,
      "loss": 0.0485,
      "step": 27225
    },
    {
      "epoch": 1.382303670237068,
      "grad_norm": 0.3799017071723938,
      "learning_rate": 1.0784642198419548e-05,
      "loss": 0.0591,
      "step": 27230
    },
    {
      "epoch": 1.3825574902279303,
      "grad_norm": 0.45561301708221436,
      "learning_rate": 1.078295006514713e-05,
      "loss": 0.0433,
      "step": 27235
    },
    {
      "epoch": 1.3828113102187929,
      "grad_norm": 0.4006117582321167,
      "learning_rate": 1.0781257931874715e-05,
      "loss": 0.051,
      "step": 27240
    },
    {
      "epoch": 1.3830651302096553,
      "grad_norm": 0.477120965719223,
      "learning_rate": 1.0779565798602299e-05,
      "loss": 0.0479,
      "step": 27245
    },
    {
      "epoch": 1.3833189502005179,
      "grad_norm": 0.29165032505989075,
      "learning_rate": 1.0777873665329884e-05,
      "loss": 0.0536,
      "step": 27250
    },
    {
      "epoch": 1.3835727701913803,
      "grad_norm": 0.3408038914203644,
      "learning_rate": 1.0776181532057466e-05,
      "loss": 0.0502,
      "step": 27255
    },
    {
      "epoch": 1.3838265901822426,
      "grad_norm": 1.091934323310852,
      "learning_rate": 1.0774489398785048e-05,
      "loss": 0.048,
      "step": 27260
    },
    {
      "epoch": 1.3840804101731052,
      "grad_norm": 0.330030620098114,
      "learning_rate": 1.0772797265512633e-05,
      "loss": 0.0488,
      "step": 27265
    },
    {
      "epoch": 1.3843342301639678,
      "grad_norm": 0.24753251671791077,
      "learning_rate": 1.0771105132240217e-05,
      "loss": 0.0491,
      "step": 27270
    },
    {
      "epoch": 1.3845880501548302,
      "grad_norm": 0.31028640270233154,
      "learning_rate": 1.0769412998967798e-05,
      "loss": 0.0481,
      "step": 27275
    },
    {
      "epoch": 1.3848418701456926,
      "grad_norm": 0.295444130897522,
      "learning_rate": 1.0767720865695384e-05,
      "loss": 0.0543,
      "step": 27280
    },
    {
      "epoch": 1.3850956901365552,
      "grad_norm": 0.7282490730285645,
      "learning_rate": 1.0766028732422965e-05,
      "loss": 0.055,
      "step": 27285
    },
    {
      "epoch": 1.3853495101274176,
      "grad_norm": 0.41168054938316345,
      "learning_rate": 1.076433659915055e-05,
      "loss": 0.0567,
      "step": 27290
    },
    {
      "epoch": 1.3856033301182802,
      "grad_norm": 0.447490930557251,
      "learning_rate": 1.0762644465878134e-05,
      "loss": 0.0501,
      "step": 27295
    },
    {
      "epoch": 1.3858571501091426,
      "grad_norm": 0.3839946687221527,
      "learning_rate": 1.0760952332605716e-05,
      "loss": 0.0479,
      "step": 27300
    },
    {
      "epoch": 1.3861109701000052,
      "grad_norm": 0.4760317802429199,
      "learning_rate": 1.0759260199333301e-05,
      "loss": 0.0464,
      "step": 27305
    },
    {
      "epoch": 1.3863647900908675,
      "grad_norm": 0.32983019948005676,
      "learning_rate": 1.0757568066060883e-05,
      "loss": 0.0453,
      "step": 27310
    },
    {
      "epoch": 1.38661861008173,
      "grad_norm": 0.7022455930709839,
      "learning_rate": 1.0755875932788467e-05,
      "loss": 0.0533,
      "step": 27315
    },
    {
      "epoch": 1.3868724300725925,
      "grad_norm": 0.44730642437934875,
      "learning_rate": 1.0754183799516052e-05,
      "loss": 0.0411,
      "step": 27320
    },
    {
      "epoch": 1.3871262500634551,
      "grad_norm": 0.25385260581970215,
      "learning_rate": 1.0752491666243634e-05,
      "loss": 0.0486,
      "step": 27325
    },
    {
      "epoch": 1.3873800700543175,
      "grad_norm": 0.29286813735961914,
      "learning_rate": 1.0750799532971217e-05,
      "loss": 0.0485,
      "step": 27330
    },
    {
      "epoch": 1.38763389004518,
      "grad_norm": 0.43927282094955444,
      "learning_rate": 1.0749107399698801e-05,
      "loss": 0.0425,
      "step": 27335
    },
    {
      "epoch": 1.3878877100360425,
      "grad_norm": 0.3693688213825226,
      "learning_rate": 1.0747415266426384e-05,
      "loss": 0.0508,
      "step": 27340
    },
    {
      "epoch": 1.3881415300269049,
      "grad_norm": 0.46396780014038086,
      "learning_rate": 1.074572313315397e-05,
      "loss": 0.0429,
      "step": 27345
    },
    {
      "epoch": 1.3883953500177675,
      "grad_norm": 0.5464805960655212,
      "learning_rate": 1.0744030999881552e-05,
      "loss": 0.0536,
      "step": 27350
    },
    {
      "epoch": 1.3886491700086299,
      "grad_norm": 0.26196610927581787,
      "learning_rate": 1.0742338866609135e-05,
      "loss": 0.0437,
      "step": 27355
    },
    {
      "epoch": 1.3889029899994925,
      "grad_norm": 0.355894535779953,
      "learning_rate": 1.0740646733336719e-05,
      "loss": 0.0447,
      "step": 27360
    },
    {
      "epoch": 1.3891568099903548,
      "grad_norm": 0.2798496186733246,
      "learning_rate": 1.0738954600064302e-05,
      "loss": 0.0414,
      "step": 27365
    },
    {
      "epoch": 1.3894106299812172,
      "grad_norm": 0.36841532588005066,
      "learning_rate": 1.0737262466791884e-05,
      "loss": 0.0469,
      "step": 27370
    },
    {
      "epoch": 1.3896644499720798,
      "grad_norm": 0.36413395404815674,
      "learning_rate": 1.073557033351947e-05,
      "loss": 0.0354,
      "step": 27375
    },
    {
      "epoch": 1.3899182699629422,
      "grad_norm": 0.3879263401031494,
      "learning_rate": 1.0733878200247053e-05,
      "loss": 0.0405,
      "step": 27380
    },
    {
      "epoch": 1.3901720899538048,
      "grad_norm": 0.4517977237701416,
      "learning_rate": 1.0732186066974635e-05,
      "loss": 0.0468,
      "step": 27385
    },
    {
      "epoch": 1.3904259099446672,
      "grad_norm": 0.4055328369140625,
      "learning_rate": 1.073049393370222e-05,
      "loss": 0.0457,
      "step": 27390
    },
    {
      "epoch": 1.3906797299355298,
      "grad_norm": 0.35788199305534363,
      "learning_rate": 1.0728801800429802e-05,
      "loss": 0.0427,
      "step": 27395
    },
    {
      "epoch": 1.3909335499263922,
      "grad_norm": 0.421030730009079,
      "learning_rate": 1.0727109667157387e-05,
      "loss": 0.0542,
      "step": 27400
    },
    {
      "epoch": 1.3911873699172546,
      "grad_norm": 0.28301897644996643,
      "learning_rate": 1.072541753388497e-05,
      "loss": 0.0486,
      "step": 27405
    },
    {
      "epoch": 1.3914411899081172,
      "grad_norm": 0.2675779461860657,
      "learning_rate": 1.0723725400612552e-05,
      "loss": 0.0502,
      "step": 27410
    },
    {
      "epoch": 1.3916950098989798,
      "grad_norm": 0.40590351819992065,
      "learning_rate": 1.0722033267340138e-05,
      "loss": 0.0493,
      "step": 27415
    },
    {
      "epoch": 1.3919488298898421,
      "grad_norm": 0.5221124887466431,
      "learning_rate": 1.072034113406772e-05,
      "loss": 0.0591,
      "step": 27420
    },
    {
      "epoch": 1.3922026498807045,
      "grad_norm": 0.6834927797317505,
      "learning_rate": 1.0718649000795303e-05,
      "loss": 0.0441,
      "step": 27425
    },
    {
      "epoch": 1.3924564698715671,
      "grad_norm": 0.3241395652294159,
      "learning_rate": 1.0716956867522888e-05,
      "loss": 0.0395,
      "step": 27430
    },
    {
      "epoch": 1.3927102898624295,
      "grad_norm": 0.3666060268878937,
      "learning_rate": 1.071526473425047e-05,
      "loss": 0.05,
      "step": 27435
    },
    {
      "epoch": 1.392964109853292,
      "grad_norm": 0.33367741107940674,
      "learning_rate": 1.0713572600978055e-05,
      "loss": 0.0406,
      "step": 27440
    },
    {
      "epoch": 1.3932179298441545,
      "grad_norm": 0.3969685137271881,
      "learning_rate": 1.0711880467705637e-05,
      "loss": 0.0427,
      "step": 27445
    },
    {
      "epoch": 1.393471749835017,
      "grad_norm": 0.3486899137496948,
      "learning_rate": 1.071018833443322e-05,
      "loss": 0.0489,
      "step": 27450
    },
    {
      "epoch": 1.3937255698258795,
      "grad_norm": 0.543165922164917,
      "learning_rate": 1.0708496201160806e-05,
      "loss": 0.0464,
      "step": 27455
    },
    {
      "epoch": 1.3939793898167419,
      "grad_norm": 0.35785973072052,
      "learning_rate": 1.0706804067888388e-05,
      "loss": 0.0502,
      "step": 27460
    },
    {
      "epoch": 1.3942332098076045,
      "grad_norm": 0.4813598692417145,
      "learning_rate": 1.070511193461597e-05,
      "loss": 0.0513,
      "step": 27465
    },
    {
      "epoch": 1.394487029798467,
      "grad_norm": 0.39403215050697327,
      "learning_rate": 1.0703419801343555e-05,
      "loss": 0.0488,
      "step": 27470
    },
    {
      "epoch": 1.3947408497893294,
      "grad_norm": 0.33684828877449036,
      "learning_rate": 1.0701727668071138e-05,
      "loss": 0.0431,
      "step": 27475
    },
    {
      "epoch": 1.3949946697801918,
      "grad_norm": 0.3423417806625366,
      "learning_rate": 1.070003553479872e-05,
      "loss": 0.0451,
      "step": 27480
    },
    {
      "epoch": 1.3952484897710544,
      "grad_norm": 0.3403189480304718,
      "learning_rate": 1.0698343401526306e-05,
      "loss": 0.0549,
      "step": 27485
    },
    {
      "epoch": 1.3955023097619168,
      "grad_norm": 0.4790232479572296,
      "learning_rate": 1.0696651268253887e-05,
      "loss": 0.0528,
      "step": 27490
    },
    {
      "epoch": 1.3957561297527794,
      "grad_norm": 0.4470588266849518,
      "learning_rate": 1.0694959134981473e-05,
      "loss": 0.0542,
      "step": 27495
    },
    {
      "epoch": 1.3960099497436418,
      "grad_norm": 0.30051055550575256,
      "learning_rate": 1.0693267001709056e-05,
      "loss": 0.0562,
      "step": 27500
    },
    {
      "epoch": 1.3962637697345044,
      "grad_norm": 0.4164995849132538,
      "learning_rate": 1.0691574868436638e-05,
      "loss": 0.0522,
      "step": 27505
    },
    {
      "epoch": 1.3965175897253668,
      "grad_norm": 0.4499371647834778,
      "learning_rate": 1.0689882735164223e-05,
      "loss": 0.0457,
      "step": 27510
    },
    {
      "epoch": 1.3967714097162292,
      "grad_norm": 0.36601021885871887,
      "learning_rate": 1.0688190601891805e-05,
      "loss": 0.048,
      "step": 27515
    },
    {
      "epoch": 1.3970252297070918,
      "grad_norm": 0.5158339738845825,
      "learning_rate": 1.0686498468619389e-05,
      "loss": 0.0538,
      "step": 27520
    },
    {
      "epoch": 1.3972790496979541,
      "grad_norm": 0.4598226845264435,
      "learning_rate": 1.0684806335346974e-05,
      "loss": 0.0517,
      "step": 27525
    },
    {
      "epoch": 1.3975328696888167,
      "grad_norm": 0.26244619488716125,
      "learning_rate": 1.0683114202074556e-05,
      "loss": 0.0486,
      "step": 27530
    },
    {
      "epoch": 1.3977866896796791,
      "grad_norm": 0.46184155344963074,
      "learning_rate": 1.0681422068802141e-05,
      "loss": 0.0475,
      "step": 27535
    },
    {
      "epoch": 1.3980405096705417,
      "grad_norm": 0.38430994749069214,
      "learning_rate": 1.0679729935529723e-05,
      "loss": 0.0461,
      "step": 27540
    },
    {
      "epoch": 1.398294329661404,
      "grad_norm": 0.4069960117340088,
      "learning_rate": 1.0678037802257306e-05,
      "loss": 0.0479,
      "step": 27545
    },
    {
      "epoch": 1.3985481496522665,
      "grad_norm": 0.3001312017440796,
      "learning_rate": 1.0676345668984892e-05,
      "loss": 0.0547,
      "step": 27550
    },
    {
      "epoch": 1.398801969643129,
      "grad_norm": 0.42500364780426025,
      "learning_rate": 1.0674653535712473e-05,
      "loss": 0.0402,
      "step": 27555
    },
    {
      "epoch": 1.3990557896339917,
      "grad_norm": 0.26252198219299316,
      "learning_rate": 1.0672961402440057e-05,
      "loss": 0.0419,
      "step": 27560
    },
    {
      "epoch": 1.399309609624854,
      "grad_norm": 0.363121896982193,
      "learning_rate": 1.067126926916764e-05,
      "loss": 0.0542,
      "step": 27565
    },
    {
      "epoch": 1.3995634296157164,
      "grad_norm": 0.3183755576610565,
      "learning_rate": 1.0669577135895224e-05,
      "loss": 0.0501,
      "step": 27570
    },
    {
      "epoch": 1.399817249606579,
      "grad_norm": 0.3802196979522705,
      "learning_rate": 1.0667885002622806e-05,
      "loss": 0.0548,
      "step": 27575
    },
    {
      "epoch": 1.4000710695974414,
      "grad_norm": 0.8211646676063538,
      "learning_rate": 1.0666192869350391e-05,
      "loss": 0.0462,
      "step": 27580
    },
    {
      "epoch": 1.400324889588304,
      "grad_norm": 0.33730578422546387,
      "learning_rate": 1.0664500736077975e-05,
      "loss": 0.0501,
      "step": 27585
    },
    {
      "epoch": 1.4005787095791664,
      "grad_norm": 0.2553010582923889,
      "learning_rate": 1.0662808602805558e-05,
      "loss": 0.0426,
      "step": 27590
    },
    {
      "epoch": 1.400832529570029,
      "grad_norm": 0.3305681049823761,
      "learning_rate": 1.0661116469533142e-05,
      "loss": 0.0483,
      "step": 27595
    },
    {
      "epoch": 1.4010863495608914,
      "grad_norm": 0.34884992241859436,
      "learning_rate": 1.0659424336260724e-05,
      "loss": 0.0383,
      "step": 27600
    },
    {
      "epoch": 1.4013401695517538,
      "grad_norm": 0.5144136548042297,
      "learning_rate": 1.0657732202988309e-05,
      "loss": 0.0438,
      "step": 27605
    },
    {
      "epoch": 1.4015939895426164,
      "grad_norm": 0.3989544212818146,
      "learning_rate": 1.0656040069715892e-05,
      "loss": 0.0427,
      "step": 27610
    },
    {
      "epoch": 1.401847809533479,
      "grad_norm": 0.4733310043811798,
      "learning_rate": 1.0654347936443474e-05,
      "loss": 0.053,
      "step": 27615
    },
    {
      "epoch": 1.4021016295243414,
      "grad_norm": 0.7690626382827759,
      "learning_rate": 1.065265580317106e-05,
      "loss": 0.0507,
      "step": 27620
    },
    {
      "epoch": 1.4023554495152037,
      "grad_norm": 0.4382520318031311,
      "learning_rate": 1.0650963669898641e-05,
      "loss": 0.0483,
      "step": 27625
    },
    {
      "epoch": 1.4026092695060663,
      "grad_norm": 0.40650293231010437,
      "learning_rate": 1.0649271536626225e-05,
      "loss": 0.0451,
      "step": 27630
    },
    {
      "epoch": 1.4028630894969287,
      "grad_norm": 0.39848485589027405,
      "learning_rate": 1.064757940335381e-05,
      "loss": 0.0529,
      "step": 27635
    },
    {
      "epoch": 1.4031169094877913,
      "grad_norm": 0.24001622200012207,
      "learning_rate": 1.0645887270081392e-05,
      "loss": 0.0403,
      "step": 27640
    },
    {
      "epoch": 1.4033707294786537,
      "grad_norm": 0.3177374303340912,
      "learning_rate": 1.0644195136808977e-05,
      "loss": 0.0462,
      "step": 27645
    },
    {
      "epoch": 1.4036245494695163,
      "grad_norm": 0.23650433123111725,
      "learning_rate": 1.0642503003536559e-05,
      "loss": 0.0465,
      "step": 27650
    },
    {
      "epoch": 1.4038783694603787,
      "grad_norm": 0.41654306650161743,
      "learning_rate": 1.0640810870264143e-05,
      "loss": 0.0499,
      "step": 27655
    },
    {
      "epoch": 1.404132189451241,
      "grad_norm": 0.2645888328552246,
      "learning_rate": 1.0639118736991726e-05,
      "loss": 0.0526,
      "step": 27660
    },
    {
      "epoch": 1.4043860094421037,
      "grad_norm": 0.2605307102203369,
      "learning_rate": 1.063742660371931e-05,
      "loss": 0.0386,
      "step": 27665
    },
    {
      "epoch": 1.404639829432966,
      "grad_norm": 0.3085188567638397,
      "learning_rate": 1.0635734470446892e-05,
      "loss": 0.0512,
      "step": 27670
    },
    {
      "epoch": 1.4048936494238287,
      "grad_norm": 0.3979353606700897,
      "learning_rate": 1.0634042337174477e-05,
      "loss": 0.05,
      "step": 27675
    },
    {
      "epoch": 1.405147469414691,
      "grad_norm": 0.8337517380714417,
      "learning_rate": 1.063235020390206e-05,
      "loss": 0.0451,
      "step": 27680
    },
    {
      "epoch": 1.4054012894055536,
      "grad_norm": 0.45434877276420593,
      "learning_rate": 1.0630658070629644e-05,
      "loss": 0.0504,
      "step": 27685
    },
    {
      "epoch": 1.405655109396416,
      "grad_norm": 0.45057016611099243,
      "learning_rate": 1.0628965937357227e-05,
      "loss": 0.0448,
      "step": 27690
    },
    {
      "epoch": 1.4059089293872784,
      "grad_norm": 0.40183669328689575,
      "learning_rate": 1.062727380408481e-05,
      "loss": 0.0379,
      "step": 27695
    },
    {
      "epoch": 1.406162749378141,
      "grad_norm": 0.36201322078704834,
      "learning_rate": 1.0625581670812395e-05,
      "loss": 0.0448,
      "step": 27700
    },
    {
      "epoch": 1.4064165693690036,
      "grad_norm": 0.3698229491710663,
      "learning_rate": 1.0623889537539978e-05,
      "loss": 0.0401,
      "step": 27705
    },
    {
      "epoch": 1.406670389359866,
      "grad_norm": 0.3653453588485718,
      "learning_rate": 1.062219740426756e-05,
      "loss": 0.0444,
      "step": 27710
    },
    {
      "epoch": 1.4069242093507284,
      "grad_norm": 0.5779272317886353,
      "learning_rate": 1.0620505270995145e-05,
      "loss": 0.0478,
      "step": 27715
    },
    {
      "epoch": 1.407178029341591,
      "grad_norm": 0.29561781883239746,
      "learning_rate": 1.0618813137722727e-05,
      "loss": 0.0501,
      "step": 27720
    },
    {
      "epoch": 1.4074318493324534,
      "grad_norm": 0.3608880937099457,
      "learning_rate": 1.061712100445031e-05,
      "loss": 0.0487,
      "step": 27725
    },
    {
      "epoch": 1.407685669323316,
      "grad_norm": 0.2934507131576538,
      "learning_rate": 1.0615428871177896e-05,
      "loss": 0.0464,
      "step": 27730
    },
    {
      "epoch": 1.4079394893141783,
      "grad_norm": 0.28474876284599304,
      "learning_rate": 1.0613736737905478e-05,
      "loss": 0.0399,
      "step": 27735
    },
    {
      "epoch": 1.408193309305041,
      "grad_norm": 0.42325231432914734,
      "learning_rate": 1.0612044604633063e-05,
      "loss": 0.0467,
      "step": 27740
    },
    {
      "epoch": 1.4084471292959033,
      "grad_norm": 0.3682181239128113,
      "learning_rate": 1.0610352471360645e-05,
      "loss": 0.0432,
      "step": 27745
    },
    {
      "epoch": 1.4087009492867657,
      "grad_norm": 0.41442859172821045,
      "learning_rate": 1.0608660338088228e-05,
      "loss": 0.0516,
      "step": 27750
    },
    {
      "epoch": 1.4089547692776283,
      "grad_norm": 0.4755087196826935,
      "learning_rate": 1.0606968204815814e-05,
      "loss": 0.0555,
      "step": 27755
    },
    {
      "epoch": 1.409208589268491,
      "grad_norm": 0.44574877619743347,
      "learning_rate": 1.0605276071543395e-05,
      "loss": 0.0485,
      "step": 27760
    },
    {
      "epoch": 1.4094624092593533,
      "grad_norm": 0.39875462651252747,
      "learning_rate": 1.0603583938270979e-05,
      "loss": 0.0537,
      "step": 27765
    },
    {
      "epoch": 1.4097162292502157,
      "grad_norm": 0.2607244551181793,
      "learning_rate": 1.0601891804998562e-05,
      "loss": 0.0465,
      "step": 27770
    },
    {
      "epoch": 1.4099700492410783,
      "grad_norm": 0.36298197507858276,
      "learning_rate": 1.0600199671726146e-05,
      "loss": 0.05,
      "step": 27775
    },
    {
      "epoch": 1.4102238692319407,
      "grad_norm": 0.5285542011260986,
      "learning_rate": 1.0598507538453728e-05,
      "loss": 0.0523,
      "step": 27780
    },
    {
      "epoch": 1.4104776892228033,
      "grad_norm": 0.49905699491500854,
      "learning_rate": 1.0596815405181313e-05,
      "loss": 0.0451,
      "step": 27785
    },
    {
      "epoch": 1.4107315092136656,
      "grad_norm": 0.5594649314880371,
      "learning_rate": 1.0595123271908897e-05,
      "loss": 0.0506,
      "step": 27790
    },
    {
      "epoch": 1.4109853292045282,
      "grad_norm": 0.2815134525299072,
      "learning_rate": 1.059343113863648e-05,
      "loss": 0.048,
      "step": 27795
    },
    {
      "epoch": 1.4112391491953906,
      "grad_norm": 0.302711546421051,
      "learning_rate": 1.0591739005364064e-05,
      "loss": 0.0453,
      "step": 27800
    },
    {
      "epoch": 1.411492969186253,
      "grad_norm": 0.42682504653930664,
      "learning_rate": 1.0590046872091646e-05,
      "loss": 0.0523,
      "step": 27805
    },
    {
      "epoch": 1.4117467891771156,
      "grad_norm": 0.39569196105003357,
      "learning_rate": 1.058835473881923e-05,
      "loss": 0.0556,
      "step": 27810
    },
    {
      "epoch": 1.4120006091679782,
      "grad_norm": 0.5938218832015991,
      "learning_rate": 1.0586662605546814e-05,
      "loss": 0.0525,
      "step": 27815
    },
    {
      "epoch": 1.4122544291588406,
      "grad_norm": 0.3381215035915375,
      "learning_rate": 1.0584970472274396e-05,
      "loss": 0.052,
      "step": 27820
    },
    {
      "epoch": 1.412508249149703,
      "grad_norm": 0.3109475374221802,
      "learning_rate": 1.0583278339001981e-05,
      "loss": 0.0463,
      "step": 27825
    },
    {
      "epoch": 1.4127620691405656,
      "grad_norm": 0.4209303557872772,
      "learning_rate": 1.0581586205729563e-05,
      "loss": 0.0469,
      "step": 27830
    },
    {
      "epoch": 1.413015889131428,
      "grad_norm": 0.3299580216407776,
      "learning_rate": 1.0579894072457149e-05,
      "loss": 0.0453,
      "step": 27835
    },
    {
      "epoch": 1.4132697091222906,
      "grad_norm": 0.31590738892555237,
      "learning_rate": 1.057820193918473e-05,
      "loss": 0.0449,
      "step": 27840
    },
    {
      "epoch": 1.413523529113153,
      "grad_norm": 0.3760964274406433,
      "learning_rate": 1.0576509805912314e-05,
      "loss": 0.0495,
      "step": 27845
    },
    {
      "epoch": 1.4137773491040155,
      "grad_norm": 0.37932348251342773,
      "learning_rate": 1.05748176726399e-05,
      "loss": 0.0494,
      "step": 27850
    },
    {
      "epoch": 1.414031169094878,
      "grad_norm": 0.4534890055656433,
      "learning_rate": 1.0573125539367481e-05,
      "loss": 0.0508,
      "step": 27855
    },
    {
      "epoch": 1.4142849890857403,
      "grad_norm": 0.323983371257782,
      "learning_rate": 1.0571433406095065e-05,
      "loss": 0.055,
      "step": 27860
    },
    {
      "epoch": 1.414538809076603,
      "grad_norm": 0.34417563676834106,
      "learning_rate": 1.0569741272822648e-05,
      "loss": 0.0567,
      "step": 27865
    },
    {
      "epoch": 1.4147926290674653,
      "grad_norm": 0.36441168189048767,
      "learning_rate": 1.0568049139550232e-05,
      "loss": 0.043,
      "step": 27870
    },
    {
      "epoch": 1.4150464490583279,
      "grad_norm": 0.49125048518180847,
      "learning_rate": 1.0566357006277814e-05,
      "loss": 0.0435,
      "step": 27875
    },
    {
      "epoch": 1.4153002690491903,
      "grad_norm": 0.40159016847610474,
      "learning_rate": 1.0564664873005399e-05,
      "loss": 0.0494,
      "step": 27880
    },
    {
      "epoch": 1.4155540890400529,
      "grad_norm": 0.5556924343109131,
      "learning_rate": 1.0562972739732982e-05,
      "loss": 0.0569,
      "step": 27885
    },
    {
      "epoch": 1.4158079090309152,
      "grad_norm": 0.8219660520553589,
      "learning_rate": 1.0561280606460566e-05,
      "loss": 0.0609,
      "step": 27890
    },
    {
      "epoch": 1.4160617290217776,
      "grad_norm": 0.318620890378952,
      "learning_rate": 1.055958847318815e-05,
      "loss": 0.0472,
      "step": 27895
    },
    {
      "epoch": 1.4163155490126402,
      "grad_norm": 0.2669641971588135,
      "learning_rate": 1.0557896339915731e-05,
      "loss": 0.0513,
      "step": 27900
    },
    {
      "epoch": 1.4165693690035028,
      "grad_norm": 0.5739254355430603,
      "learning_rate": 1.0556204206643316e-05,
      "loss": 0.043,
      "step": 27905
    },
    {
      "epoch": 1.4168231889943652,
      "grad_norm": 0.6050904989242554,
      "learning_rate": 1.05545120733709e-05,
      "loss": 0.0492,
      "step": 27910
    },
    {
      "epoch": 1.4170770089852276,
      "grad_norm": 0.47114139795303345,
      "learning_rate": 1.0552819940098482e-05,
      "loss": 0.0483,
      "step": 27915
    },
    {
      "epoch": 1.4173308289760902,
      "grad_norm": 1.0412184000015259,
      "learning_rate": 1.0551127806826067e-05,
      "loss": 0.0563,
      "step": 27920
    },
    {
      "epoch": 1.4175846489669526,
      "grad_norm": 0.3775009512901306,
      "learning_rate": 1.0549435673553649e-05,
      "loss": 0.0506,
      "step": 27925
    },
    {
      "epoch": 1.4178384689578152,
      "grad_norm": 0.3336693346500397,
      "learning_rate": 1.0547743540281234e-05,
      "loss": 0.0525,
      "step": 27930
    },
    {
      "epoch": 1.4180922889486776,
      "grad_norm": 0.26565924286842346,
      "learning_rate": 1.0546051407008818e-05,
      "loss": 0.0468,
      "step": 27935
    },
    {
      "epoch": 1.4183461089395402,
      "grad_norm": 0.3058395981788635,
      "learning_rate": 1.05443592737364e-05,
      "loss": 0.0366,
      "step": 27940
    },
    {
      "epoch": 1.4185999289304025,
      "grad_norm": 0.3325693905353546,
      "learning_rate": 1.0542667140463985e-05,
      "loss": 0.0452,
      "step": 27945
    },
    {
      "epoch": 1.418853748921265,
      "grad_norm": 0.3314320743083954,
      "learning_rate": 1.0540975007191567e-05,
      "loss": 0.0494,
      "step": 27950
    },
    {
      "epoch": 1.4191075689121275,
      "grad_norm": 0.27716904878616333,
      "learning_rate": 1.053928287391915e-05,
      "loss": 0.0416,
      "step": 27955
    },
    {
      "epoch": 1.4193613889029901,
      "grad_norm": 0.3261719346046448,
      "learning_rate": 1.0537590740646735e-05,
      "loss": 0.0557,
      "step": 27960
    },
    {
      "epoch": 1.4196152088938525,
      "grad_norm": 0.4958856701850891,
      "learning_rate": 1.0535898607374317e-05,
      "loss": 0.047,
      "step": 27965
    },
    {
      "epoch": 1.419869028884715,
      "grad_norm": 0.2569836378097534,
      "learning_rate": 1.0534206474101901e-05,
      "loss": 0.0514,
      "step": 27970
    },
    {
      "epoch": 1.4201228488755775,
      "grad_norm": 0.3573864996433258,
      "learning_rate": 1.0532514340829484e-05,
      "loss": 0.0431,
      "step": 27975
    },
    {
      "epoch": 1.4203766688664399,
      "grad_norm": 1.4592692852020264,
      "learning_rate": 1.0530822207557068e-05,
      "loss": 0.0483,
      "step": 27980
    },
    {
      "epoch": 1.4206304888573025,
      "grad_norm": 0.7606174945831299,
      "learning_rate": 1.0529130074284653e-05,
      "loss": 0.0512,
      "step": 27985
    },
    {
      "epoch": 1.4208843088481649,
      "grad_norm": 0.49157556891441345,
      "learning_rate": 1.0527437941012235e-05,
      "loss": 0.0461,
      "step": 27990
    },
    {
      "epoch": 1.4211381288390275,
      "grad_norm": 1.5763497352600098,
      "learning_rate": 1.0525745807739819e-05,
      "loss": 0.056,
      "step": 27995
    },
    {
      "epoch": 1.4213919488298898,
      "grad_norm": 0.3884940445423126,
      "learning_rate": 1.0524053674467402e-05,
      "loss": 0.0425,
      "step": 28000
    },
    {
      "epoch": 1.4216457688207522,
      "grad_norm": 0.36167848110198975,
      "learning_rate": 1.0522361541194986e-05,
      "loss": 0.0472,
      "step": 28005
    },
    {
      "epoch": 1.4218995888116148,
      "grad_norm": 0.6486092209815979,
      "learning_rate": 1.0520669407922568e-05,
      "loss": 0.0487,
      "step": 28010
    },
    {
      "epoch": 1.4221534088024772,
      "grad_norm": 0.4026010036468506,
      "learning_rate": 1.0518977274650153e-05,
      "loss": 0.0511,
      "step": 28015
    },
    {
      "epoch": 1.4224072287933398,
      "grad_norm": 0.4928683042526245,
      "learning_rate": 1.0517285141377736e-05,
      "loss": 0.0583,
      "step": 28020
    },
    {
      "epoch": 1.4226610487842022,
      "grad_norm": 0.706096351146698,
      "learning_rate": 1.0515593008105318e-05,
      "loss": 0.0557,
      "step": 28025
    },
    {
      "epoch": 1.4229148687750648,
      "grad_norm": 0.33938637375831604,
      "learning_rate": 1.0513900874832903e-05,
      "loss": 0.0418,
      "step": 28030
    },
    {
      "epoch": 1.4231686887659272,
      "grad_norm": 0.3287295699119568,
      "learning_rate": 1.0512208741560485e-05,
      "loss": 0.0486,
      "step": 28035
    },
    {
      "epoch": 1.4234225087567896,
      "grad_norm": 0.3345835506916046,
      "learning_rate": 1.051051660828807e-05,
      "loss": 0.054,
      "step": 28040
    },
    {
      "epoch": 1.4236763287476522,
      "grad_norm": 0.29467958211898804,
      "learning_rate": 1.0508824475015652e-05,
      "loss": 0.0495,
      "step": 28045
    },
    {
      "epoch": 1.4239301487385148,
      "grad_norm": 0.37132373452186584,
      "learning_rate": 1.0507132341743236e-05,
      "loss": 0.051,
      "step": 28050
    },
    {
      "epoch": 1.4241839687293771,
      "grad_norm": 0.385649710893631,
      "learning_rate": 1.0505440208470821e-05,
      "loss": 0.0441,
      "step": 28055
    },
    {
      "epoch": 1.4244377887202395,
      "grad_norm": 0.43290475010871887,
      "learning_rate": 1.0503748075198403e-05,
      "loss": 0.0464,
      "step": 28060
    },
    {
      "epoch": 1.4246916087111021,
      "grad_norm": 0.37711504101753235,
      "learning_rate": 1.0502055941925987e-05,
      "loss": 0.0407,
      "step": 28065
    },
    {
      "epoch": 1.4249454287019645,
      "grad_norm": 0.29967761039733887,
      "learning_rate": 1.050036380865357e-05,
      "loss": 0.0472,
      "step": 28070
    },
    {
      "epoch": 1.425199248692827,
      "grad_norm": 0.3553757667541504,
      "learning_rate": 1.0498671675381154e-05,
      "loss": 0.0432,
      "step": 28075
    },
    {
      "epoch": 1.4254530686836895,
      "grad_norm": 0.5949152708053589,
      "learning_rate": 1.0496979542108739e-05,
      "loss": 0.0441,
      "step": 28080
    },
    {
      "epoch": 1.425706888674552,
      "grad_norm": 0.45510491728782654,
      "learning_rate": 1.049528740883632e-05,
      "loss": 0.0473,
      "step": 28085
    },
    {
      "epoch": 1.4259607086654145,
      "grad_norm": 1.5510005950927734,
      "learning_rate": 1.0493595275563904e-05,
      "loss": 0.0497,
      "step": 28090
    },
    {
      "epoch": 1.4262145286562768,
      "grad_norm": 0.4228087365627289,
      "learning_rate": 1.0491903142291488e-05,
      "loss": 0.0436,
      "step": 28095
    },
    {
      "epoch": 1.4264683486471395,
      "grad_norm": 0.45731356739997864,
      "learning_rate": 1.0490211009019071e-05,
      "loss": 0.0584,
      "step": 28100
    },
    {
      "epoch": 1.426722168638002,
      "grad_norm": 0.3750070631504059,
      "learning_rate": 1.0488518875746653e-05,
      "loss": 0.0477,
      "step": 28105
    },
    {
      "epoch": 1.4269759886288644,
      "grad_norm": 0.530326783657074,
      "learning_rate": 1.0486826742474238e-05,
      "loss": 0.0491,
      "step": 28110
    },
    {
      "epoch": 1.4272298086197268,
      "grad_norm": 0.45974087715148926,
      "learning_rate": 1.0485134609201822e-05,
      "loss": 0.0503,
      "step": 28115
    },
    {
      "epoch": 1.4274836286105894,
      "grad_norm": 0.3488108217716217,
      "learning_rate": 1.0483442475929404e-05,
      "loss": 0.0493,
      "step": 28120
    },
    {
      "epoch": 1.4277374486014518,
      "grad_norm": 0.3617185950279236,
      "learning_rate": 1.0481750342656989e-05,
      "loss": 0.0463,
      "step": 28125
    },
    {
      "epoch": 1.4279912685923144,
      "grad_norm": 0.4106167256832123,
      "learning_rate": 1.0480058209384571e-05,
      "loss": 0.0542,
      "step": 28130
    },
    {
      "epoch": 1.4282450885831768,
      "grad_norm": 0.3748033940792084,
      "learning_rate": 1.0478366076112156e-05,
      "loss": 0.0385,
      "step": 28135
    },
    {
      "epoch": 1.4284989085740394,
      "grad_norm": 0.3587561547756195,
      "learning_rate": 1.047667394283974e-05,
      "loss": 0.0561,
      "step": 28140
    },
    {
      "epoch": 1.4287527285649018,
      "grad_norm": 0.3485838770866394,
      "learning_rate": 1.0474981809567322e-05,
      "loss": 0.0468,
      "step": 28145
    },
    {
      "epoch": 1.4290065485557641,
      "grad_norm": 0.6947336196899414,
      "learning_rate": 1.0473289676294907e-05,
      "loss": 0.0433,
      "step": 28150
    },
    {
      "epoch": 1.4292603685466267,
      "grad_norm": 0.3088648319244385,
      "learning_rate": 1.0471597543022489e-05,
      "loss": 0.0476,
      "step": 28155
    },
    {
      "epoch": 1.4295141885374893,
      "grad_norm": 0.38387876749038696,
      "learning_rate": 1.0469905409750072e-05,
      "loss": 0.0406,
      "step": 28160
    },
    {
      "epoch": 1.4297680085283517,
      "grad_norm": 0.3302471339702606,
      "learning_rate": 1.0468213276477657e-05,
      "loss": 0.0406,
      "step": 28165
    },
    {
      "epoch": 1.430021828519214,
      "grad_norm": 0.9328307509422302,
      "learning_rate": 1.046652114320524e-05,
      "loss": 0.0459,
      "step": 28170
    },
    {
      "epoch": 1.4302756485100767,
      "grad_norm": 0.3534489870071411,
      "learning_rate": 1.0464829009932825e-05,
      "loss": 0.0384,
      "step": 28175
    },
    {
      "epoch": 1.430529468500939,
      "grad_norm": 0.2702120542526245,
      "learning_rate": 1.0463136876660406e-05,
      "loss": 0.0431,
      "step": 28180
    },
    {
      "epoch": 1.4307832884918017,
      "grad_norm": 0.9567867517471313,
      "learning_rate": 1.046144474338799e-05,
      "loss": 0.0561,
      "step": 28185
    },
    {
      "epoch": 1.431037108482664,
      "grad_norm": 0.35010412335395813,
      "learning_rate": 1.0459752610115575e-05,
      "loss": 0.0548,
      "step": 28190
    },
    {
      "epoch": 1.4312909284735267,
      "grad_norm": 0.6569327712059021,
      "learning_rate": 1.0458060476843157e-05,
      "loss": 0.0461,
      "step": 28195
    },
    {
      "epoch": 1.431544748464389,
      "grad_norm": 0.305759996175766,
      "learning_rate": 1.045636834357074e-05,
      "loss": 0.0475,
      "step": 28200
    },
    {
      "epoch": 1.4317985684552514,
      "grad_norm": 0.36200013756752014,
      "learning_rate": 1.0454676210298324e-05,
      "loss": 0.0469,
      "step": 28205
    },
    {
      "epoch": 1.432052388446114,
      "grad_norm": 0.3904263377189636,
      "learning_rate": 1.0452984077025908e-05,
      "loss": 0.0448,
      "step": 28210
    },
    {
      "epoch": 1.4323062084369764,
      "grad_norm": 0.7225484251976013,
      "learning_rate": 1.045129194375349e-05,
      "loss": 0.0452,
      "step": 28215
    },
    {
      "epoch": 1.432560028427839,
      "grad_norm": 0.4833544194698334,
      "learning_rate": 1.0449599810481075e-05,
      "loss": 0.0476,
      "step": 28220
    },
    {
      "epoch": 1.4328138484187014,
      "grad_norm": 0.39925500750541687,
      "learning_rate": 1.0447907677208657e-05,
      "loss": 0.0342,
      "step": 28225
    },
    {
      "epoch": 1.433067668409564,
      "grad_norm": 0.26874077320098877,
      "learning_rate": 1.0446215543936242e-05,
      "loss": 0.0444,
      "step": 28230
    },
    {
      "epoch": 1.4333214884004264,
      "grad_norm": 0.3140677809715271,
      "learning_rate": 1.0444523410663825e-05,
      "loss": 0.037,
      "step": 28235
    },
    {
      "epoch": 1.4335753083912888,
      "grad_norm": 0.3089941740036011,
      "learning_rate": 1.0442831277391407e-05,
      "loss": 0.0428,
      "step": 28240
    },
    {
      "epoch": 1.4338291283821514,
      "grad_norm": 0.7370154857635498,
      "learning_rate": 1.0441139144118992e-05,
      "loss": 0.0417,
      "step": 28245
    },
    {
      "epoch": 1.434082948373014,
      "grad_norm": 0.2998804748058319,
      "learning_rate": 1.0439447010846574e-05,
      "loss": 0.0415,
      "step": 28250
    },
    {
      "epoch": 1.4343367683638764,
      "grad_norm": 0.2329416424036026,
      "learning_rate": 1.0437754877574158e-05,
      "loss": 0.0479,
      "step": 28255
    },
    {
      "epoch": 1.4345905883547387,
      "grad_norm": 0.3600409924983978,
      "learning_rate": 1.0436062744301743e-05,
      "loss": 0.052,
      "step": 28260
    },
    {
      "epoch": 1.4348444083456013,
      "grad_norm": 0.3627881705760956,
      "learning_rate": 1.0434370611029325e-05,
      "loss": 0.0457,
      "step": 28265
    },
    {
      "epoch": 1.4350982283364637,
      "grad_norm": 0.3902629017829895,
      "learning_rate": 1.0432678477756908e-05,
      "loss": 0.0484,
      "step": 28270
    },
    {
      "epoch": 1.4353520483273263,
      "grad_norm": 0.42769598960876465,
      "learning_rate": 1.0430986344484492e-05,
      "loss": 0.0504,
      "step": 28275
    },
    {
      "epoch": 1.4356058683181887,
      "grad_norm": 0.3344421982765198,
      "learning_rate": 1.0429294211212076e-05,
      "loss": 0.047,
      "step": 28280
    },
    {
      "epoch": 1.4358596883090513,
      "grad_norm": 0.24225115776062012,
      "learning_rate": 1.042760207793966e-05,
      "loss": 0.0381,
      "step": 28285
    },
    {
      "epoch": 1.4361135082999137,
      "grad_norm": 0.47621268033981323,
      "learning_rate": 1.0425909944667243e-05,
      "loss": 0.0483,
      "step": 28290
    },
    {
      "epoch": 1.436367328290776,
      "grad_norm": 0.3224206566810608,
      "learning_rate": 1.0424217811394826e-05,
      "loss": 0.0391,
      "step": 28295
    },
    {
      "epoch": 1.4366211482816387,
      "grad_norm": 0.6168380379676819,
      "learning_rate": 1.042252567812241e-05,
      "loss": 0.0455,
      "step": 28300
    },
    {
      "epoch": 1.4368749682725013,
      "grad_norm": 0.6423296928405762,
      "learning_rate": 1.0420833544849993e-05,
      "loss": 0.0557,
      "step": 28305
    },
    {
      "epoch": 1.4371287882633637,
      "grad_norm": 0.4000137150287628,
      "learning_rate": 1.0419141411577575e-05,
      "loss": 0.0383,
      "step": 28310
    },
    {
      "epoch": 1.437382608254226,
      "grad_norm": 0.8485282063484192,
      "learning_rate": 1.041744927830516e-05,
      "loss": 0.0499,
      "step": 28315
    },
    {
      "epoch": 1.4376364282450886,
      "grad_norm": 0.4467218816280365,
      "learning_rate": 1.0415757145032744e-05,
      "loss": 0.039,
      "step": 28320
    },
    {
      "epoch": 1.437890248235951,
      "grad_norm": 0.5513525009155273,
      "learning_rate": 1.0414065011760327e-05,
      "loss": 0.0505,
      "step": 28325
    },
    {
      "epoch": 1.4381440682268136,
      "grad_norm": 0.36793988943099976,
      "learning_rate": 1.0412372878487911e-05,
      "loss": 0.0448,
      "step": 28330
    },
    {
      "epoch": 1.438397888217676,
      "grad_norm": 0.500225305557251,
      "learning_rate": 1.0410680745215493e-05,
      "loss": 0.0443,
      "step": 28335
    },
    {
      "epoch": 1.4386517082085386,
      "grad_norm": 0.4077879786491394,
      "learning_rate": 1.0408988611943078e-05,
      "loss": 0.0456,
      "step": 28340
    },
    {
      "epoch": 1.438905528199401,
      "grad_norm": 0.28105783462524414,
      "learning_rate": 1.0407296478670662e-05,
      "loss": 0.0425,
      "step": 28345
    },
    {
      "epoch": 1.4391593481902634,
      "grad_norm": 1.5346609354019165,
      "learning_rate": 1.0405604345398243e-05,
      "loss": 0.0493,
      "step": 28350
    },
    {
      "epoch": 1.439413168181126,
      "grad_norm": 0.256663978099823,
      "learning_rate": 1.0403912212125829e-05,
      "loss": 0.0445,
      "step": 28355
    },
    {
      "epoch": 1.4396669881719883,
      "grad_norm": 0.5210541486740112,
      "learning_rate": 1.040222007885341e-05,
      "loss": 0.0441,
      "step": 28360
    },
    {
      "epoch": 1.439920808162851,
      "grad_norm": 0.3798723518848419,
      "learning_rate": 1.0400527945580994e-05,
      "loss": 0.048,
      "step": 28365
    },
    {
      "epoch": 1.4401746281537133,
      "grad_norm": 0.3566092550754547,
      "learning_rate": 1.039883581230858e-05,
      "loss": 0.0417,
      "step": 28370
    },
    {
      "epoch": 1.440428448144576,
      "grad_norm": 0.358541339635849,
      "learning_rate": 1.0397143679036161e-05,
      "loss": 0.0444,
      "step": 28375
    },
    {
      "epoch": 1.4406822681354383,
      "grad_norm": 0.3433448374271393,
      "learning_rate": 1.0395451545763746e-05,
      "loss": 0.0462,
      "step": 28380
    },
    {
      "epoch": 1.4409360881263007,
      "grad_norm": 0.3503459095954895,
      "learning_rate": 1.0393759412491328e-05,
      "loss": 0.0598,
      "step": 28385
    },
    {
      "epoch": 1.4411899081171633,
      "grad_norm": 0.5386857986450195,
      "learning_rate": 1.0392067279218912e-05,
      "loss": 0.0478,
      "step": 28390
    },
    {
      "epoch": 1.441443728108026,
      "grad_norm": 0.39092475175857544,
      "learning_rate": 1.0390375145946497e-05,
      "loss": 0.0463,
      "step": 28395
    },
    {
      "epoch": 1.4416975480988883,
      "grad_norm": 1.031701922416687,
      "learning_rate": 1.0388683012674079e-05,
      "loss": 0.0485,
      "step": 28400
    },
    {
      "epoch": 1.4419513680897507,
      "grad_norm": 0.32506614923477173,
      "learning_rate": 1.038699087940166e-05,
      "loss": 0.0484,
      "step": 28405
    },
    {
      "epoch": 1.4422051880806133,
      "grad_norm": 0.4326903522014618,
      "learning_rate": 1.0385298746129246e-05,
      "loss": 0.0486,
      "step": 28410
    },
    {
      "epoch": 1.4424590080714756,
      "grad_norm": 0.36183029413223267,
      "learning_rate": 1.038360661285683e-05,
      "loss": 0.0436,
      "step": 28415
    },
    {
      "epoch": 1.4427128280623382,
      "grad_norm": 0.37883949279785156,
      "learning_rate": 1.0381914479584415e-05,
      "loss": 0.0523,
      "step": 28420
    },
    {
      "epoch": 1.4429666480532006,
      "grad_norm": 0.26860812306404114,
      "learning_rate": 1.0380222346311997e-05,
      "loss": 0.0474,
      "step": 28425
    },
    {
      "epoch": 1.4432204680440632,
      "grad_norm": 0.706049382686615,
      "learning_rate": 1.0378530213039579e-05,
      "loss": 0.0511,
      "step": 28430
    },
    {
      "epoch": 1.4434742880349256,
      "grad_norm": 0.628073513507843,
      "learning_rate": 1.0376838079767164e-05,
      "loss": 0.0491,
      "step": 28435
    },
    {
      "epoch": 1.443728108025788,
      "grad_norm": 0.45285552740097046,
      "learning_rate": 1.0375145946494747e-05,
      "loss": 0.0448,
      "step": 28440
    },
    {
      "epoch": 1.4439819280166506,
      "grad_norm": 0.22899897396564484,
      "learning_rate": 1.0373453813222329e-05,
      "loss": 0.0463,
      "step": 28445
    },
    {
      "epoch": 1.4442357480075132,
      "grad_norm": 0.2924014627933502,
      "learning_rate": 1.0371761679949914e-05,
      "loss": 0.0488,
      "step": 28450
    },
    {
      "epoch": 1.4444895679983756,
      "grad_norm": 0.2901057302951813,
      "learning_rate": 1.0370069546677496e-05,
      "loss": 0.043,
      "step": 28455
    },
    {
      "epoch": 1.444743387989238,
      "grad_norm": 0.24489150941371918,
      "learning_rate": 1.036837741340508e-05,
      "loss": 0.0444,
      "step": 28460
    },
    {
      "epoch": 1.4449972079801006,
      "grad_norm": 0.2966690957546234,
      "learning_rate": 1.0366685280132665e-05,
      "loss": 0.0572,
      "step": 28465
    },
    {
      "epoch": 1.445251027970963,
      "grad_norm": 0.2543412744998932,
      "learning_rate": 1.0364993146860247e-05,
      "loss": 0.0413,
      "step": 28470
    },
    {
      "epoch": 1.4455048479618255,
      "grad_norm": 0.3248431384563446,
      "learning_rate": 1.0363301013587832e-05,
      "loss": 0.0474,
      "step": 28475
    },
    {
      "epoch": 1.445758667952688,
      "grad_norm": 0.2841273248195648,
      "learning_rate": 1.0361608880315414e-05,
      "loss": 0.0456,
      "step": 28480
    },
    {
      "epoch": 1.4460124879435505,
      "grad_norm": 0.7544896006584167,
      "learning_rate": 1.0359916747042997e-05,
      "loss": 0.0529,
      "step": 28485
    },
    {
      "epoch": 1.446266307934413,
      "grad_norm": 0.3466797173023224,
      "learning_rate": 1.0358224613770583e-05,
      "loss": 0.043,
      "step": 28490
    },
    {
      "epoch": 1.4465201279252753,
      "grad_norm": 0.3613210618495941,
      "learning_rate": 1.0356532480498165e-05,
      "loss": 0.0458,
      "step": 28495
    },
    {
      "epoch": 1.446773947916138,
      "grad_norm": 0.32847079634666443,
      "learning_rate": 1.0354840347225748e-05,
      "loss": 0.0359,
      "step": 28500
    },
    {
      "epoch": 1.4470277679070003,
      "grad_norm": 0.9837027192115784,
      "learning_rate": 1.0353148213953332e-05,
      "loss": 0.044,
      "step": 28505
    },
    {
      "epoch": 1.4472815878978629,
      "grad_norm": 0.2532666027545929,
      "learning_rate": 1.0351456080680915e-05,
      "loss": 0.0427,
      "step": 28510
    },
    {
      "epoch": 1.4475354078887253,
      "grad_norm": 0.430247038602829,
      "learning_rate": 1.0349763947408497e-05,
      "loss": 0.0456,
      "step": 28515
    },
    {
      "epoch": 1.4477892278795879,
      "grad_norm": 0.6153171062469482,
      "learning_rate": 1.0348071814136082e-05,
      "loss": 0.0404,
      "step": 28520
    },
    {
      "epoch": 1.4480430478704502,
      "grad_norm": 0.24143315851688385,
      "learning_rate": 1.0346379680863666e-05,
      "loss": 0.0368,
      "step": 28525
    },
    {
      "epoch": 1.4482968678613126,
      "grad_norm": 0.36326682567596436,
      "learning_rate": 1.034468754759125e-05,
      "loss": 0.0453,
      "step": 28530
    },
    {
      "epoch": 1.4485506878521752,
      "grad_norm": 0.34203532338142395,
      "learning_rate": 1.0342995414318833e-05,
      "loss": 0.042,
      "step": 28535
    },
    {
      "epoch": 1.4488045078430378,
      "grad_norm": 0.6145029664039612,
      "learning_rate": 1.0341303281046415e-05,
      "loss": 0.0521,
      "step": 28540
    },
    {
      "epoch": 1.4490583278339002,
      "grad_norm": 0.3803384602069855,
      "learning_rate": 1.0339611147774e-05,
      "loss": 0.0453,
      "step": 28545
    },
    {
      "epoch": 1.4493121478247626,
      "grad_norm": 0.36794525384902954,
      "learning_rate": 1.0337919014501584e-05,
      "loss": 0.0524,
      "step": 28550
    },
    {
      "epoch": 1.4495659678156252,
      "grad_norm": 0.28114748001098633,
      "learning_rate": 1.0336226881229165e-05,
      "loss": 0.0347,
      "step": 28555
    },
    {
      "epoch": 1.4498197878064876,
      "grad_norm": 0.7040995359420776,
      "learning_rate": 1.033453474795675e-05,
      "loss": 0.0548,
      "step": 28560
    },
    {
      "epoch": 1.4500736077973502,
      "grad_norm": 0.3672772943973541,
      "learning_rate": 1.0332842614684333e-05,
      "loss": 0.053,
      "step": 28565
    },
    {
      "epoch": 1.4503274277882126,
      "grad_norm": 0.29484474658966064,
      "learning_rate": 1.0331150481411918e-05,
      "loss": 0.0462,
      "step": 28570
    },
    {
      "epoch": 1.4505812477790752,
      "grad_norm": 0.32553452253341675,
      "learning_rate": 1.0329458348139501e-05,
      "loss": 0.0391,
      "step": 28575
    },
    {
      "epoch": 1.4508350677699375,
      "grad_norm": 0.32741355895996094,
      "learning_rate": 1.0327766214867083e-05,
      "loss": 0.0424,
      "step": 28580
    },
    {
      "epoch": 1.4510888877608,
      "grad_norm": 0.3792947232723236,
      "learning_rate": 1.0326074081594668e-05,
      "loss": 0.0457,
      "step": 28585
    },
    {
      "epoch": 1.4513427077516625,
      "grad_norm": 0.47558140754699707,
      "learning_rate": 1.032438194832225e-05,
      "loss": 0.0485,
      "step": 28590
    },
    {
      "epoch": 1.4515965277425251,
      "grad_norm": 0.4144081175327301,
      "learning_rate": 1.0322689815049834e-05,
      "loss": 0.0492,
      "step": 28595
    },
    {
      "epoch": 1.4518503477333875,
      "grad_norm": 0.41987255215644836,
      "learning_rate": 1.0320997681777419e-05,
      "loss": 0.0488,
      "step": 28600
    },
    {
      "epoch": 1.4521041677242499,
      "grad_norm": 0.3136137127876282,
      "learning_rate": 1.0319305548505001e-05,
      "loss": 0.0441,
      "step": 28605
    },
    {
      "epoch": 1.4523579877151125,
      "grad_norm": 0.4296714663505554,
      "learning_rate": 1.0317613415232583e-05,
      "loss": 0.0501,
      "step": 28610
    },
    {
      "epoch": 1.4526118077059749,
      "grad_norm": 0.3685763478279114,
      "learning_rate": 1.0315921281960168e-05,
      "loss": 0.0526,
      "step": 28615
    },
    {
      "epoch": 1.4528656276968375,
      "grad_norm": 0.426189661026001,
      "learning_rate": 1.0314229148687752e-05,
      "loss": 0.0493,
      "step": 28620
    },
    {
      "epoch": 1.4531194476876999,
      "grad_norm": 0.38130417466163635,
      "learning_rate": 1.0312537015415337e-05,
      "loss": 0.0429,
      "step": 28625
    },
    {
      "epoch": 1.4533732676785625,
      "grad_norm": 0.4051515460014343,
      "learning_rate": 1.0310844882142919e-05,
      "loss": 0.0481,
      "step": 28630
    },
    {
      "epoch": 1.4536270876694248,
      "grad_norm": 0.4713396728038788,
      "learning_rate": 1.03091527488705e-05,
      "loss": 0.0461,
      "step": 28635
    },
    {
      "epoch": 1.4538809076602872,
      "grad_norm": 0.30494242906570435,
      "learning_rate": 1.0307460615598086e-05,
      "loss": 0.0418,
      "step": 28640
    },
    {
      "epoch": 1.4541347276511498,
      "grad_norm": 0.3397122621536255,
      "learning_rate": 1.030576848232567e-05,
      "loss": 0.0505,
      "step": 28645
    },
    {
      "epoch": 1.4543885476420124,
      "grad_norm": 0.4459775686264038,
      "learning_rate": 1.0304076349053251e-05,
      "loss": 0.0471,
      "step": 28650
    },
    {
      "epoch": 1.4546423676328748,
      "grad_norm": 0.6436803936958313,
      "learning_rate": 1.0302384215780836e-05,
      "loss": 0.0447,
      "step": 28655
    },
    {
      "epoch": 1.4548961876237372,
      "grad_norm": 0.3298328220844269,
      "learning_rate": 1.0300692082508418e-05,
      "loss": 0.051,
      "step": 28660
    },
    {
      "epoch": 1.4551500076145998,
      "grad_norm": 0.3109586238861084,
      "learning_rate": 1.0298999949236003e-05,
      "loss": 0.0456,
      "step": 28665
    },
    {
      "epoch": 1.4554038276054622,
      "grad_norm": 0.39373233914375305,
      "learning_rate": 1.0297307815963587e-05,
      "loss": 0.0458,
      "step": 28670
    },
    {
      "epoch": 1.4556576475963248,
      "grad_norm": 0.34695670008659363,
      "learning_rate": 1.0295615682691169e-05,
      "loss": 0.045,
      "step": 28675
    },
    {
      "epoch": 1.4559114675871871,
      "grad_norm": 0.369632750749588,
      "learning_rate": 1.0293923549418754e-05,
      "loss": 0.0497,
      "step": 28680
    },
    {
      "epoch": 1.4561652875780497,
      "grad_norm": 0.41509467363357544,
      "learning_rate": 1.0292231416146336e-05,
      "loss": 0.0525,
      "step": 28685
    },
    {
      "epoch": 1.4564191075689121,
      "grad_norm": 0.4681684672832489,
      "learning_rate": 1.029053928287392e-05,
      "loss": 0.0543,
      "step": 28690
    },
    {
      "epoch": 1.4566729275597745,
      "grad_norm": 0.46059858798980713,
      "learning_rate": 1.0288847149601505e-05,
      "loss": 0.0432,
      "step": 28695
    },
    {
      "epoch": 1.4569267475506371,
      "grad_norm": 0.33197858929634094,
      "learning_rate": 1.0287155016329087e-05,
      "loss": 0.0458,
      "step": 28700
    },
    {
      "epoch": 1.4571805675414995,
      "grad_norm": 0.3484487235546112,
      "learning_rate": 1.028546288305667e-05,
      "loss": 0.0455,
      "step": 28705
    },
    {
      "epoch": 1.457434387532362,
      "grad_norm": 0.3765062689781189,
      "learning_rate": 1.0283770749784254e-05,
      "loss": 0.0399,
      "step": 28710
    },
    {
      "epoch": 1.4576882075232245,
      "grad_norm": 0.3412276804447174,
      "learning_rate": 1.0282078616511837e-05,
      "loss": 0.0444,
      "step": 28715
    },
    {
      "epoch": 1.457942027514087,
      "grad_norm": 0.5508232116699219,
      "learning_rate": 1.0280386483239422e-05,
      "loss": 0.0599,
      "step": 28720
    },
    {
      "epoch": 1.4581958475049495,
      "grad_norm": 0.294949471950531,
      "learning_rate": 1.0278694349967004e-05,
      "loss": 0.0422,
      "step": 28725
    },
    {
      "epoch": 1.4584496674958118,
      "grad_norm": 0.501362681388855,
      "learning_rate": 1.0277002216694588e-05,
      "loss": 0.0411,
      "step": 28730
    },
    {
      "epoch": 1.4587034874866744,
      "grad_norm": 0.3812694847583771,
      "learning_rate": 1.0275310083422171e-05,
      "loss": 0.0492,
      "step": 28735
    },
    {
      "epoch": 1.458957307477537,
      "grad_norm": 0.3240166902542114,
      "learning_rate": 1.0273617950149755e-05,
      "loss": 0.0436,
      "step": 28740
    },
    {
      "epoch": 1.4592111274683994,
      "grad_norm": 0.3954068720340729,
      "learning_rate": 1.0271925816877337e-05,
      "loss": 0.0529,
      "step": 28745
    },
    {
      "epoch": 1.4594649474592618,
      "grad_norm": 0.2987024486064911,
      "learning_rate": 1.0270233683604922e-05,
      "loss": 0.0485,
      "step": 28750
    },
    {
      "epoch": 1.4597187674501244,
      "grad_norm": 0.4231718182563782,
      "learning_rate": 1.0268541550332506e-05,
      "loss": 0.0434,
      "step": 28755
    },
    {
      "epoch": 1.4599725874409868,
      "grad_norm": 1.153788447380066,
      "learning_rate": 1.0266849417060087e-05,
      "loss": 0.041,
      "step": 28760
    },
    {
      "epoch": 1.4602264074318494,
      "grad_norm": 0.3164049983024597,
      "learning_rate": 1.0265157283787673e-05,
      "loss": 0.0395,
      "step": 28765
    },
    {
      "epoch": 1.4604802274227118,
      "grad_norm": 0.36508455872535706,
      "learning_rate": 1.0263465150515254e-05,
      "loss": 0.0499,
      "step": 28770
    },
    {
      "epoch": 1.4607340474135744,
      "grad_norm": 0.26266711950302124,
      "learning_rate": 1.026177301724284e-05,
      "loss": 0.0441,
      "step": 28775
    },
    {
      "epoch": 1.4609878674044368,
      "grad_norm": 0.29270073771476746,
      "learning_rate": 1.0260080883970423e-05,
      "loss": 0.0472,
      "step": 28780
    },
    {
      "epoch": 1.4612416873952991,
      "grad_norm": 0.302064448595047,
      "learning_rate": 1.0258388750698005e-05,
      "loss": 0.0453,
      "step": 28785
    },
    {
      "epoch": 1.4614955073861617,
      "grad_norm": 0.2645045220851898,
      "learning_rate": 1.025669661742559e-05,
      "loss": 0.0443,
      "step": 28790
    },
    {
      "epoch": 1.4617493273770243,
      "grad_norm": 0.3246147334575653,
      "learning_rate": 1.0255004484153172e-05,
      "loss": 0.0364,
      "step": 28795
    },
    {
      "epoch": 1.4620031473678867,
      "grad_norm": 0.3109263479709625,
      "learning_rate": 1.0253312350880756e-05,
      "loss": 0.0487,
      "step": 28800
    },
    {
      "epoch": 1.462256967358749,
      "grad_norm": 0.3940425217151642,
      "learning_rate": 1.0251620217608341e-05,
      "loss": 0.0499,
      "step": 28805
    },
    {
      "epoch": 1.4625107873496117,
      "grad_norm": 0.38812288641929626,
      "learning_rate": 1.0249928084335923e-05,
      "loss": 0.0474,
      "step": 28810
    },
    {
      "epoch": 1.462764607340474,
      "grad_norm": 0.2860491871833801,
      "learning_rate": 1.0248235951063508e-05,
      "loss": 0.0391,
      "step": 28815
    },
    {
      "epoch": 1.4630184273313367,
      "grad_norm": 0.4146775007247925,
      "learning_rate": 1.024654381779109e-05,
      "loss": 0.0463,
      "step": 28820
    },
    {
      "epoch": 1.463272247322199,
      "grad_norm": 0.5292524695396423,
      "learning_rate": 1.0244851684518673e-05,
      "loss": 0.0467,
      "step": 28825
    },
    {
      "epoch": 1.4635260673130617,
      "grad_norm": 0.2877132296562195,
      "learning_rate": 1.0243159551246257e-05,
      "loss": 0.0435,
      "step": 28830
    },
    {
      "epoch": 1.463779887303924,
      "grad_norm": 0.31168654561042786,
      "learning_rate": 1.024146741797384e-05,
      "loss": 0.039,
      "step": 28835
    },
    {
      "epoch": 1.4640337072947864,
      "grad_norm": 0.37059271335601807,
      "learning_rate": 1.0239775284701422e-05,
      "loss": 0.0513,
      "step": 28840
    },
    {
      "epoch": 1.464287527285649,
      "grad_norm": 0.3526490330696106,
      "learning_rate": 1.0238083151429008e-05,
      "loss": 0.0445,
      "step": 28845
    },
    {
      "epoch": 1.4645413472765114,
      "grad_norm": 0.43100833892822266,
      "learning_rate": 1.0236391018156591e-05,
      "loss": 0.047,
      "step": 28850
    },
    {
      "epoch": 1.464795167267374,
      "grad_norm": 0.45979759097099304,
      "learning_rate": 1.0234698884884173e-05,
      "loss": 0.0461,
      "step": 28855
    },
    {
      "epoch": 1.4650489872582364,
      "grad_norm": 0.34121495485305786,
      "learning_rate": 1.0233006751611758e-05,
      "loss": 0.0493,
      "step": 28860
    },
    {
      "epoch": 1.465302807249099,
      "grad_norm": 0.6851451396942139,
      "learning_rate": 1.023131461833934e-05,
      "loss": 0.0424,
      "step": 28865
    },
    {
      "epoch": 1.4655566272399614,
      "grad_norm": 0.3656606674194336,
      "learning_rate": 1.0229622485066925e-05,
      "loss": 0.053,
      "step": 28870
    },
    {
      "epoch": 1.4658104472308238,
      "grad_norm": 0.31337806582450867,
      "learning_rate": 1.0227930351794509e-05,
      "loss": 0.047,
      "step": 28875
    },
    {
      "epoch": 1.4660642672216864,
      "grad_norm": 0.4288000464439392,
      "learning_rate": 1.022623821852209e-05,
      "loss": 0.0553,
      "step": 28880
    },
    {
      "epoch": 1.466318087212549,
      "grad_norm": 0.34418991208076477,
      "learning_rate": 1.0224546085249676e-05,
      "loss": 0.0542,
      "step": 28885
    },
    {
      "epoch": 1.4665719072034114,
      "grad_norm": 0.40049436688423157,
      "learning_rate": 1.0222853951977258e-05,
      "loss": 0.0441,
      "step": 28890
    },
    {
      "epoch": 1.4668257271942737,
      "grad_norm": 0.35026365518569946,
      "learning_rate": 1.0221161818704841e-05,
      "loss": 0.0461,
      "step": 28895
    },
    {
      "epoch": 1.4670795471851363,
      "grad_norm": 0.24924367666244507,
      "learning_rate": 1.0219469685432427e-05,
      "loss": 0.0509,
      "step": 28900
    },
    {
      "epoch": 1.4673333671759987,
      "grad_norm": 0.2462109625339508,
      "learning_rate": 1.0217777552160008e-05,
      "loss": 0.0362,
      "step": 28905
    },
    {
      "epoch": 1.4675871871668613,
      "grad_norm": 0.49023517966270447,
      "learning_rate": 1.0216085418887594e-05,
      "loss": 0.044,
      "step": 28910
    },
    {
      "epoch": 1.4678410071577237,
      "grad_norm": 0.24693015217781067,
      "learning_rate": 1.0214393285615176e-05,
      "loss": 0.0455,
      "step": 28915
    },
    {
      "epoch": 1.4680948271485863,
      "grad_norm": 0.31813114881515503,
      "learning_rate": 1.0212701152342759e-05,
      "loss": 0.0462,
      "step": 28920
    },
    {
      "epoch": 1.4683486471394487,
      "grad_norm": 0.367294579744339,
      "learning_rate": 1.0211009019070344e-05,
      "loss": 0.0441,
      "step": 28925
    },
    {
      "epoch": 1.468602467130311,
      "grad_norm": 0.48306939005851746,
      "learning_rate": 1.0209316885797926e-05,
      "loss": 0.0578,
      "step": 28930
    },
    {
      "epoch": 1.4688562871211737,
      "grad_norm": 0.6266639232635498,
      "learning_rate": 1.020762475252551e-05,
      "loss": 0.0431,
      "step": 28935
    },
    {
      "epoch": 1.4691101071120363,
      "grad_norm": 0.4411011338233948,
      "learning_rate": 1.0205932619253093e-05,
      "loss": 0.0436,
      "step": 28940
    },
    {
      "epoch": 1.4693639271028986,
      "grad_norm": 0.6849778294563293,
      "learning_rate": 1.0204240485980677e-05,
      "loss": 0.0499,
      "step": 28945
    },
    {
      "epoch": 1.469617747093761,
      "grad_norm": 0.2983316481113434,
      "learning_rate": 1.0202548352708259e-05,
      "loss": 0.0375,
      "step": 28950
    },
    {
      "epoch": 1.4698715670846236,
      "grad_norm": 0.3210887908935547,
      "learning_rate": 1.0200856219435844e-05,
      "loss": 0.0518,
      "step": 28955
    },
    {
      "epoch": 1.470125387075486,
      "grad_norm": 0.4435349404811859,
      "learning_rate": 1.0199164086163427e-05,
      "loss": 0.0461,
      "step": 28960
    },
    {
      "epoch": 1.4703792070663486,
      "grad_norm": 0.4102541208267212,
      "learning_rate": 1.0197471952891011e-05,
      "loss": 0.0443,
      "step": 28965
    },
    {
      "epoch": 1.470633027057211,
      "grad_norm": 0.3138459324836731,
      "learning_rate": 1.0195779819618595e-05,
      "loss": 0.0485,
      "step": 28970
    },
    {
      "epoch": 1.4708868470480736,
      "grad_norm": 0.46446776390075684,
      "learning_rate": 1.0194087686346176e-05,
      "loss": 0.0462,
      "step": 28975
    },
    {
      "epoch": 1.471140667038936,
      "grad_norm": 0.3791068494319916,
      "learning_rate": 1.0192395553073762e-05,
      "loss": 0.0433,
      "step": 28980
    },
    {
      "epoch": 1.4713944870297984,
      "grad_norm": 0.4010803997516632,
      "learning_rate": 1.0190703419801345e-05,
      "loss": 0.0515,
      "step": 28985
    },
    {
      "epoch": 1.471648307020661,
      "grad_norm": 0.3441152274608612,
      "learning_rate": 1.0189011286528927e-05,
      "loss": 0.0498,
      "step": 28990
    },
    {
      "epoch": 1.4719021270115236,
      "grad_norm": 0.40290719270706177,
      "learning_rate": 1.0187319153256512e-05,
      "loss": 0.0454,
      "step": 28995
    },
    {
      "epoch": 1.472155947002386,
      "grad_norm": 0.9115328192710876,
      "learning_rate": 1.0185627019984094e-05,
      "loss": 0.0412,
      "step": 29000
    },
    {
      "epoch": 1.4724097669932483,
      "grad_norm": 0.3189408779144287,
      "learning_rate": 1.0183934886711678e-05,
      "loss": 0.0493,
      "step": 29005
    },
    {
      "epoch": 1.472663586984111,
      "grad_norm": 0.3260115385055542,
      "learning_rate": 1.0182242753439263e-05,
      "loss": 0.0418,
      "step": 29010
    },
    {
      "epoch": 1.4729174069749733,
      "grad_norm": 0.3087772727012634,
      "learning_rate": 1.0180550620166845e-05,
      "loss": 0.0489,
      "step": 29015
    },
    {
      "epoch": 1.473171226965836,
      "grad_norm": 0.39418232440948486,
      "learning_rate": 1.017885848689443e-05,
      "loss": 0.0396,
      "step": 29020
    },
    {
      "epoch": 1.4734250469566983,
      "grad_norm": 0.2795569896697998,
      "learning_rate": 1.0177166353622012e-05,
      "loss": 0.0512,
      "step": 29025
    },
    {
      "epoch": 1.473678866947561,
      "grad_norm": 0.5018810629844666,
      "learning_rate": 1.0175474220349595e-05,
      "loss": 0.0571,
      "step": 29030
    },
    {
      "epoch": 1.4739326869384233,
      "grad_norm": 0.49596336483955383,
      "learning_rate": 1.0173782087077179e-05,
      "loss": 0.0487,
      "step": 29035
    },
    {
      "epoch": 1.4741865069292857,
      "grad_norm": 0.3247216045856476,
      "learning_rate": 1.0172089953804762e-05,
      "loss": 0.043,
      "step": 29040
    },
    {
      "epoch": 1.4744403269201483,
      "grad_norm": 0.43180331587791443,
      "learning_rate": 1.0170397820532344e-05,
      "loss": 0.0447,
      "step": 29045
    },
    {
      "epoch": 1.4746941469110106,
      "grad_norm": 0.44530582427978516,
      "learning_rate": 1.016870568725993e-05,
      "loss": 0.0385,
      "step": 29050
    },
    {
      "epoch": 1.4749479669018732,
      "grad_norm": 0.37632861733436584,
      "learning_rate": 1.0167013553987513e-05,
      "loss": 0.0455,
      "step": 29055
    },
    {
      "epoch": 1.4752017868927356,
      "grad_norm": 0.3150545060634613,
      "learning_rate": 1.0165321420715097e-05,
      "loss": 0.0464,
      "step": 29060
    },
    {
      "epoch": 1.4754556068835982,
      "grad_norm": 0.32982978224754333,
      "learning_rate": 1.016362928744268e-05,
      "loss": 0.0439,
      "step": 29065
    },
    {
      "epoch": 1.4757094268744606,
      "grad_norm": 0.22459256649017334,
      "learning_rate": 1.0161937154170262e-05,
      "loss": 0.0457,
      "step": 29070
    },
    {
      "epoch": 1.475963246865323,
      "grad_norm": 0.6925598978996277,
      "learning_rate": 1.0160245020897847e-05,
      "loss": 0.0487,
      "step": 29075
    },
    {
      "epoch": 1.4762170668561856,
      "grad_norm": 0.2393013834953308,
      "learning_rate": 1.015855288762543e-05,
      "loss": 0.0414,
      "step": 29080
    },
    {
      "epoch": 1.4764708868470482,
      "grad_norm": 0.40662673115730286,
      "learning_rate": 1.0156860754353013e-05,
      "loss": 0.0491,
      "step": 29085
    },
    {
      "epoch": 1.4767247068379106,
      "grad_norm": 0.35922446846961975,
      "learning_rate": 1.0155168621080598e-05,
      "loss": 0.0472,
      "step": 29090
    },
    {
      "epoch": 1.476978526828773,
      "grad_norm": 0.284529447555542,
      "learning_rate": 1.015347648780818e-05,
      "loss": 0.046,
      "step": 29095
    },
    {
      "epoch": 1.4772323468196356,
      "grad_norm": 0.3706730604171753,
      "learning_rate": 1.0151784354535763e-05,
      "loss": 0.05,
      "step": 29100
    },
    {
      "epoch": 1.477486166810498,
      "grad_norm": 0.24887824058532715,
      "learning_rate": 1.0150092221263349e-05,
      "loss": 0.0528,
      "step": 29105
    },
    {
      "epoch": 1.4777399868013605,
      "grad_norm": 0.458808571100235,
      "learning_rate": 1.014840008799093e-05,
      "loss": 0.0519,
      "step": 29110
    },
    {
      "epoch": 1.477993806792223,
      "grad_norm": 0.32688531279563904,
      "learning_rate": 1.0146707954718516e-05,
      "loss": 0.0529,
      "step": 29115
    },
    {
      "epoch": 1.4782476267830855,
      "grad_norm": 0.22939306497573853,
      "learning_rate": 1.0145015821446097e-05,
      "loss": 0.0368,
      "step": 29120
    },
    {
      "epoch": 1.478501446773948,
      "grad_norm": 0.38046857714653015,
      "learning_rate": 1.0143323688173681e-05,
      "loss": 0.0492,
      "step": 29125
    },
    {
      "epoch": 1.4787552667648103,
      "grad_norm": 0.5389629602432251,
      "learning_rate": 1.0141631554901266e-05,
      "loss": 0.049,
      "step": 29130
    },
    {
      "epoch": 1.4790090867556729,
      "grad_norm": 0.6151430606842041,
      "learning_rate": 1.0139939421628848e-05,
      "loss": 0.0551,
      "step": 29135
    },
    {
      "epoch": 1.4792629067465355,
      "grad_norm": 0.33038952946662903,
      "learning_rate": 1.0138247288356432e-05,
      "loss": 0.0449,
      "step": 29140
    },
    {
      "epoch": 1.4795167267373979,
      "grad_norm": 0.388681024312973,
      "learning_rate": 1.0136555155084015e-05,
      "loss": 0.0511,
      "step": 29145
    },
    {
      "epoch": 1.4797705467282602,
      "grad_norm": 0.3388515114784241,
      "learning_rate": 1.0134863021811599e-05,
      "loss": 0.0479,
      "step": 29150
    },
    {
      "epoch": 1.4800243667191229,
      "grad_norm": 0.400495707988739,
      "learning_rate": 1.013317088853918e-05,
      "loss": 0.0588,
      "step": 29155
    },
    {
      "epoch": 1.4802781867099852,
      "grad_norm": 0.3645167946815491,
      "learning_rate": 1.0131478755266766e-05,
      "loss": 0.0497,
      "step": 29160
    },
    {
      "epoch": 1.4805320067008478,
      "grad_norm": 0.3763105869293213,
      "learning_rate": 1.012978662199435e-05,
      "loss": 0.0515,
      "step": 29165
    },
    {
      "epoch": 1.4807858266917102,
      "grad_norm": 0.34039294719696045,
      "learning_rate": 1.0128094488721933e-05,
      "loss": 0.0427,
      "step": 29170
    },
    {
      "epoch": 1.4810396466825728,
      "grad_norm": 0.2790921628475189,
      "learning_rate": 1.0126402355449516e-05,
      "loss": 0.038,
      "step": 29175
    },
    {
      "epoch": 1.4812934666734352,
      "grad_norm": 0.41854605078697205,
      "learning_rate": 1.0124710222177098e-05,
      "loss": 0.0457,
      "step": 29180
    },
    {
      "epoch": 1.4815472866642976,
      "grad_norm": 0.2072964310646057,
      "learning_rate": 1.0123018088904684e-05,
      "loss": 0.0408,
      "step": 29185
    },
    {
      "epoch": 1.4818011066551602,
      "grad_norm": 0.4667484164237976,
      "learning_rate": 1.0121325955632267e-05,
      "loss": 0.0449,
      "step": 29190
    },
    {
      "epoch": 1.4820549266460226,
      "grad_norm": 0.445342481136322,
      "learning_rate": 1.0119633822359849e-05,
      "loss": 0.0531,
      "step": 29195
    },
    {
      "epoch": 1.4823087466368852,
      "grad_norm": 0.3367258608341217,
      "learning_rate": 1.0117941689087434e-05,
      "loss": 0.0494,
      "step": 29200
    },
    {
      "epoch": 1.4825625666277475,
      "grad_norm": 0.5606625080108643,
      "learning_rate": 1.0116249555815016e-05,
      "loss": 0.0381,
      "step": 29205
    },
    {
      "epoch": 1.4828163866186101,
      "grad_norm": 0.5759447813034058,
      "learning_rate": 1.0114557422542601e-05,
      "loss": 0.042,
      "step": 29210
    },
    {
      "epoch": 1.4830702066094725,
      "grad_norm": 0.5777857899665833,
      "learning_rate": 1.0112865289270183e-05,
      "loss": 0.0453,
      "step": 29215
    },
    {
      "epoch": 1.483324026600335,
      "grad_norm": 0.5903242230415344,
      "learning_rate": 1.0111173155997767e-05,
      "loss": 0.0451,
      "step": 29220
    },
    {
      "epoch": 1.4835778465911975,
      "grad_norm": 0.39073795080184937,
      "learning_rate": 1.0109481022725352e-05,
      "loss": 0.0489,
      "step": 29225
    },
    {
      "epoch": 1.4838316665820601,
      "grad_norm": 0.4036313593387604,
      "learning_rate": 1.0107788889452934e-05,
      "loss": 0.0467,
      "step": 29230
    },
    {
      "epoch": 1.4840854865729225,
      "grad_norm": 0.35615524649620056,
      "learning_rate": 1.0106096756180517e-05,
      "loss": 0.0471,
      "step": 29235
    },
    {
      "epoch": 1.4843393065637849,
      "grad_norm": 0.34579744935035706,
      "learning_rate": 1.0104404622908101e-05,
      "loss": 0.0424,
      "step": 29240
    },
    {
      "epoch": 1.4845931265546475,
      "grad_norm": 0.2929094731807709,
      "learning_rate": 1.0102712489635684e-05,
      "loss": 0.0362,
      "step": 29245
    },
    {
      "epoch": 1.4848469465455099,
      "grad_norm": 0.39636480808258057,
      "learning_rate": 1.0101020356363266e-05,
      "loss": 0.0509,
      "step": 29250
    },
    {
      "epoch": 1.4851007665363725,
      "grad_norm": 0.33898094296455383,
      "learning_rate": 1.0099328223090851e-05,
      "loss": 0.0415,
      "step": 29255
    },
    {
      "epoch": 1.4853545865272348,
      "grad_norm": 0.4324086904525757,
      "learning_rate": 1.0097636089818435e-05,
      "loss": 0.0473,
      "step": 29260
    },
    {
      "epoch": 1.4856084065180974,
      "grad_norm": 0.37893491983413696,
      "learning_rate": 1.0095943956546019e-05,
      "loss": 0.0479,
      "step": 29265
    },
    {
      "epoch": 1.4858622265089598,
      "grad_norm": 0.4576340615749359,
      "learning_rate": 1.0094251823273602e-05,
      "loss": 0.0458,
      "step": 29270
    },
    {
      "epoch": 1.4861160464998222,
      "grad_norm": 0.2759975492954254,
      "learning_rate": 1.0092559690001184e-05,
      "loss": 0.0476,
      "step": 29275
    },
    {
      "epoch": 1.4863698664906848,
      "grad_norm": 0.3189256489276886,
      "learning_rate": 1.009086755672877e-05,
      "loss": 0.0433,
      "step": 29280
    },
    {
      "epoch": 1.4866236864815474,
      "grad_norm": 0.3150682747364044,
      "learning_rate": 1.0089175423456353e-05,
      "loss": 0.0427,
      "step": 29285
    },
    {
      "epoch": 1.4868775064724098,
      "grad_norm": 0.5731126666069031,
      "learning_rate": 1.0087483290183935e-05,
      "loss": 0.0456,
      "step": 29290
    },
    {
      "epoch": 1.4871313264632722,
      "grad_norm": 0.24449680745601654,
      "learning_rate": 1.008579115691152e-05,
      "loss": 0.0426,
      "step": 29295
    },
    {
      "epoch": 1.4873851464541348,
      "grad_norm": 0.487131804227829,
      "learning_rate": 1.0084099023639102e-05,
      "loss": 0.0487,
      "step": 29300
    },
    {
      "epoch": 1.4876389664449972,
      "grad_norm": 0.37090784311294556,
      "learning_rate": 1.0082406890366687e-05,
      "loss": 0.0483,
      "step": 29305
    },
    {
      "epoch": 1.4878927864358598,
      "grad_norm": 0.32718345522880554,
      "learning_rate": 1.008071475709427e-05,
      "loss": 0.0407,
      "step": 29310
    },
    {
      "epoch": 1.4881466064267221,
      "grad_norm": 0.7287764549255371,
      "learning_rate": 1.0079022623821852e-05,
      "loss": 0.0437,
      "step": 29315
    },
    {
      "epoch": 1.4884004264175847,
      "grad_norm": 0.44735628366470337,
      "learning_rate": 1.0077330490549438e-05,
      "loss": 0.0428,
      "step": 29320
    },
    {
      "epoch": 1.4886542464084471,
      "grad_norm": 0.4608113765716553,
      "learning_rate": 1.007563835727702e-05,
      "loss": 0.0481,
      "step": 29325
    },
    {
      "epoch": 1.4889080663993095,
      "grad_norm": 0.4977395534515381,
      "learning_rate": 1.0073946224004603e-05,
      "loss": 0.0496,
      "step": 29330
    },
    {
      "epoch": 1.489161886390172,
      "grad_norm": 0.308709055185318,
      "learning_rate": 1.0072254090732188e-05,
      "loss": 0.0465,
      "step": 29335
    },
    {
      "epoch": 1.4894157063810345,
      "grad_norm": 0.4213113486766815,
      "learning_rate": 1.007056195745977e-05,
      "loss": 0.0398,
      "step": 29340
    },
    {
      "epoch": 1.489669526371897,
      "grad_norm": 0.6217330694198608,
      "learning_rate": 1.0068869824187354e-05,
      "loss": 0.0457,
      "step": 29345
    },
    {
      "epoch": 1.4899233463627595,
      "grad_norm": 0.583604097366333,
      "learning_rate": 1.0067177690914937e-05,
      "loss": 0.0477,
      "step": 29350
    },
    {
      "epoch": 1.490177166353622,
      "grad_norm": 0.36942973732948303,
      "learning_rate": 1.006548555764252e-05,
      "loss": 0.0445,
      "step": 29355
    },
    {
      "epoch": 1.4904309863444845,
      "grad_norm": 0.28825482726097107,
      "learning_rate": 1.0063793424370106e-05,
      "loss": 0.0426,
      "step": 29360
    },
    {
      "epoch": 1.4906848063353468,
      "grad_norm": 0.3700205981731415,
      "learning_rate": 1.0062101291097688e-05,
      "loss": 0.0488,
      "step": 29365
    },
    {
      "epoch": 1.4909386263262094,
      "grad_norm": 0.9633322954177856,
      "learning_rate": 1.0060409157825271e-05,
      "loss": 0.0432,
      "step": 29370
    },
    {
      "epoch": 1.491192446317072,
      "grad_norm": 0.43203192949295044,
      "learning_rate": 1.0058717024552855e-05,
      "loss": 0.0464,
      "step": 29375
    },
    {
      "epoch": 1.4914462663079344,
      "grad_norm": 0.3489255905151367,
      "learning_rate": 1.0057024891280438e-05,
      "loss": 0.0456,
      "step": 29380
    },
    {
      "epoch": 1.4917000862987968,
      "grad_norm": 0.40560370683670044,
      "learning_rate": 1.005533275800802e-05,
      "loss": 0.0448,
      "step": 29385
    },
    {
      "epoch": 1.4919539062896594,
      "grad_norm": 0.3906811475753784,
      "learning_rate": 1.0053640624735606e-05,
      "loss": 0.0447,
      "step": 29390
    },
    {
      "epoch": 1.4922077262805218,
      "grad_norm": 0.42720428109169006,
      "learning_rate": 1.0051948491463187e-05,
      "loss": 0.0376,
      "step": 29395
    },
    {
      "epoch": 1.4924615462713844,
      "grad_norm": 0.5189972519874573,
      "learning_rate": 1.0050256358190771e-05,
      "loss": 0.0532,
      "step": 29400
    },
    {
      "epoch": 1.4927153662622468,
      "grad_norm": 1.2741813659667969,
      "learning_rate": 1.0048564224918356e-05,
      "loss": 0.0445,
      "step": 29405
    },
    {
      "epoch": 1.4929691862531094,
      "grad_norm": 0.43833017349243164,
      "learning_rate": 1.0046872091645938e-05,
      "loss": 0.0491,
      "step": 29410
    },
    {
      "epoch": 1.4932230062439718,
      "grad_norm": 0.28552454710006714,
      "learning_rate": 1.0045179958373523e-05,
      "loss": 0.0411,
      "step": 29415
    },
    {
      "epoch": 1.4934768262348341,
      "grad_norm": 0.39519840478897095,
      "learning_rate": 1.0043487825101105e-05,
      "loss": 0.0441,
      "step": 29420
    },
    {
      "epoch": 1.4937306462256967,
      "grad_norm": 0.5029251575469971,
      "learning_rate": 1.0041795691828689e-05,
      "loss": 0.0521,
      "step": 29425
    },
    {
      "epoch": 1.4939844662165593,
      "grad_norm": 0.3176358640193939,
      "learning_rate": 1.0040103558556274e-05,
      "loss": 0.042,
      "step": 29430
    },
    {
      "epoch": 1.4942382862074217,
      "grad_norm": 0.3860771059989929,
      "learning_rate": 1.0038411425283856e-05,
      "loss": 0.0435,
      "step": 29435
    },
    {
      "epoch": 1.494492106198284,
      "grad_norm": 0.5486845374107361,
      "learning_rate": 1.003671929201144e-05,
      "loss": 0.0478,
      "step": 29440
    },
    {
      "epoch": 1.4947459261891467,
      "grad_norm": 0.3547370135784149,
      "learning_rate": 1.0035027158739023e-05,
      "loss": 0.0401,
      "step": 29445
    },
    {
      "epoch": 1.494999746180009,
      "grad_norm": 0.21240559220314026,
      "learning_rate": 1.0033335025466606e-05,
      "loss": 0.0438,
      "step": 29450
    },
    {
      "epoch": 1.4952535661708717,
      "grad_norm": 0.2859129309654236,
      "learning_rate": 1.0031642892194192e-05,
      "loss": 0.0405,
      "step": 29455
    },
    {
      "epoch": 1.495507386161734,
      "grad_norm": 0.6197316646575928,
      "learning_rate": 1.0029950758921773e-05,
      "loss": 0.0468,
      "step": 29460
    },
    {
      "epoch": 1.4957612061525967,
      "grad_norm": 0.46611955761909485,
      "learning_rate": 1.0028258625649357e-05,
      "loss": 0.0433,
      "step": 29465
    },
    {
      "epoch": 1.496015026143459,
      "grad_norm": 0.3373320698738098,
      "learning_rate": 1.002656649237694e-05,
      "loss": 0.0438,
      "step": 29470
    },
    {
      "epoch": 1.4962688461343214,
      "grad_norm": 0.40270277857780457,
      "learning_rate": 1.0024874359104524e-05,
      "loss": 0.0568,
      "step": 29475
    },
    {
      "epoch": 1.496522666125184,
      "grad_norm": 0.5500027537345886,
      "learning_rate": 1.0023182225832106e-05,
      "loss": 0.0428,
      "step": 29480
    },
    {
      "epoch": 1.4967764861160466,
      "grad_norm": 0.5985025763511658,
      "learning_rate": 1.0021490092559691e-05,
      "loss": 0.053,
      "step": 29485
    },
    {
      "epoch": 1.497030306106909,
      "grad_norm": 0.39436739683151245,
      "learning_rate": 1.0019797959287275e-05,
      "loss": 0.0457,
      "step": 29490
    },
    {
      "epoch": 1.4972841260977714,
      "grad_norm": 0.8494074940681458,
      "learning_rate": 1.0018105826014857e-05,
      "loss": 0.0453,
      "step": 29495
    },
    {
      "epoch": 1.497537946088634,
      "grad_norm": 0.3544905185699463,
      "learning_rate": 1.0016413692742442e-05,
      "loss": 0.0446,
      "step": 29500
    },
    {
      "epoch": 1.4977917660794964,
      "grad_norm": 0.3078000545501709,
      "learning_rate": 1.0014721559470024e-05,
      "loss": 0.0426,
      "step": 29505
    },
    {
      "epoch": 1.498045586070359,
      "grad_norm": 0.3942676782608032,
      "learning_rate": 1.0013029426197609e-05,
      "loss": 0.0475,
      "step": 29510
    },
    {
      "epoch": 1.4982994060612214,
      "grad_norm": 0.3200058341026306,
      "learning_rate": 1.0011337292925192e-05,
      "loss": 0.0461,
      "step": 29515
    },
    {
      "epoch": 1.498553226052084,
      "grad_norm": 0.4130484163761139,
      "learning_rate": 1.0009645159652774e-05,
      "loss": 0.0488,
      "step": 29520
    },
    {
      "epoch": 1.4988070460429463,
      "grad_norm": 0.44190675020217896,
      "learning_rate": 1.000795302638036e-05,
      "loss": 0.048,
      "step": 29525
    },
    {
      "epoch": 1.4990608660338087,
      "grad_norm": 0.5355786681175232,
      "learning_rate": 1.0006260893107941e-05,
      "loss": 0.0464,
      "step": 29530
    },
    {
      "epoch": 1.4993146860246713,
      "grad_norm": 0.27352702617645264,
      "learning_rate": 1.0004568759835525e-05,
      "loss": 0.0422,
      "step": 29535
    },
    {
      "epoch": 1.4995685060155337,
      "grad_norm": 0.3291056156158447,
      "learning_rate": 1.000287662656311e-05,
      "loss": 0.0498,
      "step": 29540
    },
    {
      "epoch": 1.4998223260063963,
      "grad_norm": 0.35799673199653625,
      "learning_rate": 1.0001184493290692e-05,
      "loss": 0.0427,
      "step": 29545
    },
    {
      "epoch": 1.5000761459972587,
      "grad_norm": 0.33578240871429443,
      "learning_rate": 9.999492360018276e-06,
      "loss": 0.046,
      "step": 29550
    },
    {
      "epoch": 1.5003299659881213,
      "grad_norm": 0.27118977904319763,
      "learning_rate": 9.997800226745859e-06,
      "loss": 0.0354,
      "step": 29555
    },
    {
      "epoch": 1.5005837859789837,
      "grad_norm": 0.5098280906677246,
      "learning_rate": 9.996108093473443e-06,
      "loss": 0.051,
      "step": 29560
    },
    {
      "epoch": 1.500837605969846,
      "grad_norm": 0.34415581822395325,
      "learning_rate": 9.994415960201026e-06,
      "loss": 0.0428,
      "step": 29565
    },
    {
      "epoch": 1.5010914259607087,
      "grad_norm": 0.2864645719528198,
      "learning_rate": 9.99272382692861e-06,
      "loss": 0.0371,
      "step": 29570
    },
    {
      "epoch": 1.5013452459515713,
      "grad_norm": 0.316386878490448,
      "learning_rate": 9.991031693656193e-06,
      "loss": 0.0451,
      "step": 29575
    },
    {
      "epoch": 1.5015990659424336,
      "grad_norm": 0.3148564100265503,
      "learning_rate": 9.989339560383777e-06,
      "loss": 0.0447,
      "step": 29580
    },
    {
      "epoch": 1.501852885933296,
      "grad_norm": 0.28054192662239075,
      "learning_rate": 9.98764742711136e-06,
      "loss": 0.0425,
      "step": 29585
    },
    {
      "epoch": 1.5021067059241586,
      "grad_norm": 0.34092408418655396,
      "learning_rate": 9.985955293838944e-06,
      "loss": 0.0434,
      "step": 29590
    },
    {
      "epoch": 1.5023605259150212,
      "grad_norm": 0.3001309633255005,
      "learning_rate": 9.984263160566527e-06,
      "loss": 0.0404,
      "step": 29595
    },
    {
      "epoch": 1.5026143459058834,
      "grad_norm": 0.36474132537841797,
      "learning_rate": 9.98257102729411e-06,
      "loss": 0.0449,
      "step": 29600
    },
    {
      "epoch": 1.502868165896746,
      "grad_norm": 0.35133421421051025,
      "learning_rate": 9.980878894021695e-06,
      "loss": 0.0585,
      "step": 29605
    },
    {
      "epoch": 1.5031219858876086,
      "grad_norm": 0.4537728726863861,
      "learning_rate": 9.979186760749278e-06,
      "loss": 0.0371,
      "step": 29610
    },
    {
      "epoch": 1.503375805878471,
      "grad_norm": 0.2741606831550598,
      "learning_rate": 9.977494627476862e-06,
      "loss": 0.0378,
      "step": 29615
    },
    {
      "epoch": 1.5036296258693334,
      "grad_norm": 0.30056142807006836,
      "learning_rate": 9.975802494204443e-06,
      "loss": 0.0387,
      "step": 29620
    },
    {
      "epoch": 1.503883445860196,
      "grad_norm": 0.34249797463417053,
      "learning_rate": 9.974110360932027e-06,
      "loss": 0.042,
      "step": 29625
    },
    {
      "epoch": 1.5041372658510586,
      "grad_norm": 0.2840827703475952,
      "learning_rate": 9.972418227659612e-06,
      "loss": 0.0473,
      "step": 29630
    },
    {
      "epoch": 1.504391085841921,
      "grad_norm": 0.40764328837394714,
      "learning_rate": 9.970726094387196e-06,
      "loss": 0.044,
      "step": 29635
    },
    {
      "epoch": 1.5046449058327833,
      "grad_norm": 0.4199784994125366,
      "learning_rate": 9.969033961114778e-06,
      "loss": 0.0411,
      "step": 29640
    },
    {
      "epoch": 1.504898725823646,
      "grad_norm": 0.33828359842300415,
      "learning_rate": 9.967341827842361e-06,
      "loss": 0.0427,
      "step": 29645
    },
    {
      "epoch": 1.5051525458145083,
      "grad_norm": 0.3699682056903839,
      "learning_rate": 9.965649694569945e-06,
      "loss": 0.0549,
      "step": 29650
    },
    {
      "epoch": 1.5054063658053707,
      "grad_norm": 0.41304442286491394,
      "learning_rate": 9.963957561297528e-06,
      "loss": 0.0498,
      "step": 29655
    },
    {
      "epoch": 1.5056601857962333,
      "grad_norm": 0.46015387773513794,
      "learning_rate": 9.962265428025112e-06,
      "loss": 0.0422,
      "step": 29660
    },
    {
      "epoch": 1.5059140057870959,
      "grad_norm": 0.28946584463119507,
      "learning_rate": 9.960573294752695e-06,
      "loss": 0.0441,
      "step": 29665
    },
    {
      "epoch": 1.5061678257779583,
      "grad_norm": 0.592179536819458,
      "learning_rate": 9.958881161480279e-06,
      "loss": 0.0466,
      "step": 29670
    },
    {
      "epoch": 1.5064216457688206,
      "grad_norm": 0.31929734349250793,
      "learning_rate": 9.957189028207862e-06,
      "loss": 0.04,
      "step": 29675
    },
    {
      "epoch": 1.5066754657596833,
      "grad_norm": 0.5968852043151855,
      "learning_rate": 9.955496894935446e-06,
      "loss": 0.0539,
      "step": 29680
    },
    {
      "epoch": 1.5069292857505459,
      "grad_norm": 0.43727225065231323,
      "learning_rate": 9.95380476166303e-06,
      "loss": 0.0558,
      "step": 29685
    },
    {
      "epoch": 1.5071831057414082,
      "grad_norm": 0.2910405397415161,
      "learning_rate": 9.952112628390613e-06,
      "loss": 0.042,
      "step": 29690
    },
    {
      "epoch": 1.5074369257322706,
      "grad_norm": 0.5035144686698914,
      "learning_rate": 9.950420495118197e-06,
      "loss": 0.0495,
      "step": 29695
    },
    {
      "epoch": 1.5076907457231332,
      "grad_norm": 0.8541820645332336,
      "learning_rate": 9.94872836184578e-06,
      "loss": 0.0416,
      "step": 29700
    },
    {
      "epoch": 1.5079445657139956,
      "grad_norm": 0.49969613552093506,
      "learning_rate": 9.947036228573364e-06,
      "loss": 0.0468,
      "step": 29705
    },
    {
      "epoch": 1.508198385704858,
      "grad_norm": 0.22982028126716614,
      "learning_rate": 9.945344095300947e-06,
      "loss": 0.0418,
      "step": 29710
    },
    {
      "epoch": 1.5084522056957206,
      "grad_norm": 0.3136741816997528,
      "learning_rate": 9.943651962028529e-06,
      "loss": 0.0452,
      "step": 29715
    },
    {
      "epoch": 1.5087060256865832,
      "grad_norm": 0.2914392352104187,
      "learning_rate": 9.941959828756114e-06,
      "loss": 0.0469,
      "step": 29720
    },
    {
      "epoch": 1.5089598456774456,
      "grad_norm": 0.390132337808609,
      "learning_rate": 9.940267695483698e-06,
      "loss": 0.0549,
      "step": 29725
    },
    {
      "epoch": 1.509213665668308,
      "grad_norm": 0.4769875109195709,
      "learning_rate": 9.93857556221128e-06,
      "loss": 0.0404,
      "step": 29730
    },
    {
      "epoch": 1.5094674856591705,
      "grad_norm": 0.4255264401435852,
      "learning_rate": 9.936883428938863e-06,
      "loss": 0.0423,
      "step": 29735
    },
    {
      "epoch": 1.5097213056500332,
      "grad_norm": 0.621986448764801,
      "learning_rate": 9.935191295666447e-06,
      "loss": 0.043,
      "step": 29740
    },
    {
      "epoch": 1.5099751256408953,
      "grad_norm": 0.47021618485450745,
      "learning_rate": 9.933499162394032e-06,
      "loss": 0.0372,
      "step": 29745
    },
    {
      "epoch": 1.510228945631758,
      "grad_norm": 0.5070352554321289,
      "learning_rate": 9.931807029121614e-06,
      "loss": 0.0458,
      "step": 29750
    },
    {
      "epoch": 1.5104827656226205,
      "grad_norm": 0.357382595539093,
      "learning_rate": 9.930114895849197e-06,
      "loss": 0.0444,
      "step": 29755
    },
    {
      "epoch": 1.510736585613483,
      "grad_norm": 0.303201287984848,
      "learning_rate": 9.928422762576781e-06,
      "loss": 0.0433,
      "step": 29760
    },
    {
      "epoch": 1.5109904056043453,
      "grad_norm": 0.4169237017631531,
      "learning_rate": 9.926730629304365e-06,
      "loss": 0.0507,
      "step": 29765
    },
    {
      "epoch": 1.5112442255952079,
      "grad_norm": 0.4273361265659332,
      "learning_rate": 9.925038496031948e-06,
      "loss": 0.047,
      "step": 29770
    },
    {
      "epoch": 1.5114980455860705,
      "grad_norm": 0.3794945180416107,
      "learning_rate": 9.923346362759532e-06,
      "loss": 0.0432,
      "step": 29775
    },
    {
      "epoch": 1.5117518655769329,
      "grad_norm": 0.29941776394844055,
      "learning_rate": 9.921654229487115e-06,
      "loss": 0.0384,
      "step": 29780
    },
    {
      "epoch": 1.5120056855677952,
      "grad_norm": 0.4585740566253662,
      "learning_rate": 9.919962096214699e-06,
      "loss": 0.0493,
      "step": 29785
    },
    {
      "epoch": 1.5122595055586578,
      "grad_norm": 0.30415454506874084,
      "learning_rate": 9.918269962942282e-06,
      "loss": 0.0423,
      "step": 29790
    },
    {
      "epoch": 1.5125133255495202,
      "grad_norm": 0.2759764790534973,
      "learning_rate": 9.916577829669866e-06,
      "loss": 0.0414,
      "step": 29795
    },
    {
      "epoch": 1.5127671455403826,
      "grad_norm": 0.3170830011367798,
      "learning_rate": 9.91488569639745e-06,
      "loss": 0.043,
      "step": 29800
    },
    {
      "epoch": 1.5130209655312452,
      "grad_norm": 0.3159930408000946,
      "learning_rate": 9.913193563125031e-06,
      "loss": 0.0419,
      "step": 29805
    },
    {
      "epoch": 1.5132747855221078,
      "grad_norm": 0.35223549604415894,
      "learning_rate": 9.911501429852616e-06,
      "loss": 0.044,
      "step": 29810
    },
    {
      "epoch": 1.5135286055129702,
      "grad_norm": 0.2940140962600708,
      "learning_rate": 9.9098092965802e-06,
      "loss": 0.0506,
      "step": 29815
    },
    {
      "epoch": 1.5137824255038326,
      "grad_norm": 0.44202086329460144,
      "learning_rate": 9.908117163307784e-06,
      "loss": 0.0503,
      "step": 29820
    },
    {
      "epoch": 1.5140362454946952,
      "grad_norm": 0.37494203448295593,
      "learning_rate": 9.906425030035365e-06,
      "loss": 0.0557,
      "step": 29825
    },
    {
      "epoch": 1.5142900654855578,
      "grad_norm": 0.34451350569725037,
      "learning_rate": 9.904732896762949e-06,
      "loss": 0.0382,
      "step": 29830
    },
    {
      "epoch": 1.5145438854764202,
      "grad_norm": 0.35963883996009827,
      "learning_rate": 9.903040763490534e-06,
      "loss": 0.0482,
      "step": 29835
    },
    {
      "epoch": 1.5147977054672825,
      "grad_norm": 0.3412439525127411,
      "learning_rate": 9.901348630218118e-06,
      "loss": 0.042,
      "step": 29840
    },
    {
      "epoch": 1.5150515254581451,
      "grad_norm": 0.32426542043685913,
      "learning_rate": 9.8996564969457e-06,
      "loss": 0.045,
      "step": 29845
    },
    {
      "epoch": 1.5153053454490075,
      "grad_norm": 0.2650112807750702,
      "learning_rate": 9.897964363673283e-06,
      "loss": 0.0443,
      "step": 29850
    },
    {
      "epoch": 1.51555916543987,
      "grad_norm": 0.38973918557167053,
      "learning_rate": 9.896272230400867e-06,
      "loss": 0.0402,
      "step": 29855
    },
    {
      "epoch": 1.5158129854307325,
      "grad_norm": 0.49945682287216187,
      "learning_rate": 9.894580097128452e-06,
      "loss": 0.045,
      "step": 29860
    },
    {
      "epoch": 1.516066805421595,
      "grad_norm": 0.283811092376709,
      "learning_rate": 9.892887963856034e-06,
      "loss": 0.034,
      "step": 29865
    },
    {
      "epoch": 1.5163206254124575,
      "grad_norm": 0.381591260433197,
      "learning_rate": 9.891195830583617e-06,
      "loss": 0.0445,
      "step": 29870
    },
    {
      "epoch": 1.5165744454033199,
      "grad_norm": 0.43761926889419556,
      "learning_rate": 9.889503697311201e-06,
      "loss": 0.051,
      "step": 29875
    },
    {
      "epoch": 1.5168282653941825,
      "grad_norm": 0.3689984381198883,
      "learning_rate": 9.887811564038784e-06,
      "loss": 0.0456,
      "step": 29880
    },
    {
      "epoch": 1.517082085385045,
      "grad_norm": 0.41369885206222534,
      "learning_rate": 9.886119430766368e-06,
      "loss": 0.0478,
      "step": 29885
    },
    {
      "epoch": 1.5173359053759075,
      "grad_norm": 0.3490171730518341,
      "learning_rate": 9.884427297493951e-06,
      "loss": 0.0467,
      "step": 29890
    },
    {
      "epoch": 1.5175897253667698,
      "grad_norm": 0.41968002915382385,
      "learning_rate": 9.882735164221535e-06,
      "loss": 0.052,
      "step": 29895
    },
    {
      "epoch": 1.5178435453576324,
      "grad_norm": 0.3344227075576782,
      "learning_rate": 9.881043030949119e-06,
      "loss": 0.0497,
      "step": 29900
    },
    {
      "epoch": 1.5180973653484948,
      "grad_norm": 0.2427772432565689,
      "learning_rate": 9.879350897676702e-06,
      "loss": 0.0369,
      "step": 29905
    },
    {
      "epoch": 1.5183511853393572,
      "grad_norm": 0.2409926950931549,
      "learning_rate": 9.877658764404286e-06,
      "loss": 0.0416,
      "step": 29910
    },
    {
      "epoch": 1.5186050053302198,
      "grad_norm": 0.31125301122665405,
      "learning_rate": 9.87596663113187e-06,
      "loss": 0.0419,
      "step": 29915
    },
    {
      "epoch": 1.5188588253210824,
      "grad_norm": 0.3816099464893341,
      "learning_rate": 9.874274497859451e-06,
      "loss": 0.0452,
      "step": 29920
    },
    {
      "epoch": 1.5191126453119448,
      "grad_norm": 0.2479274719953537,
      "learning_rate": 9.872582364587036e-06,
      "loss": 0.0487,
      "step": 29925
    },
    {
      "epoch": 1.5193664653028072,
      "grad_norm": 0.4390551447868347,
      "learning_rate": 9.87089023131462e-06,
      "loss": 0.0506,
      "step": 29930
    },
    {
      "epoch": 1.5196202852936698,
      "grad_norm": 0.36978304386138916,
      "learning_rate": 9.869198098042203e-06,
      "loss": 0.0438,
      "step": 29935
    },
    {
      "epoch": 1.5198741052845324,
      "grad_norm": 0.35317033529281616,
      "learning_rate": 9.867505964769785e-06,
      "loss": 0.0395,
      "step": 29940
    },
    {
      "epoch": 1.5201279252753945,
      "grad_norm": 0.3761212229728699,
      "learning_rate": 9.865813831497369e-06,
      "loss": 0.0505,
      "step": 29945
    },
    {
      "epoch": 1.5203817452662571,
      "grad_norm": 0.3977470099925995,
      "learning_rate": 9.864121698224954e-06,
      "loss": 0.051,
      "step": 29950
    },
    {
      "epoch": 1.5206355652571197,
      "grad_norm": 0.43119311332702637,
      "learning_rate": 9.862429564952538e-06,
      "loss": 0.0485,
      "step": 29955
    },
    {
      "epoch": 1.5208893852479821,
      "grad_norm": 0.3914470076560974,
      "learning_rate": 9.86073743168012e-06,
      "loss": 0.0425,
      "step": 29960
    },
    {
      "epoch": 1.5211432052388445,
      "grad_norm": 0.30373236536979675,
      "learning_rate": 9.859045298407703e-06,
      "loss": 0.0473,
      "step": 29965
    },
    {
      "epoch": 1.521397025229707,
      "grad_norm": 1.0446465015411377,
      "learning_rate": 9.857353165135287e-06,
      "loss": 0.0439,
      "step": 29970
    },
    {
      "epoch": 1.5216508452205697,
      "grad_norm": 0.5108188986778259,
      "learning_rate": 9.85566103186287e-06,
      "loss": 0.0432,
      "step": 29975
    },
    {
      "epoch": 1.521904665211432,
      "grad_norm": 0.257059782743454,
      "learning_rate": 9.853968898590454e-06,
      "loss": 0.0403,
      "step": 29980
    },
    {
      "epoch": 1.5221584852022945,
      "grad_norm": 0.3410789370536804,
      "learning_rate": 9.852276765318037e-06,
      "loss": 0.0456,
      "step": 29985
    },
    {
      "epoch": 1.522412305193157,
      "grad_norm": 0.28535956144332886,
      "learning_rate": 9.85058463204562e-06,
      "loss": 0.041,
      "step": 29990
    },
    {
      "epoch": 1.5226661251840194,
      "grad_norm": 0.3559688627719879,
      "learning_rate": 9.848892498773204e-06,
      "loss": 0.0401,
      "step": 29995
    },
    {
      "epoch": 1.5229199451748818,
      "grad_norm": 0.3361237943172455,
      "learning_rate": 9.847200365500788e-06,
      "loss": 0.0492,
      "step": 30000
    },
    {
      "epoch": 1.5231737651657444,
      "grad_norm": 0.535846471786499,
      "learning_rate": 9.845508232228371e-06,
      "loss": 0.0549,
      "step": 30005
    },
    {
      "epoch": 1.523427585156607,
      "grad_norm": 0.4088345170021057,
      "learning_rate": 9.843816098955955e-06,
      "loss": 0.0447,
      "step": 30010
    },
    {
      "epoch": 1.5236814051474694,
      "grad_norm": 0.36553964018821716,
      "learning_rate": 9.842123965683538e-06,
      "loss": 0.0458,
      "step": 30015
    },
    {
      "epoch": 1.5239352251383318,
      "grad_norm": 0.4269677400588989,
      "learning_rate": 9.840431832411122e-06,
      "loss": 0.043,
      "step": 30020
    },
    {
      "epoch": 1.5241890451291944,
      "grad_norm": 0.2518591284751892,
      "learning_rate": 9.838739699138705e-06,
      "loss": 0.0414,
      "step": 30025
    },
    {
      "epoch": 1.524442865120057,
      "grad_norm": 0.2876974940299988,
      "learning_rate": 9.837047565866289e-06,
      "loss": 0.039,
      "step": 30030
    },
    {
      "epoch": 1.5246966851109194,
      "grad_norm": 0.40843939781188965,
      "learning_rate": 9.835355432593871e-06,
      "loss": 0.0465,
      "step": 30035
    },
    {
      "epoch": 1.5249505051017818,
      "grad_norm": 1.2598471641540527,
      "learning_rate": 9.833663299321456e-06,
      "loss": 0.04,
      "step": 30040
    },
    {
      "epoch": 1.5252043250926444,
      "grad_norm": 0.39028775691986084,
      "learning_rate": 9.83197116604904e-06,
      "loss": 0.0524,
      "step": 30045
    },
    {
      "epoch": 1.5254581450835067,
      "grad_norm": 0.5589412450790405,
      "learning_rate": 9.830279032776622e-06,
      "loss": 0.0535,
      "step": 30050
    },
    {
      "epoch": 1.5257119650743691,
      "grad_norm": 0.3849193751811981,
      "learning_rate": 9.828586899504205e-06,
      "loss": 0.0482,
      "step": 30055
    },
    {
      "epoch": 1.5259657850652317,
      "grad_norm": 0.3056574761867523,
      "learning_rate": 9.826894766231789e-06,
      "loss": 0.0426,
      "step": 30060
    },
    {
      "epoch": 1.5262196050560943,
      "grad_norm": 0.43994876742362976,
      "learning_rate": 9.825202632959374e-06,
      "loss": 0.0424,
      "step": 30065
    },
    {
      "epoch": 1.5264734250469567,
      "grad_norm": 0.33953234553337097,
      "learning_rate": 9.823510499686956e-06,
      "loss": 0.0438,
      "step": 30070
    },
    {
      "epoch": 1.526727245037819,
      "grad_norm": 0.4996977746486664,
      "learning_rate": 9.82181836641454e-06,
      "loss": 0.0487,
      "step": 30075
    },
    {
      "epoch": 1.5269810650286817,
      "grad_norm": 0.4753670394420624,
      "learning_rate": 9.820126233142123e-06,
      "loss": 0.0401,
      "step": 30080
    },
    {
      "epoch": 1.5272348850195443,
      "grad_norm": 0.3781532645225525,
      "learning_rate": 9.818434099869706e-06,
      "loss": 0.0457,
      "step": 30085
    },
    {
      "epoch": 1.5274887050104065,
      "grad_norm": 0.4716370105743408,
      "learning_rate": 9.81674196659729e-06,
      "loss": 0.0457,
      "step": 30090
    },
    {
      "epoch": 1.527742525001269,
      "grad_norm": 0.2860237658023834,
      "learning_rate": 9.815049833324873e-06,
      "loss": 0.0379,
      "step": 30095
    },
    {
      "epoch": 1.5279963449921317,
      "grad_norm": 0.5502962470054626,
      "learning_rate": 9.813357700052457e-06,
      "loss": 0.0441,
      "step": 30100
    },
    {
      "epoch": 1.528250164982994,
      "grad_norm": 0.32045990228652954,
      "learning_rate": 9.81166556678004e-06,
      "loss": 0.0384,
      "step": 30105
    },
    {
      "epoch": 1.5285039849738564,
      "grad_norm": 0.2748587727546692,
      "learning_rate": 9.809973433507624e-06,
      "loss": 0.0428,
      "step": 30110
    },
    {
      "epoch": 1.528757804964719,
      "grad_norm": 0.3589244484901428,
      "learning_rate": 9.808281300235208e-06,
      "loss": 0.0423,
      "step": 30115
    },
    {
      "epoch": 1.5290116249555816,
      "grad_norm": 0.30734512209892273,
      "learning_rate": 9.806589166962791e-06,
      "loss": 0.0368,
      "step": 30120
    },
    {
      "epoch": 1.529265444946444,
      "grad_norm": 0.5760895013809204,
      "learning_rate": 9.804897033690373e-06,
      "loss": 0.0494,
      "step": 30125
    },
    {
      "epoch": 1.5295192649373064,
      "grad_norm": 0.3356100916862488,
      "learning_rate": 9.803204900417958e-06,
      "loss": 0.0465,
      "step": 30130
    },
    {
      "epoch": 1.529773084928169,
      "grad_norm": 0.7250293493270874,
      "learning_rate": 9.801512767145542e-06,
      "loss": 0.0421,
      "step": 30135
    },
    {
      "epoch": 1.5300269049190314,
      "grad_norm": 0.22375506162643433,
      "learning_rate": 9.799820633873125e-06,
      "loss": 0.0389,
      "step": 30140
    },
    {
      "epoch": 1.5302807249098938,
      "grad_norm": 0.5991578102111816,
      "learning_rate": 9.798128500600707e-06,
      "loss": 0.043,
      "step": 30145
    },
    {
      "epoch": 1.5305345449007564,
      "grad_norm": 0.3490102291107178,
      "learning_rate": 9.79643636732829e-06,
      "loss": 0.0413,
      "step": 30150
    },
    {
      "epoch": 1.530788364891619,
      "grad_norm": 0.3767857551574707,
      "learning_rate": 9.794744234055876e-06,
      "loss": 0.0541,
      "step": 30155
    },
    {
      "epoch": 1.5310421848824813,
      "grad_norm": 0.6094956994056702,
      "learning_rate": 9.79305210078346e-06,
      "loss": 0.041,
      "step": 30160
    },
    {
      "epoch": 1.5312960048733437,
      "grad_norm": 0.6001025438308716,
      "learning_rate": 9.791359967511041e-06,
      "loss": 0.0453,
      "step": 30165
    },
    {
      "epoch": 1.5315498248642063,
      "grad_norm": 0.36103636026382446,
      "learning_rate": 9.789667834238625e-06,
      "loss": 0.0466,
      "step": 30170
    },
    {
      "epoch": 1.531803644855069,
      "grad_norm": 0.48035770654678345,
      "learning_rate": 9.787975700966208e-06,
      "loss": 0.045,
      "step": 30175
    },
    {
      "epoch": 1.5320574648459313,
      "grad_norm": 0.43839773535728455,
      "learning_rate": 9.786283567693794e-06,
      "loss": 0.0489,
      "step": 30180
    },
    {
      "epoch": 1.5323112848367937,
      "grad_norm": 0.7059385776519775,
      "learning_rate": 9.784591434421376e-06,
      "loss": 0.0422,
      "step": 30185
    },
    {
      "epoch": 1.5325651048276563,
      "grad_norm": 0.3195935785770416,
      "learning_rate": 9.782899301148959e-06,
      "loss": 0.0413,
      "step": 30190
    },
    {
      "epoch": 1.5328189248185187,
      "grad_norm": 0.25066328048706055,
      "learning_rate": 9.781207167876543e-06,
      "loss": 0.0379,
      "step": 30195
    },
    {
      "epoch": 1.533072744809381,
      "grad_norm": 0.3572368025779724,
      "learning_rate": 9.779515034604126e-06,
      "loss": 0.0465,
      "step": 30200
    },
    {
      "epoch": 1.5333265648002437,
      "grad_norm": 0.21863994002342224,
      "learning_rate": 9.77782290133171e-06,
      "loss": 0.0421,
      "step": 30205
    },
    {
      "epoch": 1.5335803847911063,
      "grad_norm": 0.427107036113739,
      "learning_rate": 9.776130768059293e-06,
      "loss": 0.0446,
      "step": 30210
    },
    {
      "epoch": 1.5338342047819686,
      "grad_norm": 0.2890668213367462,
      "learning_rate": 9.774438634786877e-06,
      "loss": 0.0491,
      "step": 30215
    },
    {
      "epoch": 1.534088024772831,
      "grad_norm": 0.34447863698005676,
      "learning_rate": 9.77274650151446e-06,
      "loss": 0.0369,
      "step": 30220
    },
    {
      "epoch": 1.5343418447636936,
      "grad_norm": 0.3056979775428772,
      "learning_rate": 9.771054368242044e-06,
      "loss": 0.0544,
      "step": 30225
    },
    {
      "epoch": 1.5345956647545562,
      "grad_norm": 0.27630293369293213,
      "learning_rate": 9.769362234969627e-06,
      "loss": 0.0385,
      "step": 30230
    },
    {
      "epoch": 1.5348494847454186,
      "grad_norm": 0.42689627408981323,
      "learning_rate": 9.767670101697211e-06,
      "loss": 0.0424,
      "step": 30235
    },
    {
      "epoch": 1.535103304736281,
      "grad_norm": 0.40753817558288574,
      "learning_rate": 9.765977968424793e-06,
      "loss": 0.0568,
      "step": 30240
    },
    {
      "epoch": 1.5353571247271436,
      "grad_norm": 0.2552444040775299,
      "learning_rate": 9.764285835152378e-06,
      "loss": 0.0437,
      "step": 30245
    },
    {
      "epoch": 1.535610944718006,
      "grad_norm": 0.27542778849601746,
      "learning_rate": 9.762593701879962e-06,
      "loss": 0.0397,
      "step": 30250
    },
    {
      "epoch": 1.5358647647088683,
      "grad_norm": 0.257028728723526,
      "learning_rate": 9.760901568607545e-06,
      "loss": 0.043,
      "step": 30255
    },
    {
      "epoch": 1.536118584699731,
      "grad_norm": 0.7336534857749939,
      "learning_rate": 9.759209435335127e-06,
      "loss": 0.0448,
      "step": 30260
    },
    {
      "epoch": 1.5363724046905936,
      "grad_norm": 0.7926486134529114,
      "learning_rate": 9.75751730206271e-06,
      "loss": 0.0473,
      "step": 30265
    },
    {
      "epoch": 1.536626224681456,
      "grad_norm": 0.37469780445098877,
      "learning_rate": 9.755825168790296e-06,
      "loss": 0.0434,
      "step": 30270
    },
    {
      "epoch": 1.5368800446723183,
      "grad_norm": 0.2389148771762848,
      "learning_rate": 9.75413303551788e-06,
      "loss": 0.0328,
      "step": 30275
    },
    {
      "epoch": 1.537133864663181,
      "grad_norm": 0.18880969285964966,
      "learning_rate": 9.752440902245461e-06,
      "loss": 0.0333,
      "step": 30280
    },
    {
      "epoch": 1.5373876846540435,
      "grad_norm": 0.29365938901901245,
      "learning_rate": 9.750748768973045e-06,
      "loss": 0.0365,
      "step": 30285
    },
    {
      "epoch": 1.5376415046449057,
      "grad_norm": 0.4020933508872986,
      "learning_rate": 9.749056635700628e-06,
      "loss": 0.0451,
      "step": 30290
    },
    {
      "epoch": 1.5378953246357683,
      "grad_norm": 0.3207293748855591,
      "learning_rate": 9.747364502428212e-06,
      "loss": 0.0465,
      "step": 30295
    },
    {
      "epoch": 1.5381491446266309,
      "grad_norm": 0.5186519026756287,
      "learning_rate": 9.745672369155795e-06,
      "loss": 0.0478,
      "step": 30300
    },
    {
      "epoch": 1.5384029646174933,
      "grad_norm": 0.657397985458374,
      "learning_rate": 9.743980235883379e-06,
      "loss": 0.0366,
      "step": 30305
    },
    {
      "epoch": 1.5386567846083556,
      "grad_norm": 0.3749079704284668,
      "learning_rate": 9.742288102610962e-06,
      "loss": 0.0419,
      "step": 30310
    },
    {
      "epoch": 1.5389106045992182,
      "grad_norm": 0.47508054971694946,
      "learning_rate": 9.740595969338546e-06,
      "loss": 0.0492,
      "step": 30315
    },
    {
      "epoch": 1.5391644245900808,
      "grad_norm": 0.3623720109462738,
      "learning_rate": 9.73890383606613e-06,
      "loss": 0.0413,
      "step": 30320
    },
    {
      "epoch": 1.5394182445809432,
      "grad_norm": 0.32053571939468384,
      "learning_rate": 9.737211702793713e-06,
      "loss": 0.0432,
      "step": 30325
    },
    {
      "epoch": 1.5396720645718056,
      "grad_norm": 0.3546943962574005,
      "learning_rate": 9.735519569521297e-06,
      "loss": 0.0487,
      "step": 30330
    },
    {
      "epoch": 1.5399258845626682,
      "grad_norm": 0.3554565906524658,
      "learning_rate": 9.73382743624888e-06,
      "loss": 0.0424,
      "step": 30335
    },
    {
      "epoch": 1.5401797045535306,
      "grad_norm": 0.44860467314720154,
      "learning_rate": 9.732135302976464e-06,
      "loss": 0.0416,
      "step": 30340
    },
    {
      "epoch": 1.540433524544393,
      "grad_norm": 0.28419896960258484,
      "learning_rate": 9.730443169704047e-06,
      "loss": 0.0422,
      "step": 30345
    },
    {
      "epoch": 1.5406873445352556,
      "grad_norm": 0.2352607697248459,
      "learning_rate": 9.72875103643163e-06,
      "loss": 0.0397,
      "step": 30350
    },
    {
      "epoch": 1.5409411645261182,
      "grad_norm": 0.348967969417572,
      "learning_rate": 9.727058903159213e-06,
      "loss": 0.0443,
      "step": 30355
    },
    {
      "epoch": 1.5411949845169806,
      "grad_norm": 0.5574663281440735,
      "learning_rate": 9.725366769886798e-06,
      "loss": 0.0476,
      "step": 30360
    },
    {
      "epoch": 1.541448804507843,
      "grad_norm": 0.3044932186603546,
      "learning_rate": 9.723674636614381e-06,
      "loss": 0.0497,
      "step": 30365
    },
    {
      "epoch": 1.5417026244987055,
      "grad_norm": 0.3024437427520752,
      "learning_rate": 9.721982503341963e-06,
      "loss": 0.0392,
      "step": 30370
    },
    {
      "epoch": 1.5419564444895681,
      "grad_norm": 0.9230430126190186,
      "learning_rate": 9.720290370069547e-06,
      "loss": 0.0457,
      "step": 30375
    },
    {
      "epoch": 1.5422102644804305,
      "grad_norm": 0.29543060064315796,
      "learning_rate": 9.71859823679713e-06,
      "loss": 0.0407,
      "step": 30380
    },
    {
      "epoch": 1.542464084471293,
      "grad_norm": 0.3375641107559204,
      "learning_rate": 9.716906103524714e-06,
      "loss": 0.04,
      "step": 30385
    },
    {
      "epoch": 1.5427179044621555,
      "grad_norm": 0.31575915217399597,
      "learning_rate": 9.715213970252297e-06,
      "loss": 0.049,
      "step": 30390
    },
    {
      "epoch": 1.542971724453018,
      "grad_norm": 0.2939968705177307,
      "learning_rate": 9.713521836979881e-06,
      "loss": 0.0469,
      "step": 30395
    },
    {
      "epoch": 1.5432255444438803,
      "grad_norm": 0.47484177350997925,
      "learning_rate": 9.711829703707465e-06,
      "loss": 0.0539,
      "step": 30400
    },
    {
      "epoch": 1.5434793644347429,
      "grad_norm": 0.29723361134529114,
      "learning_rate": 9.710137570435048e-06,
      "loss": 0.0388,
      "step": 30405
    },
    {
      "epoch": 1.5437331844256055,
      "grad_norm": 0.6943442225456238,
      "learning_rate": 9.708445437162632e-06,
      "loss": 0.0447,
      "step": 30410
    },
    {
      "epoch": 1.5439870044164679,
      "grad_norm": 0.3720008134841919,
      "learning_rate": 9.706753303890215e-06,
      "loss": 0.0427,
      "step": 30415
    },
    {
      "epoch": 1.5442408244073302,
      "grad_norm": 0.2879853844642639,
      "learning_rate": 9.705061170617799e-06,
      "loss": 0.047,
      "step": 30420
    },
    {
      "epoch": 1.5444946443981928,
      "grad_norm": 0.5231357216835022,
      "learning_rate": 9.703369037345382e-06,
      "loss": 0.0394,
      "step": 30425
    },
    {
      "epoch": 1.5447484643890554,
      "grad_norm": 0.33853334188461304,
      "learning_rate": 9.701676904072966e-06,
      "loss": 0.0488,
      "step": 30430
    },
    {
      "epoch": 1.5450022843799176,
      "grad_norm": 0.5183348655700684,
      "learning_rate": 9.69998477080055e-06,
      "loss": 0.0382,
      "step": 30435
    },
    {
      "epoch": 1.5452561043707802,
      "grad_norm": 0.34625086188316345,
      "learning_rate": 9.698292637528133e-06,
      "loss": 0.038,
      "step": 30440
    },
    {
      "epoch": 1.5455099243616428,
      "grad_norm": 0.5586714744567871,
      "learning_rate": 9.696600504255715e-06,
      "loss": 0.0367,
      "step": 30445
    },
    {
      "epoch": 1.5457637443525052,
      "grad_norm": 0.7301913499832153,
      "learning_rate": 9.6949083709833e-06,
      "loss": 0.0479,
      "step": 30450
    },
    {
      "epoch": 1.5460175643433676,
      "grad_norm": 0.5144783854484558,
      "learning_rate": 9.693216237710884e-06,
      "loss": 0.0416,
      "step": 30455
    },
    {
      "epoch": 1.5462713843342302,
      "grad_norm": 0.33749499917030334,
      "learning_rate": 9.691524104438467e-06,
      "loss": 0.0465,
      "step": 30460
    },
    {
      "epoch": 1.5465252043250928,
      "grad_norm": 0.28034669160842896,
      "learning_rate": 9.689831971166049e-06,
      "loss": 0.0442,
      "step": 30465
    },
    {
      "epoch": 1.5467790243159552,
      "grad_norm": 0.47718653082847595,
      "learning_rate": 9.688139837893632e-06,
      "loss": 0.0417,
      "step": 30470
    },
    {
      "epoch": 1.5470328443068175,
      "grad_norm": 0.543878436088562,
      "learning_rate": 9.686447704621216e-06,
      "loss": 0.0499,
      "step": 30475
    },
    {
      "epoch": 1.5472866642976801,
      "grad_norm": 0.30951932072639465,
      "learning_rate": 9.684755571348801e-06,
      "loss": 0.0548,
      "step": 30480
    },
    {
      "epoch": 1.5475404842885425,
      "grad_norm": 0.39540088176727295,
      "learning_rate": 9.683063438076383e-06,
      "loss": 0.0394,
      "step": 30485
    },
    {
      "epoch": 1.547794304279405,
      "grad_norm": 0.37335509061813354,
      "learning_rate": 9.681371304803967e-06,
      "loss": 0.0466,
      "step": 30490
    },
    {
      "epoch": 1.5480481242702675,
      "grad_norm": 0.39366480708122253,
      "learning_rate": 9.67967917153155e-06,
      "loss": 0.0391,
      "step": 30495
    },
    {
      "epoch": 1.54830194426113,
      "grad_norm": 0.2801901698112488,
      "learning_rate": 9.677987038259134e-06,
      "loss": 0.0421,
      "step": 30500
    },
    {
      "epoch": 1.5485557642519925,
      "grad_norm": 0.37131232023239136,
      "learning_rate": 9.676294904986717e-06,
      "loss": 0.0496,
      "step": 30505
    },
    {
      "epoch": 1.5488095842428549,
      "grad_norm": 0.3372965455055237,
      "learning_rate": 9.674602771714301e-06,
      "loss": 0.0588,
      "step": 30510
    },
    {
      "epoch": 1.5490634042337175,
      "grad_norm": 0.8736234307289124,
      "learning_rate": 9.672910638441884e-06,
      "loss": 0.0464,
      "step": 30515
    },
    {
      "epoch": 1.54931722422458,
      "grad_norm": 0.24658441543579102,
      "learning_rate": 9.671218505169468e-06,
      "loss": 0.0338,
      "step": 30520
    },
    {
      "epoch": 1.5495710442154425,
      "grad_norm": 0.2501583397388458,
      "learning_rate": 9.669526371897051e-06,
      "loss": 0.0386,
      "step": 30525
    },
    {
      "epoch": 1.5498248642063048,
      "grad_norm": 0.3103393018245697,
      "learning_rate": 9.667834238624635e-06,
      "loss": 0.0476,
      "step": 30530
    },
    {
      "epoch": 1.5500786841971674,
      "grad_norm": 0.36764755845069885,
      "learning_rate": 9.666142105352219e-06,
      "loss": 0.0414,
      "step": 30535
    },
    {
      "epoch": 1.5503325041880298,
      "grad_norm": 0.46840208768844604,
      "learning_rate": 9.664449972079802e-06,
      "loss": 0.0451,
      "step": 30540
    },
    {
      "epoch": 1.5505863241788922,
      "grad_norm": 0.5336713194847107,
      "learning_rate": 9.662757838807386e-06,
      "loss": 0.0537,
      "step": 30545
    },
    {
      "epoch": 1.5508401441697548,
      "grad_norm": 0.4022626578807831,
      "learning_rate": 9.66106570553497e-06,
      "loss": 0.0472,
      "step": 30550
    },
    {
      "epoch": 1.5510939641606174,
      "grad_norm": 0.22053535282611847,
      "learning_rate": 9.659373572262553e-06,
      "loss": 0.052,
      "step": 30555
    },
    {
      "epoch": 1.5513477841514798,
      "grad_norm": 0.3512711226940155,
      "learning_rate": 9.657681438990135e-06,
      "loss": 0.044,
      "step": 30560
    },
    {
      "epoch": 1.5516016041423422,
      "grad_norm": 0.31947484612464905,
      "learning_rate": 9.655989305717718e-06,
      "loss": 0.0444,
      "step": 30565
    },
    {
      "epoch": 1.5518554241332048,
      "grad_norm": 0.6187324523925781,
      "learning_rate": 9.654297172445303e-06,
      "loss": 0.0499,
      "step": 30570
    },
    {
      "epoch": 1.5521092441240674,
      "grad_norm": 0.3712591230869293,
      "learning_rate": 9.652605039172887e-06,
      "loss": 0.0443,
      "step": 30575
    },
    {
      "epoch": 1.5523630641149297,
      "grad_norm": 0.6312601566314697,
      "learning_rate": 9.650912905900469e-06,
      "loss": 0.0468,
      "step": 30580
    },
    {
      "epoch": 1.5526168841057921,
      "grad_norm": 0.3091924786567688,
      "learning_rate": 9.649220772628052e-06,
      "loss": 0.0395,
      "step": 30585
    },
    {
      "epoch": 1.5528707040966547,
      "grad_norm": 0.23264305293560028,
      "learning_rate": 9.647528639355636e-06,
      "loss": 0.0401,
      "step": 30590
    },
    {
      "epoch": 1.553124524087517,
      "grad_norm": 0.4403635263442993,
      "learning_rate": 9.645836506083221e-06,
      "loss": 0.0394,
      "step": 30595
    },
    {
      "epoch": 1.5533783440783795,
      "grad_norm": 0.6066749095916748,
      "learning_rate": 9.644144372810803e-06,
      "loss": 0.0555,
      "step": 30600
    },
    {
      "epoch": 1.553632164069242,
      "grad_norm": 0.43328630924224854,
      "learning_rate": 9.642452239538387e-06,
      "loss": 0.0418,
      "step": 30605
    },
    {
      "epoch": 1.5538859840601047,
      "grad_norm": 0.3233341574668884,
      "learning_rate": 9.64076010626597e-06,
      "loss": 0.0411,
      "step": 30610
    },
    {
      "epoch": 1.554139804050967,
      "grad_norm": 0.47174957394599915,
      "learning_rate": 9.639067972993554e-06,
      "loss": 0.0457,
      "step": 30615
    },
    {
      "epoch": 1.5543936240418295,
      "grad_norm": 0.31749293208122253,
      "learning_rate": 9.637375839721137e-06,
      "loss": 0.0467,
      "step": 30620
    },
    {
      "epoch": 1.554647444032692,
      "grad_norm": 0.7397310137748718,
      "learning_rate": 9.63568370644872e-06,
      "loss": 0.0388,
      "step": 30625
    },
    {
      "epoch": 1.5549012640235544,
      "grad_norm": 0.24703024327754974,
      "learning_rate": 9.633991573176304e-06,
      "loss": 0.041,
      "step": 30630
    },
    {
      "epoch": 1.5551550840144168,
      "grad_norm": 0.3380805253982544,
      "learning_rate": 9.632299439903888e-06,
      "loss": 0.039,
      "step": 30635
    },
    {
      "epoch": 1.5554089040052794,
      "grad_norm": 0.30269336700439453,
      "learning_rate": 9.630607306631471e-06,
      "loss": 0.0409,
      "step": 30640
    },
    {
      "epoch": 1.555662723996142,
      "grad_norm": 0.20982228219509125,
      "learning_rate": 9.628915173359055e-06,
      "loss": 0.0426,
      "step": 30645
    },
    {
      "epoch": 1.5559165439870044,
      "grad_norm": 0.30065563321113586,
      "learning_rate": 9.627223040086638e-06,
      "loss": 0.0474,
      "step": 30650
    },
    {
      "epoch": 1.5561703639778668,
      "grad_norm": 0.5171536803245544,
      "learning_rate": 9.62553090681422e-06,
      "loss": 0.0457,
      "step": 30655
    },
    {
      "epoch": 1.5564241839687294,
      "grad_norm": 0.408906489610672,
      "learning_rate": 9.623838773541805e-06,
      "loss": 0.0403,
      "step": 30660
    },
    {
      "epoch": 1.556678003959592,
      "grad_norm": 0.2871289849281311,
      "learning_rate": 9.622146640269389e-06,
      "loss": 0.0333,
      "step": 30665
    },
    {
      "epoch": 1.5569318239504544,
      "grad_norm": 0.2980833053588867,
      "learning_rate": 9.620454506996973e-06,
      "loss": 0.0423,
      "step": 30670
    },
    {
      "epoch": 1.5571856439413168,
      "grad_norm": 0.38692790269851685,
      "learning_rate": 9.618762373724554e-06,
      "loss": 0.048,
      "step": 30675
    },
    {
      "epoch": 1.5574394639321794,
      "grad_norm": 0.9138740301132202,
      "learning_rate": 9.617070240452138e-06,
      "loss": 0.0453,
      "step": 30680
    },
    {
      "epoch": 1.5576932839230417,
      "grad_norm": 0.30773526430130005,
      "learning_rate": 9.615378107179723e-06,
      "loss": 0.0419,
      "step": 30685
    },
    {
      "epoch": 1.5579471039139041,
      "grad_norm": 0.4494325518608093,
      "learning_rate": 9.613685973907305e-06,
      "loss": 0.0355,
      "step": 30690
    },
    {
      "epoch": 1.5582009239047667,
      "grad_norm": 0.2837882339954376,
      "learning_rate": 9.611993840634889e-06,
      "loss": 0.0389,
      "step": 30695
    },
    {
      "epoch": 1.5584547438956293,
      "grad_norm": 0.3327257037162781,
      "learning_rate": 9.610301707362472e-06,
      "loss": 0.0414,
      "step": 30700
    },
    {
      "epoch": 1.5587085638864917,
      "grad_norm": 0.37678539752960205,
      "learning_rate": 9.608609574090056e-06,
      "loss": 0.0365,
      "step": 30705
    },
    {
      "epoch": 1.558962383877354,
      "grad_norm": 0.326312392950058,
      "learning_rate": 9.60691744081764e-06,
      "loss": 0.048,
      "step": 30710
    },
    {
      "epoch": 1.5592162038682167,
      "grad_norm": 0.4527011215686798,
      "learning_rate": 9.605225307545223e-06,
      "loss": 0.0405,
      "step": 30715
    },
    {
      "epoch": 1.5594700238590793,
      "grad_norm": 0.7663089036941528,
      "learning_rate": 9.603533174272806e-06,
      "loss": 0.0455,
      "step": 30720
    },
    {
      "epoch": 1.5597238438499417,
      "grad_norm": 0.3243597447872162,
      "learning_rate": 9.60184104100039e-06,
      "loss": 0.0414,
      "step": 30725
    },
    {
      "epoch": 1.559977663840804,
      "grad_norm": 0.40656790137290955,
      "learning_rate": 9.600148907727973e-06,
      "loss": 0.0553,
      "step": 30730
    },
    {
      "epoch": 1.5602314838316667,
      "grad_norm": 0.48976048827171326,
      "learning_rate": 9.598456774455557e-06,
      "loss": 0.0384,
      "step": 30735
    },
    {
      "epoch": 1.560485303822529,
      "grad_norm": 0.2989707589149475,
      "learning_rate": 9.59676464118314e-06,
      "loss": 0.0446,
      "step": 30740
    },
    {
      "epoch": 1.5607391238133914,
      "grad_norm": 0.3599499762058258,
      "learning_rate": 9.595072507910724e-06,
      "loss": 0.0385,
      "step": 30745
    },
    {
      "epoch": 1.560992943804254,
      "grad_norm": 0.36865293979644775,
      "learning_rate": 9.593380374638308e-06,
      "loss": 0.036,
      "step": 30750
    },
    {
      "epoch": 1.5612467637951166,
      "grad_norm": 0.41994988918304443,
      "learning_rate": 9.591688241365891e-06,
      "loss": 0.0463,
      "step": 30755
    },
    {
      "epoch": 1.561500583785979,
      "grad_norm": 0.7477067112922668,
      "learning_rate": 9.589996108093475e-06,
      "loss": 0.0407,
      "step": 30760
    },
    {
      "epoch": 1.5617544037768414,
      "grad_norm": 0.2858511805534363,
      "learning_rate": 9.588303974821058e-06,
      "loss": 0.0485,
      "step": 30765
    },
    {
      "epoch": 1.562008223767704,
      "grad_norm": 0.4207206964492798,
      "learning_rate": 9.58661184154864e-06,
      "loss": 0.0546,
      "step": 30770
    },
    {
      "epoch": 1.5622620437585666,
      "grad_norm": 0.7034492492675781,
      "learning_rate": 9.584919708276225e-06,
      "loss": 0.0439,
      "step": 30775
    },
    {
      "epoch": 1.5625158637494287,
      "grad_norm": 0.4226521849632263,
      "learning_rate": 9.583227575003809e-06,
      "loss": 0.0437,
      "step": 30780
    },
    {
      "epoch": 1.5627696837402913,
      "grad_norm": 0.37233424186706543,
      "learning_rate": 9.58153544173139e-06,
      "loss": 0.0396,
      "step": 30785
    },
    {
      "epoch": 1.563023503731154,
      "grad_norm": 0.35524532198905945,
      "learning_rate": 9.579843308458974e-06,
      "loss": 0.042,
      "step": 30790
    },
    {
      "epoch": 1.5632773237220163,
      "grad_norm": 0.24936921894550323,
      "learning_rate": 9.578151175186558e-06,
      "loss": 0.037,
      "step": 30795
    },
    {
      "epoch": 1.5635311437128787,
      "grad_norm": 0.35110658407211304,
      "learning_rate": 9.576459041914143e-06,
      "loss": 0.0443,
      "step": 30800
    },
    {
      "epoch": 1.5637849637037413,
      "grad_norm": 0.339292973279953,
      "learning_rate": 9.574766908641725e-06,
      "loss": 0.0433,
      "step": 30805
    },
    {
      "epoch": 1.564038783694604,
      "grad_norm": 0.39210009574890137,
      "learning_rate": 9.573074775369308e-06,
      "loss": 0.0385,
      "step": 30810
    },
    {
      "epoch": 1.5642926036854663,
      "grad_norm": 0.39238834381103516,
      "learning_rate": 9.571382642096892e-06,
      "loss": 0.0415,
      "step": 30815
    },
    {
      "epoch": 1.5645464236763287,
      "grad_norm": 0.4399608075618744,
      "learning_rate": 9.569690508824476e-06,
      "loss": 0.0466,
      "step": 30820
    },
    {
      "epoch": 1.5648002436671913,
      "grad_norm": 0.34421685338020325,
      "learning_rate": 9.567998375552059e-06,
      "loss": 0.0392,
      "step": 30825
    },
    {
      "epoch": 1.5650540636580537,
      "grad_norm": 0.256448894739151,
      "learning_rate": 9.566306242279643e-06,
      "loss": 0.041,
      "step": 30830
    },
    {
      "epoch": 1.565307883648916,
      "grad_norm": 0.2971365749835968,
      "learning_rate": 9.564614109007226e-06,
      "loss": 0.0469,
      "step": 30835
    },
    {
      "epoch": 1.5655617036397786,
      "grad_norm": 0.37906402349472046,
      "learning_rate": 9.56292197573481e-06,
      "loss": 0.0478,
      "step": 30840
    },
    {
      "epoch": 1.5658155236306412,
      "grad_norm": 0.35806217789649963,
      "learning_rate": 9.561229842462393e-06,
      "loss": 0.042,
      "step": 30845
    },
    {
      "epoch": 1.5660693436215036,
      "grad_norm": 0.5716193914413452,
      "learning_rate": 9.559537709189977e-06,
      "loss": 0.045,
      "step": 30850
    },
    {
      "epoch": 1.566323163612366,
      "grad_norm": 0.2948254644870758,
      "learning_rate": 9.55784557591756e-06,
      "loss": 0.0407,
      "step": 30855
    },
    {
      "epoch": 1.5665769836032286,
      "grad_norm": 0.4924793839454651,
      "learning_rate": 9.556153442645142e-06,
      "loss": 0.0405,
      "step": 30860
    },
    {
      "epoch": 1.5668308035940912,
      "grad_norm": 0.4333508610725403,
      "learning_rate": 9.554461309372727e-06,
      "loss": 0.0386,
      "step": 30865
    },
    {
      "epoch": 1.5670846235849536,
      "grad_norm": 0.35485759377479553,
      "learning_rate": 9.552769176100311e-06,
      "loss": 0.0356,
      "step": 30870
    },
    {
      "epoch": 1.567338443575816,
      "grad_norm": 0.2829851806163788,
      "learning_rate": 9.551077042827895e-06,
      "loss": 0.0469,
      "step": 30875
    },
    {
      "epoch": 1.5675922635666786,
      "grad_norm": 0.4814103841781616,
      "learning_rate": 9.549384909555476e-06,
      "loss": 0.05,
      "step": 30880
    },
    {
      "epoch": 1.567846083557541,
      "grad_norm": 0.3469924032688141,
      "learning_rate": 9.54769277628306e-06,
      "loss": 0.0434,
      "step": 30885
    },
    {
      "epoch": 1.5680999035484033,
      "grad_norm": 0.35943081974983215,
      "learning_rate": 9.546000643010645e-06,
      "loss": 0.0472,
      "step": 30890
    },
    {
      "epoch": 1.568353723539266,
      "grad_norm": 0.3924970030784607,
      "learning_rate": 9.544308509738229e-06,
      "loss": 0.0489,
      "step": 30895
    },
    {
      "epoch": 1.5686075435301285,
      "grad_norm": 0.3497081995010376,
      "learning_rate": 9.54261637646581e-06,
      "loss": 0.0456,
      "step": 30900
    },
    {
      "epoch": 1.568861363520991,
      "grad_norm": 0.3416511118412018,
      "learning_rate": 9.540924243193394e-06,
      "loss": 0.0357,
      "step": 30905
    },
    {
      "epoch": 1.5691151835118533,
      "grad_norm": 0.23118287324905396,
      "learning_rate": 9.539232109920978e-06,
      "loss": 0.0412,
      "step": 30910
    },
    {
      "epoch": 1.569369003502716,
      "grad_norm": 0.4158100485801697,
      "learning_rate": 9.537539976648563e-06,
      "loss": 0.0461,
      "step": 30915
    },
    {
      "epoch": 1.5696228234935785,
      "grad_norm": 0.38120755553245544,
      "learning_rate": 9.535847843376145e-06,
      "loss": 0.0384,
      "step": 30920
    },
    {
      "epoch": 1.5698766434844407,
      "grad_norm": 0.25962239503860474,
      "learning_rate": 9.534155710103728e-06,
      "loss": 0.0404,
      "step": 30925
    },
    {
      "epoch": 1.5701304634753033,
      "grad_norm": 0.31484225392341614,
      "learning_rate": 9.532463576831312e-06,
      "loss": 0.0326,
      "step": 30930
    },
    {
      "epoch": 1.5703842834661659,
      "grad_norm": 0.2863726317882538,
      "learning_rate": 9.530771443558895e-06,
      "loss": 0.0465,
      "step": 30935
    },
    {
      "epoch": 1.5706381034570283,
      "grad_norm": 0.3848249018192291,
      "learning_rate": 9.529079310286479e-06,
      "loss": 0.0449,
      "step": 30940
    },
    {
      "epoch": 1.5708919234478906,
      "grad_norm": 0.5639122724533081,
      "learning_rate": 9.527387177014062e-06,
      "loss": 0.0462,
      "step": 30945
    },
    {
      "epoch": 1.5711457434387532,
      "grad_norm": 0.33197957277297974,
      "learning_rate": 9.525695043741646e-06,
      "loss": 0.0353,
      "step": 30950
    },
    {
      "epoch": 1.5713995634296158,
      "grad_norm": 0.3288804292678833,
      "learning_rate": 9.52400291046923e-06,
      "loss": 0.0388,
      "step": 30955
    },
    {
      "epoch": 1.5716533834204782,
      "grad_norm": 0.3814419209957123,
      "learning_rate": 9.522310777196813e-06,
      "loss": 0.0413,
      "step": 30960
    },
    {
      "epoch": 1.5719072034113406,
      "grad_norm": 0.3379333019256592,
      "learning_rate": 9.520618643924397e-06,
      "loss": 0.0477,
      "step": 30965
    },
    {
      "epoch": 1.5721610234022032,
      "grad_norm": 0.438324898481369,
      "learning_rate": 9.51892651065198e-06,
      "loss": 0.0462,
      "step": 30970
    },
    {
      "epoch": 1.5724148433930656,
      "grad_norm": 0.3242974281311035,
      "learning_rate": 9.517234377379562e-06,
      "loss": 0.0372,
      "step": 30975
    },
    {
      "epoch": 1.572668663383928,
      "grad_norm": 0.9535584449768066,
      "learning_rate": 9.515542244107147e-06,
      "loss": 0.0476,
      "step": 30980
    },
    {
      "epoch": 1.5729224833747906,
      "grad_norm": 0.34823837876319885,
      "learning_rate": 9.51385011083473e-06,
      "loss": 0.0507,
      "step": 30985
    },
    {
      "epoch": 1.5731763033656532,
      "grad_norm": 0.21808657050132751,
      "learning_rate": 9.512157977562314e-06,
      "loss": 0.0376,
      "step": 30990
    },
    {
      "epoch": 1.5734301233565156,
      "grad_norm": 0.3237823247909546,
      "learning_rate": 9.510465844289896e-06,
      "loss": 0.0435,
      "step": 30995
    },
    {
      "epoch": 1.573683943347378,
      "grad_norm": 0.3559344708919525,
      "learning_rate": 9.50877371101748e-06,
      "loss": 0.0394,
      "step": 31000
    },
    {
      "epoch": 1.5739377633382405,
      "grad_norm": 0.43727895617485046,
      "learning_rate": 9.507081577745065e-06,
      "loss": 0.0489,
      "step": 31005
    },
    {
      "epoch": 1.5741915833291031,
      "grad_norm": 0.42054206132888794,
      "learning_rate": 9.505389444472649e-06,
      "loss": 0.0442,
      "step": 31010
    },
    {
      "epoch": 1.5744454033199655,
      "grad_norm": 0.19902174174785614,
      "learning_rate": 9.50369731120023e-06,
      "loss": 0.0359,
      "step": 31015
    },
    {
      "epoch": 1.574699223310828,
      "grad_norm": 0.5150263905525208,
      "learning_rate": 9.502005177927814e-06,
      "loss": 0.0499,
      "step": 31020
    },
    {
      "epoch": 1.5749530433016905,
      "grad_norm": 0.4221786856651306,
      "learning_rate": 9.500313044655397e-06,
      "loss": 0.0421,
      "step": 31025
    },
    {
      "epoch": 1.5752068632925529,
      "grad_norm": 0.5770427584648132,
      "learning_rate": 9.498620911382981e-06,
      "loss": 0.0383,
      "step": 31030
    },
    {
      "epoch": 1.5754606832834153,
      "grad_norm": 0.3271523416042328,
      "learning_rate": 9.496928778110565e-06,
      "loss": 0.0401,
      "step": 31035
    },
    {
      "epoch": 1.5757145032742779,
      "grad_norm": 0.33425188064575195,
      "learning_rate": 9.495236644838148e-06,
      "loss": 0.039,
      "step": 31040
    },
    {
      "epoch": 1.5759683232651405,
      "grad_norm": 0.6423659920692444,
      "learning_rate": 9.493544511565732e-06,
      "loss": 0.0469,
      "step": 31045
    },
    {
      "epoch": 1.5762221432560029,
      "grad_norm": 0.32353103160858154,
      "learning_rate": 9.491852378293315e-06,
      "loss": 0.0402,
      "step": 31050
    },
    {
      "epoch": 1.5764759632468652,
      "grad_norm": 0.446249395608902,
      "learning_rate": 9.490160245020899e-06,
      "loss": 0.0404,
      "step": 31055
    },
    {
      "epoch": 1.5767297832377278,
      "grad_norm": 0.5686723589897156,
      "learning_rate": 9.488468111748482e-06,
      "loss": 0.0351,
      "step": 31060
    },
    {
      "epoch": 1.5769836032285904,
      "grad_norm": 0.334552139043808,
      "learning_rate": 9.486775978476066e-06,
      "loss": 0.0442,
      "step": 31065
    },
    {
      "epoch": 1.5772374232194528,
      "grad_norm": 0.40195050835609436,
      "learning_rate": 9.48508384520365e-06,
      "loss": 0.0419,
      "step": 31070
    },
    {
      "epoch": 1.5774912432103152,
      "grad_norm": 0.36055585741996765,
      "learning_rate": 9.483391711931233e-06,
      "loss": 0.0442,
      "step": 31075
    },
    {
      "epoch": 1.5777450632011778,
      "grad_norm": 0.6583750247955322,
      "learning_rate": 9.481699578658816e-06,
      "loss": 0.0342,
      "step": 31080
    },
    {
      "epoch": 1.5779988831920402,
      "grad_norm": 0.3017454147338867,
      "learning_rate": 9.4800074453864e-06,
      "loss": 0.0476,
      "step": 31085
    },
    {
      "epoch": 1.5782527031829026,
      "grad_norm": 0.2620936334133148,
      "learning_rate": 9.478315312113982e-06,
      "loss": 0.0419,
      "step": 31090
    },
    {
      "epoch": 1.5785065231737652,
      "grad_norm": 0.347018301486969,
      "learning_rate": 9.476623178841567e-06,
      "loss": 0.0419,
      "step": 31095
    },
    {
      "epoch": 1.5787603431646278,
      "grad_norm": 0.3718807101249695,
      "learning_rate": 9.47493104556915e-06,
      "loss": 0.0412,
      "step": 31100
    },
    {
      "epoch": 1.5790141631554901,
      "grad_norm": 0.2230846881866455,
      "learning_rate": 9.473238912296732e-06,
      "loss": 0.0351,
      "step": 31105
    },
    {
      "epoch": 1.5792679831463525,
      "grad_norm": 0.2787439525127411,
      "learning_rate": 9.471546779024316e-06,
      "loss": 0.0469,
      "step": 31110
    },
    {
      "epoch": 1.5795218031372151,
      "grad_norm": 0.46202558279037476,
      "learning_rate": 9.4698546457519e-06,
      "loss": 0.0484,
      "step": 31115
    },
    {
      "epoch": 1.5797756231280777,
      "grad_norm": 0.3812348246574402,
      "learning_rate": 9.468162512479485e-06,
      "loss": 0.0389,
      "step": 31120
    },
    {
      "epoch": 1.58002944311894,
      "grad_norm": 0.2918323278427124,
      "learning_rate": 9.466470379207067e-06,
      "loss": 0.0517,
      "step": 31125
    },
    {
      "epoch": 1.5802832631098025,
      "grad_norm": 0.33513250946998596,
      "learning_rate": 9.46477824593465e-06,
      "loss": 0.0481,
      "step": 31130
    },
    {
      "epoch": 1.580537083100665,
      "grad_norm": 0.5008119344711304,
      "learning_rate": 9.463086112662234e-06,
      "loss": 0.0464,
      "step": 31135
    },
    {
      "epoch": 1.5807909030915275,
      "grad_norm": 0.31832876801490784,
      "learning_rate": 9.461393979389817e-06,
      "loss": 0.0357,
      "step": 31140
    },
    {
      "epoch": 1.5810447230823899,
      "grad_norm": 0.9408812522888184,
      "learning_rate": 9.4597018461174e-06,
      "loss": 0.0517,
      "step": 31145
    },
    {
      "epoch": 1.5812985430732525,
      "grad_norm": 0.3648240268230438,
      "learning_rate": 9.458009712844984e-06,
      "loss": 0.0475,
      "step": 31150
    },
    {
      "epoch": 1.581552363064115,
      "grad_norm": 0.3548484146595001,
      "learning_rate": 9.456317579572568e-06,
      "loss": 0.0466,
      "step": 31155
    },
    {
      "epoch": 1.5818061830549774,
      "grad_norm": 0.6021202802658081,
      "learning_rate": 9.454625446300151e-06,
      "loss": 0.0413,
      "step": 31160
    },
    {
      "epoch": 1.5820600030458398,
      "grad_norm": 0.38574841618537903,
      "learning_rate": 9.452933313027735e-06,
      "loss": 0.0476,
      "step": 31165
    },
    {
      "epoch": 1.5823138230367024,
      "grad_norm": 0.4719664752483368,
      "learning_rate": 9.451241179755319e-06,
      "loss": 0.0518,
      "step": 31170
    },
    {
      "epoch": 1.5825676430275648,
      "grad_norm": 0.2845129072666168,
      "learning_rate": 9.449549046482902e-06,
      "loss": 0.045,
      "step": 31175
    },
    {
      "epoch": 1.5828214630184272,
      "grad_norm": 0.44490012526512146,
      "learning_rate": 9.447856913210484e-06,
      "loss": 0.0414,
      "step": 31180
    },
    {
      "epoch": 1.5830752830092898,
      "grad_norm": 0.6356984376907349,
      "learning_rate": 9.44616477993807e-06,
      "loss": 0.0504,
      "step": 31185
    },
    {
      "epoch": 1.5833291030001524,
      "grad_norm": 0.47728046774864197,
      "learning_rate": 9.444472646665653e-06,
      "loss": 0.0471,
      "step": 31190
    },
    {
      "epoch": 1.5835829229910148,
      "grad_norm": 0.23689168691635132,
      "learning_rate": 9.442780513393236e-06,
      "loss": 0.0388,
      "step": 31195
    },
    {
      "epoch": 1.5838367429818772,
      "grad_norm": 0.2717723548412323,
      "learning_rate": 9.441088380120818e-06,
      "loss": 0.0476,
      "step": 31200
    },
    {
      "epoch": 1.5840905629727398,
      "grad_norm": 0.35897499322891235,
      "learning_rate": 9.439396246848402e-06,
      "loss": 0.0521,
      "step": 31205
    },
    {
      "epoch": 1.5843443829636024,
      "grad_norm": 0.4387221336364746,
      "learning_rate": 9.437704113575987e-06,
      "loss": 0.0407,
      "step": 31210
    },
    {
      "epoch": 1.5845982029544647,
      "grad_norm": 0.2721177637577057,
      "learning_rate": 9.43601198030357e-06,
      "loss": 0.0408,
      "step": 31215
    },
    {
      "epoch": 1.5848520229453271,
      "grad_norm": 0.3234291672706604,
      "learning_rate": 9.434319847031152e-06,
      "loss": 0.0348,
      "step": 31220
    },
    {
      "epoch": 1.5851058429361897,
      "grad_norm": 0.5149333477020264,
      "learning_rate": 9.432627713758736e-06,
      "loss": 0.0459,
      "step": 31225
    },
    {
      "epoch": 1.585359662927052,
      "grad_norm": 0.40333887934684753,
      "learning_rate": 9.43093558048632e-06,
      "loss": 0.0494,
      "step": 31230
    },
    {
      "epoch": 1.5856134829179145,
      "grad_norm": 0.4914284646511078,
      "learning_rate": 9.429243447213905e-06,
      "loss": 0.0429,
      "step": 31235
    },
    {
      "epoch": 1.585867302908777,
      "grad_norm": 0.571137011051178,
      "learning_rate": 9.427551313941486e-06,
      "loss": 0.0471,
      "step": 31240
    },
    {
      "epoch": 1.5861211228996397,
      "grad_norm": 0.2421867698431015,
      "learning_rate": 9.42585918066907e-06,
      "loss": 0.0425,
      "step": 31245
    },
    {
      "epoch": 1.586374942890502,
      "grad_norm": 0.54640793800354,
      "learning_rate": 9.424167047396654e-06,
      "loss": 0.0364,
      "step": 31250
    },
    {
      "epoch": 1.5866287628813645,
      "grad_norm": 0.4234645366668701,
      "learning_rate": 9.422474914124237e-06,
      "loss": 0.0562,
      "step": 31255
    },
    {
      "epoch": 1.586882582872227,
      "grad_norm": 0.44822415709495544,
      "learning_rate": 9.42078278085182e-06,
      "loss": 0.0431,
      "step": 31260
    },
    {
      "epoch": 1.5871364028630897,
      "grad_norm": 0.32454562187194824,
      "learning_rate": 9.419090647579404e-06,
      "loss": 0.0349,
      "step": 31265
    },
    {
      "epoch": 1.5873902228539518,
      "grad_norm": 0.4538845717906952,
      "learning_rate": 9.417398514306988e-06,
      "loss": 0.048,
      "step": 31270
    },
    {
      "epoch": 1.5876440428448144,
      "grad_norm": 0.35510364174842834,
      "learning_rate": 9.415706381034571e-06,
      "loss": 0.0498,
      "step": 31275
    },
    {
      "epoch": 1.587897862835677,
      "grad_norm": 0.26279670000076294,
      "learning_rate": 9.414014247762155e-06,
      "loss": 0.0395,
      "step": 31280
    },
    {
      "epoch": 1.5881516828265394,
      "grad_norm": 0.35393065214157104,
      "learning_rate": 9.412322114489738e-06,
      "loss": 0.0424,
      "step": 31285
    },
    {
      "epoch": 1.5884055028174018,
      "grad_norm": 0.2939152717590332,
      "learning_rate": 9.410629981217322e-06,
      "loss": 0.0414,
      "step": 31290
    },
    {
      "epoch": 1.5886593228082644,
      "grad_norm": 0.35889071226119995,
      "learning_rate": 9.408937847944904e-06,
      "loss": 0.0514,
      "step": 31295
    },
    {
      "epoch": 1.588913142799127,
      "grad_norm": 0.41180306673049927,
      "learning_rate": 9.407245714672489e-06,
      "loss": 0.048,
      "step": 31300
    },
    {
      "epoch": 1.5891669627899894,
      "grad_norm": 0.46586310863494873,
      "learning_rate": 9.405553581400073e-06,
      "loss": 0.0521,
      "step": 31305
    },
    {
      "epoch": 1.5894207827808517,
      "grad_norm": 0.355020672082901,
      "learning_rate": 9.403861448127656e-06,
      "loss": 0.0383,
      "step": 31310
    },
    {
      "epoch": 1.5896746027717144,
      "grad_norm": 0.47443002462387085,
      "learning_rate": 9.402169314855238e-06,
      "loss": 0.043,
      "step": 31315
    },
    {
      "epoch": 1.5899284227625767,
      "grad_norm": 0.37852799892425537,
      "learning_rate": 9.400477181582822e-06,
      "loss": 0.0389,
      "step": 31320
    },
    {
      "epoch": 1.5901822427534391,
      "grad_norm": 0.3380233645439148,
      "learning_rate": 9.398785048310407e-06,
      "loss": 0.0397,
      "step": 31325
    },
    {
      "epoch": 1.5904360627443017,
      "grad_norm": 0.34201329946517944,
      "learning_rate": 9.39709291503799e-06,
      "loss": 0.044,
      "step": 31330
    },
    {
      "epoch": 1.5906898827351643,
      "grad_norm": 0.3614378571510315,
      "learning_rate": 9.395400781765572e-06,
      "loss": 0.0437,
      "step": 31335
    },
    {
      "epoch": 1.5909437027260267,
      "grad_norm": 0.2608165144920349,
      "learning_rate": 9.393708648493156e-06,
      "loss": 0.0432,
      "step": 31340
    },
    {
      "epoch": 1.591197522716889,
      "grad_norm": 0.2886785566806793,
      "learning_rate": 9.39201651522074e-06,
      "loss": 0.0394,
      "step": 31345
    },
    {
      "epoch": 1.5914513427077517,
      "grad_norm": 0.24169431626796722,
      "learning_rate": 9.390324381948323e-06,
      "loss": 0.0449,
      "step": 31350
    },
    {
      "epoch": 1.5917051626986143,
      "grad_norm": 0.2917103171348572,
      "learning_rate": 9.388632248675906e-06,
      "loss": 0.0421,
      "step": 31355
    },
    {
      "epoch": 1.5919589826894767,
      "grad_norm": 0.32198965549468994,
      "learning_rate": 9.38694011540349e-06,
      "loss": 0.0373,
      "step": 31360
    },
    {
      "epoch": 1.592212802680339,
      "grad_norm": 0.27352675795555115,
      "learning_rate": 9.385247982131073e-06,
      "loss": 0.0422,
      "step": 31365
    },
    {
      "epoch": 1.5924666226712016,
      "grad_norm": 0.33560416102409363,
      "learning_rate": 9.383555848858657e-06,
      "loss": 0.0437,
      "step": 31370
    },
    {
      "epoch": 1.592720442662064,
      "grad_norm": 0.24007673561573029,
      "learning_rate": 9.38186371558624e-06,
      "loss": 0.0386,
      "step": 31375
    },
    {
      "epoch": 1.5929742626529264,
      "grad_norm": 0.3025837242603302,
      "learning_rate": 9.380171582313824e-06,
      "loss": 0.0498,
      "step": 31380
    },
    {
      "epoch": 1.593228082643789,
      "grad_norm": 0.4102720618247986,
      "learning_rate": 9.378479449041408e-06,
      "loss": 0.0487,
      "step": 31385
    },
    {
      "epoch": 1.5934819026346516,
      "grad_norm": 0.3684849143028259,
      "learning_rate": 9.376787315768991e-06,
      "loss": 0.0399,
      "step": 31390
    },
    {
      "epoch": 1.593735722625514,
      "grad_norm": 0.39328739047050476,
      "learning_rate": 9.375095182496575e-06,
      "loss": 0.0464,
      "step": 31395
    },
    {
      "epoch": 1.5939895426163764,
      "grad_norm": 0.30063509941101074,
      "learning_rate": 9.373403049224158e-06,
      "loss": 0.0506,
      "step": 31400
    },
    {
      "epoch": 1.594243362607239,
      "grad_norm": 0.26448220014572144,
      "learning_rate": 9.371710915951742e-06,
      "loss": 0.0343,
      "step": 31405
    },
    {
      "epoch": 1.5944971825981016,
      "grad_norm": 0.5844763517379761,
      "learning_rate": 9.370018782679324e-06,
      "loss": 0.0436,
      "step": 31410
    },
    {
      "epoch": 1.594751002588964,
      "grad_norm": 0.35715028643608093,
      "learning_rate": 9.368326649406909e-06,
      "loss": 0.0445,
      "step": 31415
    },
    {
      "epoch": 1.5950048225798263,
      "grad_norm": 0.45810723304748535,
      "learning_rate": 9.366634516134492e-06,
      "loss": 0.045,
      "step": 31420
    },
    {
      "epoch": 1.595258642570689,
      "grad_norm": 0.35017555952072144,
      "learning_rate": 9.364942382862074e-06,
      "loss": 0.045,
      "step": 31425
    },
    {
      "epoch": 1.5955124625615513,
      "grad_norm": 0.3044382631778717,
      "learning_rate": 9.363250249589658e-06,
      "loss": 0.0375,
      "step": 31430
    },
    {
      "epoch": 1.5957662825524137,
      "grad_norm": 0.36368584632873535,
      "learning_rate": 9.361558116317241e-06,
      "loss": 0.0423,
      "step": 31435
    },
    {
      "epoch": 1.5960201025432763,
      "grad_norm": 0.3246999680995941,
      "learning_rate": 9.359865983044827e-06,
      "loss": 0.0469,
      "step": 31440
    },
    {
      "epoch": 1.596273922534139,
      "grad_norm": 0.3725318908691406,
      "learning_rate": 9.358173849772408e-06,
      "loss": 0.0493,
      "step": 31445
    },
    {
      "epoch": 1.5965277425250013,
      "grad_norm": 0.3187475800514221,
      "learning_rate": 9.356481716499992e-06,
      "loss": 0.0401,
      "step": 31450
    },
    {
      "epoch": 1.5967815625158637,
      "grad_norm": 0.3895965814590454,
      "learning_rate": 9.354789583227576e-06,
      "loss": 0.0429,
      "step": 31455
    },
    {
      "epoch": 1.5970353825067263,
      "grad_norm": 0.39535629749298096,
      "learning_rate": 9.353097449955159e-06,
      "loss": 0.0391,
      "step": 31460
    },
    {
      "epoch": 1.5972892024975887,
      "grad_norm": 0.40566694736480713,
      "learning_rate": 9.351405316682743e-06,
      "loss": 0.0446,
      "step": 31465
    },
    {
      "epoch": 1.597543022488451,
      "grad_norm": 0.426129549741745,
      "learning_rate": 9.349713183410326e-06,
      "loss": 0.0412,
      "step": 31470
    },
    {
      "epoch": 1.5977968424793136,
      "grad_norm": 0.3265942931175232,
      "learning_rate": 9.34802105013791e-06,
      "loss": 0.0465,
      "step": 31475
    },
    {
      "epoch": 1.5980506624701762,
      "grad_norm": 0.27125173807144165,
      "learning_rate": 9.346328916865493e-06,
      "loss": 0.0362,
      "step": 31480
    },
    {
      "epoch": 1.5983044824610386,
      "grad_norm": 0.3212997615337372,
      "learning_rate": 9.344636783593077e-06,
      "loss": 0.0436,
      "step": 31485
    },
    {
      "epoch": 1.598558302451901,
      "grad_norm": 0.32562533020973206,
      "learning_rate": 9.34294465032066e-06,
      "loss": 0.0474,
      "step": 31490
    },
    {
      "epoch": 1.5988121224427636,
      "grad_norm": 0.37782034277915955,
      "learning_rate": 9.341252517048244e-06,
      "loss": 0.0463,
      "step": 31495
    },
    {
      "epoch": 1.5990659424336262,
      "grad_norm": 0.33579403162002563,
      "learning_rate": 9.339560383775826e-06,
      "loss": 0.0517,
      "step": 31500
    },
    {
      "epoch": 1.5993197624244886,
      "grad_norm": 0.2755340039730072,
      "learning_rate": 9.337868250503411e-06,
      "loss": 0.0348,
      "step": 31505
    },
    {
      "epoch": 1.599573582415351,
      "grad_norm": 0.3979896306991577,
      "learning_rate": 9.336176117230995e-06,
      "loss": 0.0447,
      "step": 31510
    },
    {
      "epoch": 1.5998274024062136,
      "grad_norm": 0.3220609426498413,
      "learning_rate": 9.334483983958578e-06,
      "loss": 0.0437,
      "step": 31515
    },
    {
      "epoch": 1.600081222397076,
      "grad_norm": 0.30401086807250977,
      "learning_rate": 9.33279185068616e-06,
      "loss": 0.0369,
      "step": 31520
    },
    {
      "epoch": 1.6003350423879383,
      "grad_norm": 0.48431262373924255,
      "learning_rate": 9.331099717413743e-06,
      "loss": 0.042,
      "step": 31525
    },
    {
      "epoch": 1.600588862378801,
      "grad_norm": 0.3727475106716156,
      "learning_rate": 9.329407584141329e-06,
      "loss": 0.0476,
      "step": 31530
    },
    {
      "epoch": 1.6008426823696635,
      "grad_norm": 0.3813401460647583,
      "learning_rate": 9.327715450868912e-06,
      "loss": 0.0392,
      "step": 31535
    },
    {
      "epoch": 1.601096502360526,
      "grad_norm": 0.5032063126564026,
      "learning_rate": 9.326023317596494e-06,
      "loss": 0.0458,
      "step": 31540
    },
    {
      "epoch": 1.6013503223513883,
      "grad_norm": 0.4369184672832489,
      "learning_rate": 9.324331184324078e-06,
      "loss": 0.0441,
      "step": 31545
    },
    {
      "epoch": 1.601604142342251,
      "grad_norm": 0.3979043662548065,
      "learning_rate": 9.322639051051661e-06,
      "loss": 0.0507,
      "step": 31550
    },
    {
      "epoch": 1.6018579623331135,
      "grad_norm": 0.3451641798019409,
      "learning_rate": 9.320946917779245e-06,
      "loss": 0.0451,
      "step": 31555
    },
    {
      "epoch": 1.6021117823239759,
      "grad_norm": 0.30609747767448425,
      "learning_rate": 9.319254784506828e-06,
      "loss": 0.0406,
      "step": 31560
    },
    {
      "epoch": 1.6023656023148383,
      "grad_norm": 0.2856360077857971,
      "learning_rate": 9.317562651234412e-06,
      "loss": 0.0471,
      "step": 31565
    },
    {
      "epoch": 1.6026194223057009,
      "grad_norm": 0.2808777987957001,
      "learning_rate": 9.315870517961995e-06,
      "loss": 0.0401,
      "step": 31570
    },
    {
      "epoch": 1.6028732422965632,
      "grad_norm": 0.27101564407348633,
      "learning_rate": 9.314178384689579e-06,
      "loss": 0.0367,
      "step": 31575
    },
    {
      "epoch": 1.6031270622874256,
      "grad_norm": 0.38390570878982544,
      "learning_rate": 9.312486251417162e-06,
      "loss": 0.0496,
      "step": 31580
    },
    {
      "epoch": 1.6033808822782882,
      "grad_norm": 0.44923219084739685,
      "learning_rate": 9.310794118144746e-06,
      "loss": 0.047,
      "step": 31585
    },
    {
      "epoch": 1.6036347022691508,
      "grad_norm": 0.4212184548377991,
      "learning_rate": 9.30910198487233e-06,
      "loss": 0.0504,
      "step": 31590
    },
    {
      "epoch": 1.6038885222600132,
      "grad_norm": 0.3964141011238098,
      "learning_rate": 9.307409851599913e-06,
      "loss": 0.0454,
      "step": 31595
    },
    {
      "epoch": 1.6041423422508756,
      "grad_norm": 0.4174087643623352,
      "learning_rate": 9.305717718327497e-06,
      "loss": 0.0422,
      "step": 31600
    },
    {
      "epoch": 1.6043961622417382,
      "grad_norm": 0.28624725341796875,
      "learning_rate": 9.30402558505508e-06,
      "loss": 0.0412,
      "step": 31605
    },
    {
      "epoch": 1.6046499822326008,
      "grad_norm": 0.3116838037967682,
      "learning_rate": 9.302333451782664e-06,
      "loss": 0.053,
      "step": 31610
    },
    {
      "epoch": 1.604903802223463,
      "grad_norm": 0.2775300145149231,
      "learning_rate": 9.300641318510246e-06,
      "loss": 0.0477,
      "step": 31615
    },
    {
      "epoch": 1.6051576222143256,
      "grad_norm": 0.3478650748729706,
      "learning_rate": 9.29894918523783e-06,
      "loss": 0.046,
      "step": 31620
    },
    {
      "epoch": 1.6054114422051882,
      "grad_norm": 0.35184112191200256,
      "learning_rate": 9.297257051965414e-06,
      "loss": 0.0534,
      "step": 31625
    },
    {
      "epoch": 1.6056652621960505,
      "grad_norm": 0.9067050218582153,
      "learning_rate": 9.295564918692998e-06,
      "loss": 0.0453,
      "step": 31630
    },
    {
      "epoch": 1.605919082186913,
      "grad_norm": 1.2345666885375977,
      "learning_rate": 9.29387278542058e-06,
      "loss": 0.0417,
      "step": 31635
    },
    {
      "epoch": 1.6061729021777755,
      "grad_norm": 0.39380595088005066,
      "learning_rate": 9.292180652148163e-06,
      "loss": 0.0456,
      "step": 31640
    },
    {
      "epoch": 1.6064267221686381,
      "grad_norm": 0.3252556324005127,
      "learning_rate": 9.290488518875747e-06,
      "loss": 0.0375,
      "step": 31645
    },
    {
      "epoch": 1.6066805421595005,
      "grad_norm": 0.2784633934497833,
      "learning_rate": 9.288796385603332e-06,
      "loss": 0.0367,
      "step": 31650
    },
    {
      "epoch": 1.606934362150363,
      "grad_norm": 0.3153322637081146,
      "learning_rate": 9.287104252330914e-06,
      "loss": 0.0403,
      "step": 31655
    },
    {
      "epoch": 1.6071881821412255,
      "grad_norm": 0.3245221674442291,
      "learning_rate": 9.285412119058497e-06,
      "loss": 0.0346,
      "step": 31660
    },
    {
      "epoch": 1.6074420021320879,
      "grad_norm": 0.3184836506843567,
      "learning_rate": 9.283719985786081e-06,
      "loss": 0.0375,
      "step": 31665
    },
    {
      "epoch": 1.6076958221229503,
      "grad_norm": 0.3533708453178406,
      "learning_rate": 9.282027852513665e-06,
      "loss": 0.0468,
      "step": 31670
    },
    {
      "epoch": 1.6079496421138129,
      "grad_norm": 1.2411906719207764,
      "learning_rate": 9.280335719241248e-06,
      "loss": 0.0435,
      "step": 31675
    },
    {
      "epoch": 1.6082034621046755,
      "grad_norm": 0.3504463732242584,
      "learning_rate": 9.278643585968832e-06,
      "loss": 0.0484,
      "step": 31680
    },
    {
      "epoch": 1.6084572820955378,
      "grad_norm": 0.25611740350723267,
      "learning_rate": 9.276951452696415e-06,
      "loss": 0.0379,
      "step": 31685
    },
    {
      "epoch": 1.6087111020864002,
      "grad_norm": 0.35780343413352966,
      "learning_rate": 9.275259319423999e-06,
      "loss": 0.0496,
      "step": 31690
    },
    {
      "epoch": 1.6089649220772628,
      "grad_norm": 0.3506191074848175,
      "learning_rate": 9.273567186151582e-06,
      "loss": 0.0367,
      "step": 31695
    },
    {
      "epoch": 1.6092187420681254,
      "grad_norm": 0.69794100522995,
      "learning_rate": 9.271875052879166e-06,
      "loss": 0.0459,
      "step": 31700
    },
    {
      "epoch": 1.6094725620589878,
      "grad_norm": 0.29237765073776245,
      "learning_rate": 9.27018291960675e-06,
      "loss": 0.041,
      "step": 31705
    },
    {
      "epoch": 1.6097263820498502,
      "grad_norm": 0.4747942388057709,
      "learning_rate": 9.268490786334333e-06,
      "loss": 0.0407,
      "step": 31710
    },
    {
      "epoch": 1.6099802020407128,
      "grad_norm": 0.29343801736831665,
      "learning_rate": 9.266798653061916e-06,
      "loss": 0.0394,
      "step": 31715
    },
    {
      "epoch": 1.6102340220315752,
      "grad_norm": 0.3256320059299469,
      "learning_rate": 9.2651065197895e-06,
      "loss": 0.0485,
      "step": 31720
    },
    {
      "epoch": 1.6104878420224376,
      "grad_norm": 0.2793024778366089,
      "learning_rate": 9.263414386517084e-06,
      "loss": 0.0436,
      "step": 31725
    },
    {
      "epoch": 1.6107416620133002,
      "grad_norm": 0.5951392650604248,
      "learning_rate": 9.261722253244665e-06,
      "loss": 0.0402,
      "step": 31730
    },
    {
      "epoch": 1.6109954820041628,
      "grad_norm": 0.30947548151016235,
      "learning_rate": 9.260030119972249e-06,
      "loss": 0.0373,
      "step": 31735
    },
    {
      "epoch": 1.6112493019950251,
      "grad_norm": 0.491113543510437,
      "learning_rate": 9.258337986699834e-06,
      "loss": 0.0428,
      "step": 31740
    },
    {
      "epoch": 1.6115031219858875,
      "grad_norm": 0.3150951862335205,
      "learning_rate": 9.256645853427416e-06,
      "loss": 0.0411,
      "step": 31745
    },
    {
      "epoch": 1.6117569419767501,
      "grad_norm": 0.6105877757072449,
      "learning_rate": 9.254953720155e-06,
      "loss": 0.0394,
      "step": 31750
    },
    {
      "epoch": 1.6120107619676127,
      "grad_norm": 0.3246072828769684,
      "learning_rate": 9.253261586882583e-06,
      "loss": 0.039,
      "step": 31755
    },
    {
      "epoch": 1.6122645819584749,
      "grad_norm": 0.44057467579841614,
      "learning_rate": 9.251569453610167e-06,
      "loss": 0.042,
      "step": 31760
    },
    {
      "epoch": 1.6125184019493375,
      "grad_norm": 0.39417964220046997,
      "learning_rate": 9.24987732033775e-06,
      "loss": 0.0438,
      "step": 31765
    },
    {
      "epoch": 1.6127722219402,
      "grad_norm": 0.5825387835502625,
      "learning_rate": 9.248185187065334e-06,
      "loss": 0.0444,
      "step": 31770
    },
    {
      "epoch": 1.6130260419310625,
      "grad_norm": 0.36112135648727417,
      "learning_rate": 9.246493053792917e-06,
      "loss": 0.0469,
      "step": 31775
    },
    {
      "epoch": 1.6132798619219249,
      "grad_norm": 0.488370418548584,
      "learning_rate": 9.2448009205205e-06,
      "loss": 0.036,
      "step": 31780
    },
    {
      "epoch": 1.6135336819127875,
      "grad_norm": 0.3322552442550659,
      "learning_rate": 9.243108787248084e-06,
      "loss": 0.0407,
      "step": 31785
    },
    {
      "epoch": 1.61378750190365,
      "grad_norm": 0.4039710462093353,
      "learning_rate": 9.241416653975668e-06,
      "loss": 0.0438,
      "step": 31790
    },
    {
      "epoch": 1.6140413218945124,
      "grad_norm": 0.34084275364875793,
      "learning_rate": 9.239724520703251e-06,
      "loss": 0.0445,
      "step": 31795
    },
    {
      "epoch": 1.6142951418853748,
      "grad_norm": 0.4430694282054901,
      "learning_rate": 9.238032387430835e-06,
      "loss": 0.0505,
      "step": 31800
    },
    {
      "epoch": 1.6145489618762374,
      "grad_norm": 0.3359103798866272,
      "learning_rate": 9.236340254158419e-06,
      "loss": 0.037,
      "step": 31805
    },
    {
      "epoch": 1.6148027818670998,
      "grad_norm": 0.26327624917030334,
      "learning_rate": 9.234648120886002e-06,
      "loss": 0.0391,
      "step": 31810
    },
    {
      "epoch": 1.6150566018579622,
      "grad_norm": 0.2624158561229706,
      "learning_rate": 9.232955987613586e-06,
      "loss": 0.0362,
      "step": 31815
    },
    {
      "epoch": 1.6153104218488248,
      "grad_norm": 0.36701250076293945,
      "learning_rate": 9.231263854341167e-06,
      "loss": 0.0437,
      "step": 31820
    },
    {
      "epoch": 1.6155642418396874,
      "grad_norm": 0.7499197125434875,
      "learning_rate": 9.229571721068751e-06,
      "loss": 0.04,
      "step": 31825
    },
    {
      "epoch": 1.6158180618305498,
      "grad_norm": 0.3554810881614685,
      "learning_rate": 9.227879587796336e-06,
      "loss": 0.0459,
      "step": 31830
    },
    {
      "epoch": 1.6160718818214121,
      "grad_norm": 0.4646514356136322,
      "learning_rate": 9.22618745452392e-06,
      "loss": 0.0395,
      "step": 31835
    },
    {
      "epoch": 1.6163257018122748,
      "grad_norm": 0.5427611470222473,
      "learning_rate": 9.224495321251502e-06,
      "loss": 0.0456,
      "step": 31840
    },
    {
      "epoch": 1.6165795218031374,
      "grad_norm": 0.346992552280426,
      "learning_rate": 9.222803187979085e-06,
      "loss": 0.0344,
      "step": 31845
    },
    {
      "epoch": 1.6168333417939997,
      "grad_norm": 0.473665714263916,
      "learning_rate": 9.221111054706669e-06,
      "loss": 0.0344,
      "step": 31850
    },
    {
      "epoch": 1.6170871617848621,
      "grad_norm": 0.38679835200309753,
      "learning_rate": 9.219418921434254e-06,
      "loss": 0.0375,
      "step": 31855
    },
    {
      "epoch": 1.6173409817757247,
      "grad_norm": 0.689315676689148,
      "learning_rate": 9.217726788161836e-06,
      "loss": 0.0445,
      "step": 31860
    },
    {
      "epoch": 1.617594801766587,
      "grad_norm": 0.3175433874130249,
      "learning_rate": 9.21603465488942e-06,
      "loss": 0.0482,
      "step": 31865
    },
    {
      "epoch": 1.6178486217574495,
      "grad_norm": 0.301717072725296,
      "learning_rate": 9.214342521617003e-06,
      "loss": 0.0426,
      "step": 31870
    },
    {
      "epoch": 1.618102441748312,
      "grad_norm": 0.24479809403419495,
      "learning_rate": 9.212650388344586e-06,
      "loss": 0.0344,
      "step": 31875
    },
    {
      "epoch": 1.6183562617391747,
      "grad_norm": 0.36919647455215454,
      "learning_rate": 9.21095825507217e-06,
      "loss": 0.04,
      "step": 31880
    },
    {
      "epoch": 1.618610081730037,
      "grad_norm": 0.3051331639289856,
      "learning_rate": 9.209266121799754e-06,
      "loss": 0.0337,
      "step": 31885
    },
    {
      "epoch": 1.6188639017208994,
      "grad_norm": 0.31036415696144104,
      "learning_rate": 9.207573988527337e-06,
      "loss": 0.0458,
      "step": 31890
    },
    {
      "epoch": 1.619117721711762,
      "grad_norm": 0.4276675581932068,
      "learning_rate": 9.20588185525492e-06,
      "loss": 0.0368,
      "step": 31895
    },
    {
      "epoch": 1.6193715417026247,
      "grad_norm": 0.2697307765483856,
      "learning_rate": 9.204189721982504e-06,
      "loss": 0.0381,
      "step": 31900
    },
    {
      "epoch": 1.619625361693487,
      "grad_norm": 0.4047255516052246,
      "learning_rate": 9.202497588710088e-06,
      "loss": 0.0473,
      "step": 31905
    },
    {
      "epoch": 1.6198791816843494,
      "grad_norm": 0.46836021542549133,
      "learning_rate": 9.200805455437671e-06,
      "loss": 0.038,
      "step": 31910
    },
    {
      "epoch": 1.620133001675212,
      "grad_norm": 1.1610099077224731,
      "learning_rate": 9.199113322165253e-06,
      "loss": 0.0512,
      "step": 31915
    },
    {
      "epoch": 1.6203868216660744,
      "grad_norm": 0.3843535780906677,
      "learning_rate": 9.197421188892838e-06,
      "loss": 0.042,
      "step": 31920
    },
    {
      "epoch": 1.6206406416569368,
      "grad_norm": 0.35286080837249756,
      "learning_rate": 9.195729055620422e-06,
      "loss": 0.0425,
      "step": 31925
    },
    {
      "epoch": 1.6208944616477994,
      "grad_norm": 0.30205556750297546,
      "learning_rate": 9.194036922348005e-06,
      "loss": 0.0411,
      "step": 31930
    },
    {
      "epoch": 1.621148281638662,
      "grad_norm": 0.3636847138404846,
      "learning_rate": 9.192344789075587e-06,
      "loss": 0.0399,
      "step": 31935
    },
    {
      "epoch": 1.6214021016295244,
      "grad_norm": 0.2955825626850128,
      "learning_rate": 9.190652655803171e-06,
      "loss": 0.0384,
      "step": 31940
    },
    {
      "epoch": 1.6216559216203867,
      "grad_norm": 0.5489678978919983,
      "learning_rate": 9.188960522530756e-06,
      "loss": 0.0461,
      "step": 31945
    },
    {
      "epoch": 1.6219097416112493,
      "grad_norm": 0.4751637876033783,
      "learning_rate": 9.18726838925834e-06,
      "loss": 0.0537,
      "step": 31950
    },
    {
      "epoch": 1.622163561602112,
      "grad_norm": 0.8122043609619141,
      "learning_rate": 9.185576255985922e-06,
      "loss": 0.0493,
      "step": 31955
    },
    {
      "epoch": 1.622417381592974,
      "grad_norm": 0.32820770144462585,
      "learning_rate": 9.183884122713505e-06,
      "loss": 0.0362,
      "step": 31960
    },
    {
      "epoch": 1.6226712015838367,
      "grad_norm": 0.5642192363739014,
      "learning_rate": 9.182191989441089e-06,
      "loss": 0.0402,
      "step": 31965
    },
    {
      "epoch": 1.6229250215746993,
      "grad_norm": 0.3879733085632324,
      "learning_rate": 9.180499856168674e-06,
      "loss": 0.0561,
      "step": 31970
    },
    {
      "epoch": 1.6231788415655617,
      "grad_norm": 0.4680909812450409,
      "learning_rate": 9.178807722896256e-06,
      "loss": 0.039,
      "step": 31975
    },
    {
      "epoch": 1.623432661556424,
      "grad_norm": 0.30603843927383423,
      "learning_rate": 9.17711558962384e-06,
      "loss": 0.0452,
      "step": 31980
    },
    {
      "epoch": 1.6236864815472867,
      "grad_norm": 0.36126911640167236,
      "learning_rate": 9.175423456351423e-06,
      "loss": 0.0455,
      "step": 31985
    },
    {
      "epoch": 1.6239403015381493,
      "grad_norm": 0.3678168058395386,
      "learning_rate": 9.173731323079006e-06,
      "loss": 0.0448,
      "step": 31990
    },
    {
      "epoch": 1.6241941215290117,
      "grad_norm": 0.28959447145462036,
      "learning_rate": 9.17203918980659e-06,
      "loss": 0.0383,
      "step": 31995
    },
    {
      "epoch": 1.624447941519874,
      "grad_norm": 0.4826527535915375,
      "learning_rate": 9.170347056534173e-06,
      "loss": 0.0417,
      "step": 32000
    },
    {
      "epoch": 1.6247017615107366,
      "grad_norm": 0.31721869111061096,
      "learning_rate": 9.168654923261757e-06,
      "loss": 0.0431,
      "step": 32005
    },
    {
      "epoch": 1.624955581501599,
      "grad_norm": 0.2833889424800873,
      "learning_rate": 9.16696278998934e-06,
      "loss": 0.0443,
      "step": 32010
    },
    {
      "epoch": 1.6252094014924614,
      "grad_norm": 0.37571266293525696,
      "learning_rate": 9.165270656716924e-06,
      "loss": 0.0475,
      "step": 32015
    },
    {
      "epoch": 1.625463221483324,
      "grad_norm": 0.4514489769935608,
      "learning_rate": 9.163578523444508e-06,
      "loss": 0.0465,
      "step": 32020
    },
    {
      "epoch": 1.6257170414741866,
      "grad_norm": 0.5537285208702087,
      "learning_rate": 9.161886390172091e-06,
      "loss": 0.0461,
      "step": 32025
    },
    {
      "epoch": 1.625970861465049,
      "grad_norm": 0.3658086955547333,
      "learning_rate": 9.160194256899673e-06,
      "loss": 0.0532,
      "step": 32030
    },
    {
      "epoch": 1.6262246814559114,
      "grad_norm": 0.2867622375488281,
      "learning_rate": 9.158502123627258e-06,
      "loss": 0.0388,
      "step": 32035
    },
    {
      "epoch": 1.626478501446774,
      "grad_norm": 0.6386134028434753,
      "learning_rate": 9.156809990354842e-06,
      "loss": 0.0508,
      "step": 32040
    },
    {
      "epoch": 1.6267323214376366,
      "grad_norm": 0.2683972716331482,
      "learning_rate": 9.155117857082425e-06,
      "loss": 0.0466,
      "step": 32045
    },
    {
      "epoch": 1.626986141428499,
      "grad_norm": 0.3644627034664154,
      "learning_rate": 9.153425723810007e-06,
      "loss": 0.0486,
      "step": 32050
    },
    {
      "epoch": 1.6272399614193613,
      "grad_norm": 0.4730590283870697,
      "learning_rate": 9.15173359053759e-06,
      "loss": 0.0408,
      "step": 32055
    },
    {
      "epoch": 1.627493781410224,
      "grad_norm": 0.2658335566520691,
      "learning_rate": 9.150041457265176e-06,
      "loss": 0.0431,
      "step": 32060
    },
    {
      "epoch": 1.6277476014010863,
      "grad_norm": 0.2563387453556061,
      "learning_rate": 9.148349323992758e-06,
      "loss": 0.0378,
      "step": 32065
    },
    {
      "epoch": 1.6280014213919487,
      "grad_norm": 0.3779682219028473,
      "learning_rate": 9.146657190720341e-06,
      "loss": 0.0524,
      "step": 32070
    },
    {
      "epoch": 1.6282552413828113,
      "grad_norm": 0.8039404153823853,
      "learning_rate": 9.144965057447925e-06,
      "loss": 0.0487,
      "step": 32075
    },
    {
      "epoch": 1.628509061373674,
      "grad_norm": 0.3716595768928528,
      "learning_rate": 9.143272924175508e-06,
      "loss": 0.0451,
      "step": 32080
    },
    {
      "epoch": 1.6287628813645363,
      "grad_norm": 0.30066925287246704,
      "learning_rate": 9.141580790903092e-06,
      "loss": 0.0469,
      "step": 32085
    },
    {
      "epoch": 1.6290167013553987,
      "grad_norm": 0.36805853247642517,
      "learning_rate": 9.139888657630676e-06,
      "loss": 0.0427,
      "step": 32090
    },
    {
      "epoch": 1.6292705213462613,
      "grad_norm": 0.7319865226745605,
      "learning_rate": 9.138196524358259e-06,
      "loss": 0.0382,
      "step": 32095
    },
    {
      "epoch": 1.6295243413371239,
      "grad_norm": 0.5053632855415344,
      "learning_rate": 9.136504391085843e-06,
      "loss": 0.0468,
      "step": 32100
    },
    {
      "epoch": 1.629778161327986,
      "grad_norm": 0.32306617498397827,
      "learning_rate": 9.134812257813426e-06,
      "loss": 0.0435,
      "step": 32105
    },
    {
      "epoch": 1.6300319813188486,
      "grad_norm": 0.2762267291545868,
      "learning_rate": 9.13312012454101e-06,
      "loss": 0.0425,
      "step": 32110
    },
    {
      "epoch": 1.6302858013097112,
      "grad_norm": 0.350439190864563,
      "learning_rate": 9.131427991268593e-06,
      "loss": 0.048,
      "step": 32115
    },
    {
      "epoch": 1.6305396213005736,
      "grad_norm": 0.296966552734375,
      "learning_rate": 9.129735857996177e-06,
      "loss": 0.0403,
      "step": 32120
    },
    {
      "epoch": 1.630793441291436,
      "grad_norm": 0.4328997731208801,
      "learning_rate": 9.12804372472376e-06,
      "loss": 0.0439,
      "step": 32125
    },
    {
      "epoch": 1.6310472612822986,
      "grad_norm": 0.384617418050766,
      "learning_rate": 9.126351591451344e-06,
      "loss": 0.0488,
      "step": 32130
    },
    {
      "epoch": 1.6313010812731612,
      "grad_norm": 0.43519383668899536,
      "learning_rate": 9.124659458178927e-06,
      "loss": 0.0448,
      "step": 32135
    },
    {
      "epoch": 1.6315549012640236,
      "grad_norm": 0.6686396598815918,
      "learning_rate": 9.122967324906511e-06,
      "loss": 0.0449,
      "step": 32140
    },
    {
      "epoch": 1.631808721254886,
      "grad_norm": 0.44832053780555725,
      "learning_rate": 9.121275191634093e-06,
      "loss": 0.0422,
      "step": 32145
    },
    {
      "epoch": 1.6320625412457486,
      "grad_norm": 0.3811791241168976,
      "learning_rate": 9.119583058361678e-06,
      "loss": 0.0378,
      "step": 32150
    },
    {
      "epoch": 1.632316361236611,
      "grad_norm": 0.3821338713169098,
      "learning_rate": 9.117890925089262e-06,
      "loss": 0.0439,
      "step": 32155
    },
    {
      "epoch": 1.6325701812274733,
      "grad_norm": 0.30220335721969604,
      "learning_rate": 9.116198791816843e-06,
      "loss": 0.0387,
      "step": 32160
    },
    {
      "epoch": 1.632824001218336,
      "grad_norm": 0.456337571144104,
      "learning_rate": 9.114506658544427e-06,
      "loss": 0.0492,
      "step": 32165
    },
    {
      "epoch": 1.6330778212091985,
      "grad_norm": 0.24902650713920593,
      "learning_rate": 9.11281452527201e-06,
      "loss": 0.0441,
      "step": 32170
    },
    {
      "epoch": 1.633331641200061,
      "grad_norm": 0.38890451192855835,
      "learning_rate": 9.111122391999596e-06,
      "loss": 0.0418,
      "step": 32175
    },
    {
      "epoch": 1.6335854611909233,
      "grad_norm": 1.0790613889694214,
      "learning_rate": 9.109430258727178e-06,
      "loss": 0.0367,
      "step": 32180
    },
    {
      "epoch": 1.633839281181786,
      "grad_norm": 0.46890148520469666,
      "learning_rate": 9.107738125454761e-06,
      "loss": 0.0406,
      "step": 32185
    },
    {
      "epoch": 1.6340931011726485,
      "grad_norm": 0.2735448479652405,
      "learning_rate": 9.106045992182345e-06,
      "loss": 0.0448,
      "step": 32190
    },
    {
      "epoch": 1.6343469211635109,
      "grad_norm": 0.3427935838699341,
      "learning_rate": 9.104353858909928e-06,
      "loss": 0.0396,
      "step": 32195
    },
    {
      "epoch": 1.6346007411543733,
      "grad_norm": 0.4376786947250366,
      "learning_rate": 9.102661725637512e-06,
      "loss": 0.0445,
      "step": 32200
    },
    {
      "epoch": 1.6348545611452359,
      "grad_norm": 0.29393336176872253,
      "learning_rate": 9.100969592365095e-06,
      "loss": 0.0456,
      "step": 32205
    },
    {
      "epoch": 1.6351083811360982,
      "grad_norm": 0.389175146818161,
      "learning_rate": 9.099277459092679e-06,
      "loss": 0.0441,
      "step": 32210
    },
    {
      "epoch": 1.6353622011269606,
      "grad_norm": 0.3740473687648773,
      "learning_rate": 9.097585325820262e-06,
      "loss": 0.0412,
      "step": 32215
    },
    {
      "epoch": 1.6356160211178232,
      "grad_norm": 0.3517167270183563,
      "learning_rate": 9.095893192547846e-06,
      "loss": 0.0422,
      "step": 32220
    },
    {
      "epoch": 1.6358698411086858,
      "grad_norm": 0.4503104090690613,
      "learning_rate": 9.09420105927543e-06,
      "loss": 0.0432,
      "step": 32225
    },
    {
      "epoch": 1.6361236610995482,
      "grad_norm": 0.279237300157547,
      "learning_rate": 9.092508926003013e-06,
      "loss": 0.0373,
      "step": 32230
    },
    {
      "epoch": 1.6363774810904106,
      "grad_norm": 0.3596137464046478,
      "learning_rate": 9.090816792730595e-06,
      "loss": 0.0477,
      "step": 32235
    },
    {
      "epoch": 1.6366313010812732,
      "grad_norm": 0.3686954975128174,
      "learning_rate": 9.08912465945818e-06,
      "loss": 0.041,
      "step": 32240
    },
    {
      "epoch": 1.6368851210721358,
      "grad_norm": 0.4615430235862732,
      "learning_rate": 9.087432526185764e-06,
      "loss": 0.0353,
      "step": 32245
    },
    {
      "epoch": 1.6371389410629982,
      "grad_norm": 0.3107963502407074,
      "learning_rate": 9.085740392913347e-06,
      "loss": 0.0471,
      "step": 32250
    },
    {
      "epoch": 1.6373927610538606,
      "grad_norm": 0.3667447566986084,
      "learning_rate": 9.084048259640929e-06,
      "loss": 0.0423,
      "step": 32255
    },
    {
      "epoch": 1.6376465810447232,
      "grad_norm": 0.49654021859169006,
      "learning_rate": 9.082356126368513e-06,
      "loss": 0.0432,
      "step": 32260
    },
    {
      "epoch": 1.6379004010355855,
      "grad_norm": 1.1227554082870483,
      "learning_rate": 9.080663993096098e-06,
      "loss": 0.0438,
      "step": 32265
    },
    {
      "epoch": 1.638154221026448,
      "grad_norm": 0.41635578870773315,
      "learning_rate": 9.078971859823681e-06,
      "loss": 0.0554,
      "step": 32270
    },
    {
      "epoch": 1.6384080410173105,
      "grad_norm": 0.22257228195667267,
      "learning_rate": 9.077279726551263e-06,
      "loss": 0.0416,
      "step": 32275
    },
    {
      "epoch": 1.6386618610081731,
      "grad_norm": 0.2682057321071625,
      "learning_rate": 9.075587593278847e-06,
      "loss": 0.0363,
      "step": 32280
    },
    {
      "epoch": 1.6389156809990355,
      "grad_norm": 0.305270254611969,
      "learning_rate": 9.07389546000643e-06,
      "loss": 0.0384,
      "step": 32285
    },
    {
      "epoch": 1.6391695009898979,
      "grad_norm": 0.5659728050231934,
      "learning_rate": 9.072203326734016e-06,
      "loss": 0.0497,
      "step": 32290
    },
    {
      "epoch": 1.6394233209807605,
      "grad_norm": 0.38251349329948425,
      "learning_rate": 9.070511193461597e-06,
      "loss": 0.0437,
      "step": 32295
    },
    {
      "epoch": 1.6396771409716229,
      "grad_norm": 0.7674602270126343,
      "learning_rate": 9.068819060189181e-06,
      "loss": 0.0498,
      "step": 32300
    },
    {
      "epoch": 1.6399309609624853,
      "grad_norm": 0.49687087535858154,
      "learning_rate": 9.067126926916765e-06,
      "loss": 0.0327,
      "step": 32305
    },
    {
      "epoch": 1.6401847809533479,
      "grad_norm": 0.3052130341529846,
      "learning_rate": 9.065434793644348e-06,
      "loss": 0.044,
      "step": 32310
    },
    {
      "epoch": 1.6404386009442105,
      "grad_norm": 0.31215372681617737,
      "learning_rate": 9.063742660371932e-06,
      "loss": 0.0378,
      "step": 32315
    },
    {
      "epoch": 1.6406924209350728,
      "grad_norm": 0.4322529435157776,
      "learning_rate": 9.062050527099515e-06,
      "loss": 0.0389,
      "step": 32320
    },
    {
      "epoch": 1.6409462409259352,
      "grad_norm": 0.27929675579071045,
      "learning_rate": 9.060358393827099e-06,
      "loss": 0.041,
      "step": 32325
    },
    {
      "epoch": 1.6412000609167978,
      "grad_norm": 0.3943149745464325,
      "learning_rate": 9.058666260554682e-06,
      "loss": 0.0454,
      "step": 32330
    },
    {
      "epoch": 1.6414538809076604,
      "grad_norm": 0.3910643756389618,
      "learning_rate": 9.056974127282266e-06,
      "loss": 0.0445,
      "step": 32335
    },
    {
      "epoch": 1.6417077008985228,
      "grad_norm": 0.42487633228302,
      "learning_rate": 9.05528199400985e-06,
      "loss": 0.0507,
      "step": 32340
    },
    {
      "epoch": 1.6419615208893852,
      "grad_norm": 0.2816854417324066,
      "learning_rate": 9.053589860737433e-06,
      "loss": 0.0319,
      "step": 32345
    },
    {
      "epoch": 1.6422153408802478,
      "grad_norm": 0.28755298256874084,
      "learning_rate": 9.051897727465015e-06,
      "loss": 0.0377,
      "step": 32350
    },
    {
      "epoch": 1.6424691608711102,
      "grad_norm": 0.2777571678161621,
      "learning_rate": 9.0502055941926e-06,
      "loss": 0.0413,
      "step": 32355
    },
    {
      "epoch": 1.6427229808619725,
      "grad_norm": 0.30235791206359863,
      "learning_rate": 9.048513460920184e-06,
      "loss": 0.0439,
      "step": 32360
    },
    {
      "epoch": 1.6429768008528352,
      "grad_norm": 0.40951234102249146,
      "learning_rate": 9.046821327647767e-06,
      "loss": 0.0468,
      "step": 32365
    },
    {
      "epoch": 1.6432306208436978,
      "grad_norm": 0.366265207529068,
      "learning_rate": 9.045129194375349e-06,
      "loss": 0.0324,
      "step": 32370
    },
    {
      "epoch": 1.6434844408345601,
      "grad_norm": 0.39244604110717773,
      "learning_rate": 9.043437061102932e-06,
      "loss": 0.0441,
      "step": 32375
    },
    {
      "epoch": 1.6437382608254225,
      "grad_norm": 0.3864644467830658,
      "learning_rate": 9.041744927830518e-06,
      "loss": 0.044,
      "step": 32380
    },
    {
      "epoch": 1.6439920808162851,
      "grad_norm": 0.4265846312046051,
      "learning_rate": 9.0400527945581e-06,
      "loss": 0.0417,
      "step": 32385
    },
    {
      "epoch": 1.6442459008071477,
      "grad_norm": 0.31097733974456787,
      "learning_rate": 9.038360661285683e-06,
      "loss": 0.0418,
      "step": 32390
    },
    {
      "epoch": 1.64449972079801,
      "grad_norm": 0.26527610421180725,
      "learning_rate": 9.036668528013267e-06,
      "loss": 0.0406,
      "step": 32395
    },
    {
      "epoch": 1.6447535407888725,
      "grad_norm": 0.3447258174419403,
      "learning_rate": 9.03497639474085e-06,
      "loss": 0.0448,
      "step": 32400
    },
    {
      "epoch": 1.645007360779735,
      "grad_norm": 0.4536108374595642,
      "learning_rate": 9.033284261468434e-06,
      "loss": 0.0449,
      "step": 32405
    },
    {
      "epoch": 1.6452611807705975,
      "grad_norm": 0.43144747614860535,
      "learning_rate": 9.031592128196017e-06,
      "loss": 0.0414,
      "step": 32410
    },
    {
      "epoch": 1.6455150007614598,
      "grad_norm": 0.26930710673332214,
      "learning_rate": 9.0298999949236e-06,
      "loss": 0.0469,
      "step": 32415
    },
    {
      "epoch": 1.6457688207523224,
      "grad_norm": 0.33055657148361206,
      "learning_rate": 9.028207861651184e-06,
      "loss": 0.0416,
      "step": 32420
    },
    {
      "epoch": 1.646022640743185,
      "grad_norm": 0.4396286606788635,
      "learning_rate": 9.026515728378768e-06,
      "loss": 0.0472,
      "step": 32425
    },
    {
      "epoch": 1.6462764607340474,
      "grad_norm": 0.3018118441104889,
      "learning_rate": 9.024823595106351e-06,
      "loss": 0.0372,
      "step": 32430
    },
    {
      "epoch": 1.6465302807249098,
      "grad_norm": 0.39372122287750244,
      "learning_rate": 9.023131461833935e-06,
      "loss": 0.0495,
      "step": 32435
    },
    {
      "epoch": 1.6467841007157724,
      "grad_norm": 0.2524060904979706,
      "learning_rate": 9.021439328561519e-06,
      "loss": 0.0392,
      "step": 32440
    },
    {
      "epoch": 1.647037920706635,
      "grad_norm": 0.5259925127029419,
      "learning_rate": 9.019747195289102e-06,
      "loss": 0.0401,
      "step": 32445
    },
    {
      "epoch": 1.6472917406974972,
      "grad_norm": 0.41529425978660583,
      "learning_rate": 9.018055062016686e-06,
      "loss": 0.0432,
      "step": 32450
    },
    {
      "epoch": 1.6475455606883598,
      "grad_norm": 0.3844156861305237,
      "learning_rate": 9.01636292874427e-06,
      "loss": 0.0361,
      "step": 32455
    },
    {
      "epoch": 1.6477993806792224,
      "grad_norm": 0.3346210718154907,
      "learning_rate": 9.014670795471853e-06,
      "loss": 0.0454,
      "step": 32460
    },
    {
      "epoch": 1.6480532006700848,
      "grad_norm": 0.3419366776943207,
      "learning_rate": 9.012978662199435e-06,
      "loss": 0.0426,
      "step": 32465
    },
    {
      "epoch": 1.6483070206609471,
      "grad_norm": 0.40948936343193054,
      "learning_rate": 9.01128652892702e-06,
      "loss": 0.0431,
      "step": 32470
    },
    {
      "epoch": 1.6485608406518097,
      "grad_norm": 0.3462151885032654,
      "learning_rate": 9.009594395654603e-06,
      "loss": 0.0424,
      "step": 32475
    },
    {
      "epoch": 1.6488146606426723,
      "grad_norm": 0.38817259669303894,
      "learning_rate": 9.007902262382185e-06,
      "loss": 0.0367,
      "step": 32480
    },
    {
      "epoch": 1.6490684806335347,
      "grad_norm": 0.33476361632347107,
      "learning_rate": 9.006210129109769e-06,
      "loss": 0.0459,
      "step": 32485
    },
    {
      "epoch": 1.649322300624397,
      "grad_norm": 0.31743040680885315,
      "learning_rate": 9.004517995837352e-06,
      "loss": 0.038,
      "step": 32490
    },
    {
      "epoch": 1.6495761206152597,
      "grad_norm": 0.3448827266693115,
      "learning_rate": 9.002825862564938e-06,
      "loss": 0.0388,
      "step": 32495
    },
    {
      "epoch": 1.649829940606122,
      "grad_norm": 0.277004599571228,
      "learning_rate": 9.00113372929252e-06,
      "loss": 0.044,
      "step": 32500
    },
    {
      "epoch": 1.6500837605969845,
      "grad_norm": 0.3807571232318878,
      "learning_rate": 8.999441596020103e-06,
      "loss": 0.0516,
      "step": 32505
    },
    {
      "epoch": 1.650337580587847,
      "grad_norm": 0.43307530879974365,
      "learning_rate": 8.997749462747686e-06,
      "loss": 0.0385,
      "step": 32510
    },
    {
      "epoch": 1.6505914005787097,
      "grad_norm": 0.36925190687179565,
      "learning_rate": 8.99605732947527e-06,
      "loss": 0.0406,
      "step": 32515
    },
    {
      "epoch": 1.650845220569572,
      "grad_norm": 0.37371015548706055,
      "learning_rate": 8.994365196202854e-06,
      "loss": 0.033,
      "step": 32520
    },
    {
      "epoch": 1.6510990405604344,
      "grad_norm": 0.3058145344257355,
      "learning_rate": 8.992673062930437e-06,
      "loss": 0.041,
      "step": 32525
    },
    {
      "epoch": 1.651352860551297,
      "grad_norm": 0.44341811537742615,
      "learning_rate": 8.99098092965802e-06,
      "loss": 0.0517,
      "step": 32530
    },
    {
      "epoch": 1.6516066805421596,
      "grad_norm": 0.4051477313041687,
      "learning_rate": 8.989288796385604e-06,
      "loss": 0.0444,
      "step": 32535
    },
    {
      "epoch": 1.651860500533022,
      "grad_norm": 0.4557914137840271,
      "learning_rate": 8.987596663113188e-06,
      "loss": 0.0471,
      "step": 32540
    },
    {
      "epoch": 1.6521143205238844,
      "grad_norm": 0.367385596036911,
      "learning_rate": 8.985904529840771e-06,
      "loss": 0.0434,
      "step": 32545
    },
    {
      "epoch": 1.652368140514747,
      "grad_norm": 0.31616297364234924,
      "learning_rate": 8.984212396568355e-06,
      "loss": 0.0441,
      "step": 32550
    },
    {
      "epoch": 1.6526219605056094,
      "grad_norm": 0.47400176525115967,
      "learning_rate": 8.982520263295937e-06,
      "loss": 0.0403,
      "step": 32555
    },
    {
      "epoch": 1.6528757804964718,
      "grad_norm": 0.37616056203842163,
      "learning_rate": 8.980828130023522e-06,
      "loss": 0.0516,
      "step": 32560
    },
    {
      "epoch": 1.6531296004873344,
      "grad_norm": 0.40422409772872925,
      "learning_rate": 8.979135996751105e-06,
      "loss": 0.0458,
      "step": 32565
    },
    {
      "epoch": 1.653383420478197,
      "grad_norm": 0.33850717544555664,
      "learning_rate": 8.977443863478689e-06,
      "loss": 0.0454,
      "step": 32570
    },
    {
      "epoch": 1.6536372404690594,
      "grad_norm": 0.2698414921760559,
      "learning_rate": 8.975751730206271e-06,
      "loss": 0.043,
      "step": 32575
    },
    {
      "epoch": 1.6538910604599217,
      "grad_norm": 0.3722778558731079,
      "learning_rate": 8.974059596933854e-06,
      "loss": 0.0477,
      "step": 32580
    },
    {
      "epoch": 1.6541448804507843,
      "grad_norm": 0.3686807453632355,
      "learning_rate": 8.97236746366144e-06,
      "loss": 0.0376,
      "step": 32585
    },
    {
      "epoch": 1.654398700441647,
      "grad_norm": 0.3176324963569641,
      "learning_rate": 8.970675330389023e-06,
      "loss": 0.0437,
      "step": 32590
    },
    {
      "epoch": 1.654652520432509,
      "grad_norm": 0.48339200019836426,
      "learning_rate": 8.968983197116605e-06,
      "loss": 0.047,
      "step": 32595
    },
    {
      "epoch": 1.6549063404233717,
      "grad_norm": 0.422828733921051,
      "learning_rate": 8.967291063844189e-06,
      "loss": 0.0413,
      "step": 32600
    },
    {
      "epoch": 1.6551601604142343,
      "grad_norm": 0.3320799469947815,
      "learning_rate": 8.965598930571772e-06,
      "loss": 0.0358,
      "step": 32605
    },
    {
      "epoch": 1.6554139804050967,
      "grad_norm": 0.8261854648590088,
      "learning_rate": 8.963906797299357e-06,
      "loss": 0.0414,
      "step": 32610
    },
    {
      "epoch": 1.655667800395959,
      "grad_norm": 0.3974917232990265,
      "learning_rate": 8.96221466402694e-06,
      "loss": 0.0353,
      "step": 32615
    },
    {
      "epoch": 1.6559216203868217,
      "grad_norm": 0.3724062144756317,
      "learning_rate": 8.960522530754523e-06,
      "loss": 0.0294,
      "step": 32620
    },
    {
      "epoch": 1.6561754403776843,
      "grad_norm": 0.4030728340148926,
      "learning_rate": 8.958830397482106e-06,
      "loss": 0.0429,
      "step": 32625
    },
    {
      "epoch": 1.6564292603685467,
      "grad_norm": 0.677786111831665,
      "learning_rate": 8.95713826420969e-06,
      "loss": 0.0495,
      "step": 32630
    },
    {
      "epoch": 1.656683080359409,
      "grad_norm": 0.21341340243816376,
      "learning_rate": 8.955446130937273e-06,
      "loss": 0.029,
      "step": 32635
    },
    {
      "epoch": 1.6569369003502716,
      "grad_norm": 0.45626917481422424,
      "learning_rate": 8.953753997664857e-06,
      "loss": 0.0363,
      "step": 32640
    },
    {
      "epoch": 1.657190720341134,
      "grad_norm": 0.29046255350112915,
      "learning_rate": 8.95206186439244e-06,
      "loss": 0.0442,
      "step": 32645
    },
    {
      "epoch": 1.6574445403319964,
      "grad_norm": 0.6526368260383606,
      "learning_rate": 8.950369731120024e-06,
      "loss": 0.0437,
      "step": 32650
    },
    {
      "epoch": 1.657698360322859,
      "grad_norm": 0.28871995210647583,
      "learning_rate": 8.948677597847608e-06,
      "loss": 0.0453,
      "step": 32655
    },
    {
      "epoch": 1.6579521803137216,
      "grad_norm": 0.3228037357330322,
      "learning_rate": 8.946985464575191e-06,
      "loss": 0.0454,
      "step": 32660
    },
    {
      "epoch": 1.658206000304584,
      "grad_norm": 0.24650952219963074,
      "learning_rate": 8.945293331302775e-06,
      "loss": 0.0368,
      "step": 32665
    },
    {
      "epoch": 1.6584598202954464,
      "grad_norm": 0.26511305570602417,
      "learning_rate": 8.943601198030357e-06,
      "loss": 0.0419,
      "step": 32670
    },
    {
      "epoch": 1.658713640286309,
      "grad_norm": 0.32455992698669434,
      "learning_rate": 8.941909064757942e-06,
      "loss": 0.0473,
      "step": 32675
    },
    {
      "epoch": 1.6589674602771716,
      "grad_norm": 0.3761235475540161,
      "learning_rate": 8.940216931485525e-06,
      "loss": 0.0417,
      "step": 32680
    },
    {
      "epoch": 1.659221280268034,
      "grad_norm": 0.2946665287017822,
      "learning_rate": 8.938524798213109e-06,
      "loss": 0.0433,
      "step": 32685
    },
    {
      "epoch": 1.6594751002588963,
      "grad_norm": 0.3487130403518677,
      "learning_rate": 8.93683266494069e-06,
      "loss": 0.0371,
      "step": 32690
    },
    {
      "epoch": 1.659728920249759,
      "grad_norm": 0.4113446772098541,
      "learning_rate": 8.935140531668274e-06,
      "loss": 0.0424,
      "step": 32695
    },
    {
      "epoch": 1.6599827402406213,
      "grad_norm": 0.22404730319976807,
      "learning_rate": 8.93344839839586e-06,
      "loss": 0.042,
      "step": 32700
    },
    {
      "epoch": 1.6602365602314837,
      "grad_norm": 0.3690703809261322,
      "learning_rate": 8.931756265123443e-06,
      "loss": 0.0503,
      "step": 32705
    },
    {
      "epoch": 1.6604903802223463,
      "grad_norm": 0.26985934376716614,
      "learning_rate": 8.930064131851025e-06,
      "loss": 0.0415,
      "step": 32710
    },
    {
      "epoch": 1.660744200213209,
      "grad_norm": 0.30366820096969604,
      "learning_rate": 8.928371998578608e-06,
      "loss": 0.0454,
      "step": 32715
    },
    {
      "epoch": 1.6609980202040713,
      "grad_norm": 0.3210146725177765,
      "learning_rate": 8.926679865306192e-06,
      "loss": 0.038,
      "step": 32720
    },
    {
      "epoch": 1.6612518401949337,
      "grad_norm": 0.38465768098831177,
      "learning_rate": 8.924987732033776e-06,
      "loss": 0.0393,
      "step": 32725
    },
    {
      "epoch": 1.6615056601857963,
      "grad_norm": 0.32459139823913574,
      "learning_rate": 8.923295598761359e-06,
      "loss": 0.0515,
      "step": 32730
    },
    {
      "epoch": 1.6617594801766589,
      "grad_norm": 0.2934326231479645,
      "learning_rate": 8.921603465488943e-06,
      "loss": 0.03,
      "step": 32735
    },
    {
      "epoch": 1.6620133001675212,
      "grad_norm": 0.35384735465049744,
      "learning_rate": 8.919911332216526e-06,
      "loss": 0.0426,
      "step": 32740
    },
    {
      "epoch": 1.6622671201583836,
      "grad_norm": 0.3279673159122467,
      "learning_rate": 8.91821919894411e-06,
      "loss": 0.0416,
      "step": 32745
    },
    {
      "epoch": 1.6625209401492462,
      "grad_norm": 0.36697596311569214,
      "learning_rate": 8.916527065671693e-06,
      "loss": 0.0477,
      "step": 32750
    },
    {
      "epoch": 1.6627747601401086,
      "grad_norm": 0.2286474108695984,
      "learning_rate": 8.914834932399277e-06,
      "loss": 0.043,
      "step": 32755
    },
    {
      "epoch": 1.663028580130971,
      "grad_norm": 0.33469220995903015,
      "learning_rate": 8.91314279912686e-06,
      "loss": 0.0415,
      "step": 32760
    },
    {
      "epoch": 1.6632824001218336,
      "grad_norm": 0.5135287642478943,
      "learning_rate": 8.911450665854444e-06,
      "loss": 0.04,
      "step": 32765
    },
    {
      "epoch": 1.6635362201126962,
      "grad_norm": 0.3337949216365814,
      "learning_rate": 8.909758532582027e-06,
      "loss": 0.0361,
      "step": 32770
    },
    {
      "epoch": 1.6637900401035586,
      "grad_norm": 0.3559700548648834,
      "learning_rate": 8.908066399309611e-06,
      "loss": 0.0422,
      "step": 32775
    },
    {
      "epoch": 1.664043860094421,
      "grad_norm": 0.44909822940826416,
      "learning_rate": 8.906374266037194e-06,
      "loss": 0.0415,
      "step": 32780
    },
    {
      "epoch": 1.6642976800852836,
      "grad_norm": 0.3622046113014221,
      "learning_rate": 8.904682132764776e-06,
      "loss": 0.0418,
      "step": 32785
    },
    {
      "epoch": 1.6645515000761462,
      "grad_norm": 0.32745665311813354,
      "learning_rate": 8.902989999492362e-06,
      "loss": 0.0423,
      "step": 32790
    },
    {
      "epoch": 1.6648053200670083,
      "grad_norm": 0.47989243268966675,
      "learning_rate": 8.901297866219945e-06,
      "loss": 0.05,
      "step": 32795
    },
    {
      "epoch": 1.665059140057871,
      "grad_norm": 0.2573833465576172,
      "learning_rate": 8.899605732947527e-06,
      "loss": 0.0363,
      "step": 32800
    },
    {
      "epoch": 1.6653129600487335,
      "grad_norm": 0.3501220643520355,
      "learning_rate": 8.89791359967511e-06,
      "loss": 0.0475,
      "step": 32805
    },
    {
      "epoch": 1.665566780039596,
      "grad_norm": 0.3637681305408478,
      "learning_rate": 8.896221466402694e-06,
      "loss": 0.0416,
      "step": 32810
    },
    {
      "epoch": 1.6658206000304583,
      "grad_norm": 0.46171268820762634,
      "learning_rate": 8.894529333130278e-06,
      "loss": 0.0412,
      "step": 32815
    },
    {
      "epoch": 1.666074420021321,
      "grad_norm": 0.3462655544281006,
      "learning_rate": 8.892837199857861e-06,
      "loss": 0.0339,
      "step": 32820
    },
    {
      "epoch": 1.6663282400121835,
      "grad_norm": 0.34022197127342224,
      "learning_rate": 8.891145066585445e-06,
      "loss": 0.0401,
      "step": 32825
    },
    {
      "epoch": 1.6665820600030459,
      "grad_norm": 0.31248193979263306,
      "learning_rate": 8.889452933313028e-06,
      "loss": 0.0404,
      "step": 32830
    },
    {
      "epoch": 1.6668358799939083,
      "grad_norm": 0.49435219168663025,
      "learning_rate": 8.887760800040612e-06,
      "loss": 0.0471,
      "step": 32835
    },
    {
      "epoch": 1.6670896999847709,
      "grad_norm": 0.47587648034095764,
      "learning_rate": 8.886068666768195e-06,
      "loss": 0.0406,
      "step": 32840
    },
    {
      "epoch": 1.6673435199756332,
      "grad_norm": 0.2598819434642792,
      "learning_rate": 8.884376533495779e-06,
      "loss": 0.037,
      "step": 32845
    },
    {
      "epoch": 1.6675973399664956,
      "grad_norm": 0.4888969361782074,
      "learning_rate": 8.882684400223362e-06,
      "loss": 0.0397,
      "step": 32850
    },
    {
      "epoch": 1.6678511599573582,
      "grad_norm": 0.37045595049858093,
      "learning_rate": 8.880992266950946e-06,
      "loss": 0.043,
      "step": 32855
    },
    {
      "epoch": 1.6681049799482208,
      "grad_norm": 0.3329499363899231,
      "learning_rate": 8.87930013367853e-06,
      "loss": 0.0469,
      "step": 32860
    },
    {
      "epoch": 1.6683587999390832,
      "grad_norm": 0.34843963384628296,
      "learning_rate": 8.877608000406113e-06,
      "loss": 0.0449,
      "step": 32865
    },
    {
      "epoch": 1.6686126199299456,
      "grad_norm": 0.45146849751472473,
      "learning_rate": 8.875915867133697e-06,
      "loss": 0.0378,
      "step": 32870
    },
    {
      "epoch": 1.6688664399208082,
      "grad_norm": 0.4130539298057556,
      "learning_rate": 8.874223733861278e-06,
      "loss": 0.0458,
      "step": 32875
    },
    {
      "epoch": 1.6691202599116708,
      "grad_norm": 0.32472503185272217,
      "learning_rate": 8.872531600588864e-06,
      "loss": 0.0354,
      "step": 32880
    },
    {
      "epoch": 1.6693740799025332,
      "grad_norm": 0.20518916845321655,
      "learning_rate": 8.870839467316447e-06,
      "loss": 0.0389,
      "step": 32885
    },
    {
      "epoch": 1.6696278998933956,
      "grad_norm": 0.31469863653182983,
      "learning_rate": 8.86914733404403e-06,
      "loss": 0.0447,
      "step": 32890
    },
    {
      "epoch": 1.6698817198842582,
      "grad_norm": 0.3323427736759186,
      "learning_rate": 8.867455200771613e-06,
      "loss": 0.0395,
      "step": 32895
    },
    {
      "epoch": 1.6701355398751205,
      "grad_norm": 0.5702518224716187,
      "learning_rate": 8.865763067499196e-06,
      "loss": 0.0468,
      "step": 32900
    },
    {
      "epoch": 1.670389359865983,
      "grad_norm": 0.4348698854446411,
      "learning_rate": 8.86407093422678e-06,
      "loss": 0.0465,
      "step": 32905
    },
    {
      "epoch": 1.6706431798568455,
      "grad_norm": 0.23130303621292114,
      "learning_rate": 8.862378800954365e-06,
      "loss": 0.0386,
      "step": 32910
    },
    {
      "epoch": 1.6708969998477081,
      "grad_norm": 0.3477792739868164,
      "learning_rate": 8.860686667681947e-06,
      "loss": 0.0371,
      "step": 32915
    },
    {
      "epoch": 1.6711508198385705,
      "grad_norm": 0.6751791834831238,
      "learning_rate": 8.85899453440953e-06,
      "loss": 0.0362,
      "step": 32920
    },
    {
      "epoch": 1.6714046398294329,
      "grad_norm": 0.29711753129959106,
      "learning_rate": 8.857302401137114e-06,
      "loss": 0.0425,
      "step": 32925
    },
    {
      "epoch": 1.6716584598202955,
      "grad_norm": 0.4057323634624481,
      "learning_rate": 8.855610267864697e-06,
      "loss": 0.0387,
      "step": 32930
    },
    {
      "epoch": 1.671912279811158,
      "grad_norm": 0.288831502199173,
      "learning_rate": 8.853918134592281e-06,
      "loss": 0.0411,
      "step": 32935
    },
    {
      "epoch": 1.6721660998020202,
      "grad_norm": 0.43437838554382324,
      "learning_rate": 8.852226001319865e-06,
      "loss": 0.0441,
      "step": 32940
    },
    {
      "epoch": 1.6724199197928828,
      "grad_norm": 0.3017238974571228,
      "learning_rate": 8.850533868047448e-06,
      "loss": 0.0374,
      "step": 32945
    },
    {
      "epoch": 1.6726737397837455,
      "grad_norm": 0.27872663736343384,
      "learning_rate": 8.848841734775032e-06,
      "loss": 0.037,
      "step": 32950
    },
    {
      "epoch": 1.6729275597746078,
      "grad_norm": 0.4081742763519287,
      "learning_rate": 8.847149601502615e-06,
      "loss": 0.0447,
      "step": 32955
    },
    {
      "epoch": 1.6731813797654702,
      "grad_norm": 0.4713224470615387,
      "learning_rate": 8.845457468230199e-06,
      "loss": 0.0402,
      "step": 32960
    },
    {
      "epoch": 1.6734351997563328,
      "grad_norm": 0.3346846401691437,
      "learning_rate": 8.843765334957782e-06,
      "loss": 0.045,
      "step": 32965
    },
    {
      "epoch": 1.6736890197471954,
      "grad_norm": 0.2436722368001938,
      "learning_rate": 8.842073201685366e-06,
      "loss": 0.0377,
      "step": 32970
    },
    {
      "epoch": 1.6739428397380578,
      "grad_norm": 0.43119826912879944,
      "learning_rate": 8.84038106841295e-06,
      "loss": 0.0514,
      "step": 32975
    },
    {
      "epoch": 1.6741966597289202,
      "grad_norm": 0.36432069540023804,
      "learning_rate": 8.838688935140533e-06,
      "loss": 0.0484,
      "step": 32980
    },
    {
      "epoch": 1.6744504797197828,
      "grad_norm": 0.2824365496635437,
      "learning_rate": 8.836996801868116e-06,
      "loss": 0.04,
      "step": 32985
    },
    {
      "epoch": 1.6747042997106452,
      "grad_norm": 0.38415277004241943,
      "learning_rate": 8.835304668595698e-06,
      "loss": 0.0335,
      "step": 32990
    },
    {
      "epoch": 1.6749581197015075,
      "grad_norm": 0.4238277077674866,
      "learning_rate": 8.833612535323282e-06,
      "loss": 0.0407,
      "step": 32995
    },
    {
      "epoch": 1.6752119396923701,
      "grad_norm": 0.49974820017814636,
      "learning_rate": 8.831920402050867e-06,
      "loss": 0.0413,
      "step": 33000
    },
    {
      "epoch": 1.6754657596832327,
      "grad_norm": 0.33166828751564026,
      "learning_rate": 8.83022826877845e-06,
      "loss": 0.0366,
      "step": 33005
    },
    {
      "epoch": 1.6757195796740951,
      "grad_norm": 0.4129970073699951,
      "learning_rate": 8.828536135506032e-06,
      "loss": 0.038,
      "step": 33010
    },
    {
      "epoch": 1.6759733996649575,
      "grad_norm": 0.23709560930728912,
      "learning_rate": 8.826844002233616e-06,
      "loss": 0.0352,
      "step": 33015
    },
    {
      "epoch": 1.67622721965582,
      "grad_norm": 0.37866827845573425,
      "learning_rate": 8.8251518689612e-06,
      "loss": 0.0447,
      "step": 33020
    },
    {
      "epoch": 1.6764810396466827,
      "grad_norm": 0.3766491711139679,
      "learning_rate": 8.823459735688785e-06,
      "loss": 0.0384,
      "step": 33025
    },
    {
      "epoch": 1.676734859637545,
      "grad_norm": 0.4483327567577362,
      "learning_rate": 8.821767602416367e-06,
      "loss": 0.0433,
      "step": 33030
    },
    {
      "epoch": 1.6769886796284075,
      "grad_norm": 0.33308711647987366,
      "learning_rate": 8.82007546914395e-06,
      "loss": 0.0307,
      "step": 33035
    },
    {
      "epoch": 1.67724249961927,
      "grad_norm": 0.3442741632461548,
      "learning_rate": 8.818383335871534e-06,
      "loss": 0.0411,
      "step": 33040
    },
    {
      "epoch": 1.6774963196101325,
      "grad_norm": 0.3743845820426941,
      "learning_rate": 8.816691202599117e-06,
      "loss": 0.0362,
      "step": 33045
    },
    {
      "epoch": 1.6777501396009948,
      "grad_norm": 0.35795456171035767,
      "learning_rate": 8.8149990693267e-06,
      "loss": 0.0414,
      "step": 33050
    },
    {
      "epoch": 1.6780039595918574,
      "grad_norm": 0.6182012557983398,
      "learning_rate": 8.813306936054284e-06,
      "loss": 0.0446,
      "step": 33055
    },
    {
      "epoch": 1.67825777958272,
      "grad_norm": 0.43161648511886597,
      "learning_rate": 8.811614802781868e-06,
      "loss": 0.0362,
      "step": 33060
    },
    {
      "epoch": 1.6785115995735824,
      "grad_norm": 0.31941068172454834,
      "learning_rate": 8.809922669509451e-06,
      "loss": 0.0414,
      "step": 33065
    },
    {
      "epoch": 1.6787654195644448,
      "grad_norm": 0.3838498890399933,
      "learning_rate": 8.808230536237035e-06,
      "loss": 0.0341,
      "step": 33070
    },
    {
      "epoch": 1.6790192395553074,
      "grad_norm": 0.38141605257987976,
      "learning_rate": 8.806538402964619e-06,
      "loss": 0.047,
      "step": 33075
    },
    {
      "epoch": 1.67927305954617,
      "grad_norm": 0.33075106143951416,
      "learning_rate": 8.804846269692202e-06,
      "loss": 0.0402,
      "step": 33080
    },
    {
      "epoch": 1.6795268795370324,
      "grad_norm": 0.32069718837738037,
      "learning_rate": 8.803154136419784e-06,
      "loss": 0.0408,
      "step": 33085
    },
    {
      "epoch": 1.6797806995278948,
      "grad_norm": 0.3646174371242523,
      "learning_rate": 8.80146200314737e-06,
      "loss": 0.0386,
      "step": 33090
    },
    {
      "epoch": 1.6800345195187574,
      "grad_norm": 0.30178675055503845,
      "learning_rate": 8.799769869874953e-06,
      "loss": 0.0466,
      "step": 33095
    },
    {
      "epoch": 1.6802883395096198,
      "grad_norm": 0.6580922603607178,
      "learning_rate": 8.798077736602536e-06,
      "loss": 0.0459,
      "step": 33100
    },
    {
      "epoch": 1.6805421595004821,
      "grad_norm": 0.48057985305786133,
      "learning_rate": 8.796385603330118e-06,
      "loss": 0.0482,
      "step": 33105
    },
    {
      "epoch": 1.6807959794913447,
      "grad_norm": 0.28072890639305115,
      "learning_rate": 8.794693470057702e-06,
      "loss": 0.0441,
      "step": 33110
    },
    {
      "epoch": 1.6810497994822073,
      "grad_norm": 0.27334949374198914,
      "learning_rate": 8.793001336785287e-06,
      "loss": 0.0333,
      "step": 33115
    },
    {
      "epoch": 1.6813036194730697,
      "grad_norm": 0.42797207832336426,
      "learning_rate": 8.791309203512869e-06,
      "loss": 0.041,
      "step": 33120
    },
    {
      "epoch": 1.681557439463932,
      "grad_norm": 0.3023119866847992,
      "learning_rate": 8.789617070240452e-06,
      "loss": 0.0367,
      "step": 33125
    },
    {
      "epoch": 1.6818112594547947,
      "grad_norm": 0.26819950342178345,
      "learning_rate": 8.787924936968036e-06,
      "loss": 0.0446,
      "step": 33130
    },
    {
      "epoch": 1.682065079445657,
      "grad_norm": 0.3586992919445038,
      "learning_rate": 8.78623280369562e-06,
      "loss": 0.0372,
      "step": 33135
    },
    {
      "epoch": 1.6823188994365195,
      "grad_norm": 0.35732945799827576,
      "learning_rate": 8.784540670423203e-06,
      "loss": 0.04,
      "step": 33140
    },
    {
      "epoch": 1.682572719427382,
      "grad_norm": 0.2647523581981659,
      "learning_rate": 8.782848537150786e-06,
      "loss": 0.0321,
      "step": 33145
    },
    {
      "epoch": 1.6828265394182447,
      "grad_norm": 0.39509347081184387,
      "learning_rate": 8.78115640387837e-06,
      "loss": 0.0396,
      "step": 33150
    },
    {
      "epoch": 1.683080359409107,
      "grad_norm": 0.3287479281425476,
      "learning_rate": 8.779464270605954e-06,
      "loss": 0.0356,
      "step": 33155
    },
    {
      "epoch": 1.6833341793999694,
      "grad_norm": 0.5807609558105469,
      "learning_rate": 8.777772137333537e-06,
      "loss": 0.0377,
      "step": 33160
    },
    {
      "epoch": 1.683587999390832,
      "grad_norm": 0.28490084409713745,
      "learning_rate": 8.77608000406112e-06,
      "loss": 0.0422,
      "step": 33165
    },
    {
      "epoch": 1.6838418193816946,
      "grad_norm": 0.37548598647117615,
      "learning_rate": 8.774387870788704e-06,
      "loss": 0.0453,
      "step": 33170
    },
    {
      "epoch": 1.684095639372557,
      "grad_norm": 0.6309858560562134,
      "learning_rate": 8.772695737516288e-06,
      "loss": 0.0359,
      "step": 33175
    },
    {
      "epoch": 1.6843494593634194,
      "grad_norm": 0.3599869906902313,
      "learning_rate": 8.771003604243871e-06,
      "loss": 0.0382,
      "step": 33180
    },
    {
      "epoch": 1.684603279354282,
      "grad_norm": 0.26254507899284363,
      "learning_rate": 8.769311470971455e-06,
      "loss": 0.0406,
      "step": 33185
    },
    {
      "epoch": 1.6848570993451444,
      "grad_norm": 0.3382471799850464,
      "learning_rate": 8.767619337699038e-06,
      "loss": 0.0412,
      "step": 33190
    },
    {
      "epoch": 1.6851109193360068,
      "grad_norm": 0.5124127268791199,
      "learning_rate": 8.76592720442662e-06,
      "loss": 0.0448,
      "step": 33195
    },
    {
      "epoch": 1.6853647393268694,
      "grad_norm": 0.31272459030151367,
      "learning_rate": 8.764235071154204e-06,
      "loss": 0.036,
      "step": 33200
    },
    {
      "epoch": 1.685618559317732,
      "grad_norm": 0.35631221532821655,
      "learning_rate": 8.762542937881789e-06,
      "loss": 0.0332,
      "step": 33205
    },
    {
      "epoch": 1.6858723793085943,
      "grad_norm": 0.2877142131328583,
      "learning_rate": 8.760850804609373e-06,
      "loss": 0.0446,
      "step": 33210
    },
    {
      "epoch": 1.6861261992994567,
      "grad_norm": 0.44849103689193726,
      "learning_rate": 8.759158671336954e-06,
      "loss": 0.0372,
      "step": 33215
    },
    {
      "epoch": 1.6863800192903193,
      "grad_norm": 0.4658852517604828,
      "learning_rate": 8.757466538064538e-06,
      "loss": 0.0372,
      "step": 33220
    },
    {
      "epoch": 1.686633839281182,
      "grad_norm": 0.6611411571502686,
      "learning_rate": 8.755774404792121e-06,
      "loss": 0.0402,
      "step": 33225
    },
    {
      "epoch": 1.6868876592720443,
      "grad_norm": 0.3149184584617615,
      "learning_rate": 8.754082271519707e-06,
      "loss": 0.0437,
      "step": 33230
    },
    {
      "epoch": 1.6871414792629067,
      "grad_norm": 0.31619152426719666,
      "learning_rate": 8.752390138247289e-06,
      "loss": 0.0393,
      "step": 33235
    },
    {
      "epoch": 1.6873952992537693,
      "grad_norm": 0.35493898391723633,
      "learning_rate": 8.750698004974872e-06,
      "loss": 0.0385,
      "step": 33240
    },
    {
      "epoch": 1.6876491192446317,
      "grad_norm": 0.48094281554222107,
      "learning_rate": 8.749005871702456e-06,
      "loss": 0.0413,
      "step": 33245
    },
    {
      "epoch": 1.687902939235494,
      "grad_norm": 0.26521581411361694,
      "learning_rate": 8.74731373843004e-06,
      "loss": 0.0373,
      "step": 33250
    },
    {
      "epoch": 1.6881567592263567,
      "grad_norm": 0.3604218065738678,
      "learning_rate": 8.745621605157623e-06,
      "loss": 0.0418,
      "step": 33255
    },
    {
      "epoch": 1.6884105792172193,
      "grad_norm": 0.5107678174972534,
      "learning_rate": 8.743929471885206e-06,
      "loss": 0.0396,
      "step": 33260
    },
    {
      "epoch": 1.6886643992080816,
      "grad_norm": 0.2442566156387329,
      "learning_rate": 8.74223733861279e-06,
      "loss": 0.038,
      "step": 33265
    },
    {
      "epoch": 1.688918219198944,
      "grad_norm": 0.3603915572166443,
      "learning_rate": 8.740545205340373e-06,
      "loss": 0.0438,
      "step": 33270
    },
    {
      "epoch": 1.6891720391898066,
      "grad_norm": 0.3329687714576721,
      "learning_rate": 8.738853072067957e-06,
      "loss": 0.0483,
      "step": 33275
    },
    {
      "epoch": 1.6894258591806692,
      "grad_norm": 0.2893288731575012,
      "learning_rate": 8.73716093879554e-06,
      "loss": 0.0381,
      "step": 33280
    },
    {
      "epoch": 1.6896796791715314,
      "grad_norm": 0.55464106798172,
      "learning_rate": 8.735468805523124e-06,
      "loss": 0.0475,
      "step": 33285
    },
    {
      "epoch": 1.689933499162394,
      "grad_norm": 0.8349003791809082,
      "learning_rate": 8.733776672250706e-06,
      "loss": 0.0399,
      "step": 33290
    },
    {
      "epoch": 1.6901873191532566,
      "grad_norm": 0.2722102105617523,
      "learning_rate": 8.732084538978291e-06,
      "loss": 0.0358,
      "step": 33295
    },
    {
      "epoch": 1.690441139144119,
      "grad_norm": 0.27093610167503357,
      "learning_rate": 8.730392405705875e-06,
      "loss": 0.0339,
      "step": 33300
    },
    {
      "epoch": 1.6906949591349814,
      "grad_norm": 0.3764627277851105,
      "learning_rate": 8.728700272433458e-06,
      "loss": 0.0397,
      "step": 33305
    },
    {
      "epoch": 1.690948779125844,
      "grad_norm": 0.3306204378604889,
      "learning_rate": 8.72700813916104e-06,
      "loss": 0.042,
      "step": 33310
    },
    {
      "epoch": 1.6912025991167066,
      "grad_norm": 0.2610977292060852,
      "learning_rate": 8.725316005888624e-06,
      "loss": 0.037,
      "step": 33315
    },
    {
      "epoch": 1.691456419107569,
      "grad_norm": 0.3594244420528412,
      "learning_rate": 8.723623872616209e-06,
      "loss": 0.0453,
      "step": 33320
    },
    {
      "epoch": 1.6917102390984313,
      "grad_norm": 0.31486913561820984,
      "learning_rate": 8.721931739343792e-06,
      "loss": 0.044,
      "step": 33325
    },
    {
      "epoch": 1.691964059089294,
      "grad_norm": 0.3887704610824585,
      "learning_rate": 8.720239606071374e-06,
      "loss": 0.0454,
      "step": 33330
    },
    {
      "epoch": 1.6922178790801563,
      "grad_norm": 0.3932081460952759,
      "learning_rate": 8.718547472798958e-06,
      "loss": 0.0446,
      "step": 33335
    },
    {
      "epoch": 1.6924716990710187,
      "grad_norm": 0.35229092836380005,
      "learning_rate": 8.716855339526541e-06,
      "loss": 0.0395,
      "step": 33340
    },
    {
      "epoch": 1.6927255190618813,
      "grad_norm": 0.3832239508628845,
      "learning_rate": 8.715163206254127e-06,
      "loss": 0.0487,
      "step": 33345
    },
    {
      "epoch": 1.692979339052744,
      "grad_norm": 0.3000941872596741,
      "learning_rate": 8.713471072981708e-06,
      "loss": 0.0329,
      "step": 33350
    },
    {
      "epoch": 1.6932331590436063,
      "grad_norm": 0.4978597164154053,
      "learning_rate": 8.711778939709292e-06,
      "loss": 0.046,
      "step": 33355
    },
    {
      "epoch": 1.6934869790344687,
      "grad_norm": 0.35227295756340027,
      "learning_rate": 8.710086806436875e-06,
      "loss": 0.0385,
      "step": 33360
    },
    {
      "epoch": 1.6937407990253313,
      "grad_norm": 0.46936389803886414,
      "learning_rate": 8.708394673164459e-06,
      "loss": 0.0437,
      "step": 33365
    },
    {
      "epoch": 1.6939946190161939,
      "grad_norm": 0.2642430067062378,
      "learning_rate": 8.706702539892043e-06,
      "loss": 0.049,
      "step": 33370
    },
    {
      "epoch": 1.6942484390070562,
      "grad_norm": 0.2733605206012726,
      "learning_rate": 8.705010406619626e-06,
      "loss": 0.0446,
      "step": 33375
    },
    {
      "epoch": 1.6945022589979186,
      "grad_norm": 0.42943286895751953,
      "learning_rate": 8.70331827334721e-06,
      "loss": 0.0432,
      "step": 33380
    },
    {
      "epoch": 1.6947560789887812,
      "grad_norm": 0.32281479239463806,
      "learning_rate": 8.701626140074793e-06,
      "loss": 0.0397,
      "step": 33385
    },
    {
      "epoch": 1.6950098989796436,
      "grad_norm": 0.2502782940864563,
      "learning_rate": 8.699934006802377e-06,
      "loss": 0.0429,
      "step": 33390
    },
    {
      "epoch": 1.695263718970506,
      "grad_norm": 0.34502100944519043,
      "learning_rate": 8.69824187352996e-06,
      "loss": 0.0473,
      "step": 33395
    },
    {
      "epoch": 1.6955175389613686,
      "grad_norm": 0.3549754023551941,
      "learning_rate": 8.696549740257544e-06,
      "loss": 0.0456,
      "step": 33400
    },
    {
      "epoch": 1.6957713589522312,
      "grad_norm": 0.4002918303012848,
      "learning_rate": 8.694857606985126e-06,
      "loss": 0.0429,
      "step": 33405
    },
    {
      "epoch": 1.6960251789430936,
      "grad_norm": 0.351992130279541,
      "learning_rate": 8.693165473712711e-06,
      "loss": 0.0405,
      "step": 33410
    },
    {
      "epoch": 1.696278998933956,
      "grad_norm": 0.27613869309425354,
      "learning_rate": 8.691473340440294e-06,
      "loss": 0.0433,
      "step": 33415
    },
    {
      "epoch": 1.6965328189248186,
      "grad_norm": 0.42446577548980713,
      "learning_rate": 8.689781207167878e-06,
      "loss": 0.0427,
      "step": 33420
    },
    {
      "epoch": 1.6967866389156812,
      "grad_norm": 0.32727229595184326,
      "learning_rate": 8.68808907389546e-06,
      "loss": 0.0358,
      "step": 33425
    },
    {
      "epoch": 1.6970404589065433,
      "grad_norm": 0.1883794367313385,
      "learning_rate": 8.686396940623043e-06,
      "loss": 0.0464,
      "step": 33430
    },
    {
      "epoch": 1.697294278897406,
      "grad_norm": 0.27226707339286804,
      "learning_rate": 8.684704807350629e-06,
      "loss": 0.0349,
      "step": 33435
    },
    {
      "epoch": 1.6975480988882685,
      "grad_norm": 0.4325341284275055,
      "learning_rate": 8.68301267407821e-06,
      "loss": 0.0471,
      "step": 33440
    },
    {
      "epoch": 1.697801918879131,
      "grad_norm": 0.33074063062667847,
      "learning_rate": 8.681320540805794e-06,
      "loss": 0.031,
      "step": 33445
    },
    {
      "epoch": 1.6980557388699933,
      "grad_norm": 0.3044057786464691,
      "learning_rate": 8.679628407533378e-06,
      "loss": 0.043,
      "step": 33450
    },
    {
      "epoch": 1.6983095588608559,
      "grad_norm": 0.43691715598106384,
      "learning_rate": 8.677936274260961e-06,
      "loss": 0.0411,
      "step": 33455
    },
    {
      "epoch": 1.6985633788517185,
      "grad_norm": 0.3442010283470154,
      "learning_rate": 8.676244140988545e-06,
      "loss": 0.0408,
      "step": 33460
    },
    {
      "epoch": 1.6988171988425809,
      "grad_norm": 0.24146975576877594,
      "learning_rate": 8.674552007716128e-06,
      "loss": 0.038,
      "step": 33465
    },
    {
      "epoch": 1.6990710188334432,
      "grad_norm": 0.28107747435569763,
      "learning_rate": 8.672859874443712e-06,
      "loss": 0.0404,
      "step": 33470
    },
    {
      "epoch": 1.6993248388243058,
      "grad_norm": 0.3662523627281189,
      "learning_rate": 8.671167741171295e-06,
      "loss": 0.0434,
      "step": 33475
    },
    {
      "epoch": 1.6995786588151682,
      "grad_norm": 0.3790246546268463,
      "learning_rate": 8.669475607898879e-06,
      "loss": 0.0359,
      "step": 33480
    },
    {
      "epoch": 1.6998324788060306,
      "grad_norm": 0.3024158179759979,
      "learning_rate": 8.667783474626462e-06,
      "loss": 0.0436,
      "step": 33485
    },
    {
      "epoch": 1.7000862987968932,
      "grad_norm": 0.5186987519264221,
      "learning_rate": 8.666091341354046e-06,
      "loss": 0.0437,
      "step": 33490
    },
    {
      "epoch": 1.7003401187877558,
      "grad_norm": 0.25226029753685,
      "learning_rate": 8.66439920808163e-06,
      "loss": 0.0431,
      "step": 33495
    },
    {
      "epoch": 1.7005939387786182,
      "grad_norm": 0.4354790151119232,
      "learning_rate": 8.662707074809213e-06,
      "loss": 0.0469,
      "step": 33500
    },
    {
      "epoch": 1.7008477587694806,
      "grad_norm": 0.2978040874004364,
      "learning_rate": 8.661014941536797e-06,
      "loss": 0.037,
      "step": 33505
    },
    {
      "epoch": 1.7011015787603432,
      "grad_norm": 0.36080655455589294,
      "learning_rate": 8.65932280826438e-06,
      "loss": 0.0396,
      "step": 33510
    },
    {
      "epoch": 1.7013553987512058,
      "grad_norm": 0.323999285697937,
      "learning_rate": 8.657630674991962e-06,
      "loss": 0.0389,
      "step": 33515
    },
    {
      "epoch": 1.7016092187420682,
      "grad_norm": 0.3238927721977234,
      "learning_rate": 8.655938541719546e-06,
      "loss": 0.0394,
      "step": 33520
    },
    {
      "epoch": 1.7018630387329305,
      "grad_norm": 0.30916842818260193,
      "learning_rate": 8.65424640844713e-06,
      "loss": 0.0367,
      "step": 33525
    },
    {
      "epoch": 1.7021168587237931,
      "grad_norm": 0.30041056871414185,
      "learning_rate": 8.652554275174714e-06,
      "loss": 0.0382,
      "step": 33530
    },
    {
      "epoch": 1.7023706787146555,
      "grad_norm": 0.40629908442497253,
      "learning_rate": 8.650862141902296e-06,
      "loss": 0.0408,
      "step": 33535
    },
    {
      "epoch": 1.702624498705518,
      "grad_norm": 0.3429891765117645,
      "learning_rate": 8.64917000862988e-06,
      "loss": 0.0355,
      "step": 33540
    },
    {
      "epoch": 1.7028783186963805,
      "grad_norm": 0.5190361142158508,
      "learning_rate": 8.647477875357463e-06,
      "loss": 0.04,
      "step": 33545
    },
    {
      "epoch": 1.7031321386872431,
      "grad_norm": 0.36994943022727966,
      "learning_rate": 8.645785742085048e-06,
      "loss": 0.04,
      "step": 33550
    },
    {
      "epoch": 1.7033859586781055,
      "grad_norm": 0.4087146818637848,
      "learning_rate": 8.64409360881263e-06,
      "loss": 0.0482,
      "step": 33555
    },
    {
      "epoch": 1.7036397786689679,
      "grad_norm": 0.44062137603759766,
      "learning_rate": 8.642401475540214e-06,
      "loss": 0.0395,
      "step": 33560
    },
    {
      "epoch": 1.7038935986598305,
      "grad_norm": 0.4262305498123169,
      "learning_rate": 8.640709342267797e-06,
      "loss": 0.0402,
      "step": 33565
    },
    {
      "epoch": 1.704147418650693,
      "grad_norm": 0.30454927682876587,
      "learning_rate": 8.639017208995381e-06,
      "loss": 0.0401,
      "step": 33570
    },
    {
      "epoch": 1.7044012386415555,
      "grad_norm": 0.3262251317501068,
      "learning_rate": 8.637325075722965e-06,
      "loss": 0.0435,
      "step": 33575
    },
    {
      "epoch": 1.7046550586324178,
      "grad_norm": 0.2448095828294754,
      "learning_rate": 8.635632942450548e-06,
      "loss": 0.0401,
      "step": 33580
    },
    {
      "epoch": 1.7049088786232804,
      "grad_norm": 0.3910999894142151,
      "learning_rate": 8.633940809178132e-06,
      "loss": 0.0417,
      "step": 33585
    },
    {
      "epoch": 1.7051626986141428,
      "grad_norm": 0.2758971154689789,
      "learning_rate": 8.632248675905715e-06,
      "loss": 0.0458,
      "step": 33590
    },
    {
      "epoch": 1.7054165186050052,
      "grad_norm": 0.27816465497016907,
      "learning_rate": 8.630556542633299e-06,
      "loss": 0.0417,
      "step": 33595
    },
    {
      "epoch": 1.7056703385958678,
      "grad_norm": 0.30175527930259705,
      "learning_rate": 8.628864409360882e-06,
      "loss": 0.0429,
      "step": 33600
    },
    {
      "epoch": 1.7059241585867304,
      "grad_norm": 0.2667434811592102,
      "learning_rate": 8.627172276088466e-06,
      "loss": 0.0343,
      "step": 33605
    },
    {
      "epoch": 1.7061779785775928,
      "grad_norm": 0.2773091495037079,
      "learning_rate": 8.625480142816048e-06,
      "loss": 0.0444,
      "step": 33610
    },
    {
      "epoch": 1.7064317985684552,
      "grad_norm": 0.29049497842788696,
      "learning_rate": 8.623788009543633e-06,
      "loss": 0.0352,
      "step": 33615
    },
    {
      "epoch": 1.7066856185593178,
      "grad_norm": 0.4345710873603821,
      "learning_rate": 8.622095876271216e-06,
      "loss": 0.0348,
      "step": 33620
    },
    {
      "epoch": 1.7069394385501804,
      "grad_norm": 0.43229636549949646,
      "learning_rate": 8.6204037429988e-06,
      "loss": 0.0362,
      "step": 33625
    },
    {
      "epoch": 1.7071932585410425,
      "grad_norm": 0.4182978570461273,
      "learning_rate": 8.618711609726382e-06,
      "loss": 0.0466,
      "step": 33630
    },
    {
      "epoch": 1.7074470785319051,
      "grad_norm": 0.36239463090896606,
      "learning_rate": 8.617019476453965e-06,
      "loss": 0.0415,
      "step": 33635
    },
    {
      "epoch": 1.7077008985227677,
      "grad_norm": 0.4394817650318146,
      "learning_rate": 8.61532734318155e-06,
      "loss": 0.0393,
      "step": 33640
    },
    {
      "epoch": 1.7079547185136301,
      "grad_norm": 0.323661208152771,
      "learning_rate": 8.613635209909134e-06,
      "loss": 0.0381,
      "step": 33645
    },
    {
      "epoch": 1.7082085385044925,
      "grad_norm": 0.6629905700683594,
      "learning_rate": 8.611943076636716e-06,
      "loss": 0.0374,
      "step": 33650
    },
    {
      "epoch": 1.708462358495355,
      "grad_norm": 0.31542661786079407,
      "learning_rate": 8.6102509433643e-06,
      "loss": 0.0445,
      "step": 33655
    },
    {
      "epoch": 1.7087161784862177,
      "grad_norm": 0.24203170835971832,
      "learning_rate": 8.608558810091883e-06,
      "loss": 0.0349,
      "step": 33660
    },
    {
      "epoch": 1.70896999847708,
      "grad_norm": 0.38370826840400696,
      "learning_rate": 8.606866676819468e-06,
      "loss": 0.0479,
      "step": 33665
    },
    {
      "epoch": 1.7092238184679425,
      "grad_norm": 0.5466206073760986,
      "learning_rate": 8.60517454354705e-06,
      "loss": 0.0398,
      "step": 33670
    },
    {
      "epoch": 1.709477638458805,
      "grad_norm": 0.3482562303543091,
      "learning_rate": 8.603482410274634e-06,
      "loss": 0.0326,
      "step": 33675
    },
    {
      "epoch": 1.7097314584496675,
      "grad_norm": 0.33356785774230957,
      "learning_rate": 8.601790277002217e-06,
      "loss": 0.0493,
      "step": 33680
    },
    {
      "epoch": 1.7099852784405298,
      "grad_norm": 0.24861811101436615,
      "learning_rate": 8.6000981437298e-06,
      "loss": 0.0498,
      "step": 33685
    },
    {
      "epoch": 1.7102390984313924,
      "grad_norm": 0.2985035479068756,
      "learning_rate": 8.598406010457384e-06,
      "loss": 0.0345,
      "step": 33690
    },
    {
      "epoch": 1.710492918422255,
      "grad_norm": 0.34388431906700134,
      "learning_rate": 8.596713877184968e-06,
      "loss": 0.0436,
      "step": 33695
    },
    {
      "epoch": 1.7107467384131174,
      "grad_norm": 0.28177011013031006,
      "learning_rate": 8.595021743912551e-06,
      "loss": 0.0412,
      "step": 33700
    },
    {
      "epoch": 1.7110005584039798,
      "grad_norm": 0.29938557744026184,
      "learning_rate": 8.593329610640135e-06,
      "loss": 0.0419,
      "step": 33705
    },
    {
      "epoch": 1.7112543783948424,
      "grad_norm": 0.3993265628814697,
      "learning_rate": 8.591637477367719e-06,
      "loss": 0.0424,
      "step": 33710
    },
    {
      "epoch": 1.711508198385705,
      "grad_norm": 0.2539103925228119,
      "learning_rate": 8.589945344095302e-06,
      "loss": 0.0402,
      "step": 33715
    },
    {
      "epoch": 1.7117620183765674,
      "grad_norm": 0.3944316506385803,
      "learning_rate": 8.588253210822886e-06,
      "loss": 0.049,
      "step": 33720
    },
    {
      "epoch": 1.7120158383674298,
      "grad_norm": 0.3147892355918884,
      "learning_rate": 8.586561077550467e-06,
      "loss": 0.0457,
      "step": 33725
    },
    {
      "epoch": 1.7122696583582924,
      "grad_norm": 0.3464619517326355,
      "learning_rate": 8.584868944278053e-06,
      "loss": 0.041,
      "step": 33730
    },
    {
      "epoch": 1.7125234783491547,
      "grad_norm": 0.33929938077926636,
      "learning_rate": 8.583176811005636e-06,
      "loss": 0.0436,
      "step": 33735
    },
    {
      "epoch": 1.7127772983400171,
      "grad_norm": 0.28738826513290405,
      "learning_rate": 8.58148467773322e-06,
      "loss": 0.0445,
      "step": 33740
    },
    {
      "epoch": 1.7130311183308797,
      "grad_norm": 0.40138235688209534,
      "learning_rate": 8.579792544460802e-06,
      "loss": 0.0483,
      "step": 33745
    },
    {
      "epoch": 1.7132849383217423,
      "grad_norm": 0.3729156255722046,
      "learning_rate": 8.578100411188385e-06,
      "loss": 0.0422,
      "step": 33750
    },
    {
      "epoch": 1.7135387583126047,
      "grad_norm": 0.32851508259773254,
      "learning_rate": 8.57640827791597e-06,
      "loss": 0.0401,
      "step": 33755
    },
    {
      "epoch": 1.713792578303467,
      "grad_norm": 0.3027333617210388,
      "learning_rate": 8.574716144643552e-06,
      "loss": 0.0483,
      "step": 33760
    },
    {
      "epoch": 1.7140463982943297,
      "grad_norm": 0.305976539850235,
      "learning_rate": 8.573024011371136e-06,
      "loss": 0.0382,
      "step": 33765
    },
    {
      "epoch": 1.7143002182851923,
      "grad_norm": 0.3750983774662018,
      "learning_rate": 8.57133187809872e-06,
      "loss": 0.0482,
      "step": 33770
    },
    {
      "epoch": 1.7145540382760545,
      "grad_norm": 0.4367023706436157,
      "learning_rate": 8.569639744826303e-06,
      "loss": 0.0475,
      "step": 33775
    },
    {
      "epoch": 1.714807858266917,
      "grad_norm": 0.5546810626983643,
      "learning_rate": 8.567947611553886e-06,
      "loss": 0.051,
      "step": 33780
    },
    {
      "epoch": 1.7150616782577797,
      "grad_norm": 0.3331943452358246,
      "learning_rate": 8.56625547828147e-06,
      "loss": 0.0453,
      "step": 33785
    },
    {
      "epoch": 1.715315498248642,
      "grad_norm": 0.5244970321655273,
      "learning_rate": 8.564563345009054e-06,
      "loss": 0.0505,
      "step": 33790
    },
    {
      "epoch": 1.7155693182395044,
      "grad_norm": 0.35232967138290405,
      "learning_rate": 8.562871211736637e-06,
      "loss": 0.0435,
      "step": 33795
    },
    {
      "epoch": 1.715823138230367,
      "grad_norm": 0.24759510159492493,
      "learning_rate": 8.56117907846422e-06,
      "loss": 0.0415,
      "step": 33800
    },
    {
      "epoch": 1.7160769582212296,
      "grad_norm": 0.32473260164260864,
      "learning_rate": 8.559486945191804e-06,
      "loss": 0.0373,
      "step": 33805
    },
    {
      "epoch": 1.716330778212092,
      "grad_norm": 0.5221812129020691,
      "learning_rate": 8.557794811919388e-06,
      "loss": 0.0352,
      "step": 33810
    },
    {
      "epoch": 1.7165845982029544,
      "grad_norm": 0.32362228631973267,
      "learning_rate": 8.556102678646971e-06,
      "loss": 0.0428,
      "step": 33815
    },
    {
      "epoch": 1.716838418193817,
      "grad_norm": 0.3280642628669739,
      "learning_rate": 8.554410545374555e-06,
      "loss": 0.0423,
      "step": 33820
    },
    {
      "epoch": 1.7170922381846794,
      "grad_norm": 0.38787609338760376,
      "learning_rate": 8.552718412102138e-06,
      "loss": 0.0425,
      "step": 33825
    },
    {
      "epoch": 1.7173460581755418,
      "grad_norm": 0.39664000272750854,
      "learning_rate": 8.551026278829722e-06,
      "loss": 0.0419,
      "step": 33830
    },
    {
      "epoch": 1.7175998781664044,
      "grad_norm": 0.3131723999977112,
      "learning_rate": 8.549334145557305e-06,
      "loss": 0.0395,
      "step": 33835
    },
    {
      "epoch": 1.717853698157267,
      "grad_norm": 0.374670147895813,
      "learning_rate": 8.547642012284887e-06,
      "loss": 0.0365,
      "step": 33840
    },
    {
      "epoch": 1.7181075181481293,
      "grad_norm": 0.37094488739967346,
      "learning_rate": 8.545949879012473e-06,
      "loss": 0.0479,
      "step": 33845
    },
    {
      "epoch": 1.7183613381389917,
      "grad_norm": 0.30541858077049255,
      "learning_rate": 8.544257745740056e-06,
      "loss": 0.0438,
      "step": 33850
    },
    {
      "epoch": 1.7186151581298543,
      "grad_norm": 0.3539015054702759,
      "learning_rate": 8.542565612467638e-06,
      "loss": 0.0447,
      "step": 33855
    },
    {
      "epoch": 1.718868978120717,
      "grad_norm": 0.23813435435295105,
      "learning_rate": 8.540873479195221e-06,
      "loss": 0.0373,
      "step": 33860
    },
    {
      "epoch": 1.7191227981115793,
      "grad_norm": 0.2674414813518524,
      "learning_rate": 8.539181345922805e-06,
      "loss": 0.0461,
      "step": 33865
    },
    {
      "epoch": 1.7193766181024417,
      "grad_norm": 0.37278127670288086,
      "learning_rate": 8.53748921265039e-06,
      "loss": 0.0405,
      "step": 33870
    },
    {
      "epoch": 1.7196304380933043,
      "grad_norm": 0.268686443567276,
      "learning_rate": 8.535797079377972e-06,
      "loss": 0.0375,
      "step": 33875
    },
    {
      "epoch": 1.7198842580841667,
      "grad_norm": 0.3514656722545624,
      "learning_rate": 8.534104946105556e-06,
      "loss": 0.0381,
      "step": 33880
    },
    {
      "epoch": 1.720138078075029,
      "grad_norm": 0.2861263155937195,
      "learning_rate": 8.53241281283314e-06,
      "loss": 0.0442,
      "step": 33885
    },
    {
      "epoch": 1.7203918980658917,
      "grad_norm": 0.375036358833313,
      "learning_rate": 8.530720679560723e-06,
      "loss": 0.0324,
      "step": 33890
    },
    {
      "epoch": 1.7206457180567543,
      "grad_norm": 0.2760937511920929,
      "learning_rate": 8.529028546288306e-06,
      "loss": 0.0402,
      "step": 33895
    },
    {
      "epoch": 1.7208995380476166,
      "grad_norm": 0.3707115948200226,
      "learning_rate": 8.52733641301589e-06,
      "loss": 0.0406,
      "step": 33900
    },
    {
      "epoch": 1.721153358038479,
      "grad_norm": 0.31657668948173523,
      "learning_rate": 8.525644279743473e-06,
      "loss": 0.0423,
      "step": 33905
    },
    {
      "epoch": 1.7214071780293416,
      "grad_norm": 0.45224529504776,
      "learning_rate": 8.523952146471057e-06,
      "loss": 0.0406,
      "step": 33910
    },
    {
      "epoch": 1.7216609980202042,
      "grad_norm": 0.3776727616786957,
      "learning_rate": 8.52226001319864e-06,
      "loss": 0.0416,
      "step": 33915
    },
    {
      "epoch": 1.7219148180110666,
      "grad_norm": 0.32330262660980225,
      "learning_rate": 8.520567879926224e-06,
      "loss": 0.0366,
      "step": 33920
    },
    {
      "epoch": 1.722168638001929,
      "grad_norm": 0.3480754494667053,
      "learning_rate": 8.518875746653808e-06,
      "loss": 0.0437,
      "step": 33925
    },
    {
      "epoch": 1.7224224579927916,
      "grad_norm": 0.2876611053943634,
      "learning_rate": 8.51718361338139e-06,
      "loss": 0.0355,
      "step": 33930
    },
    {
      "epoch": 1.722676277983654,
      "grad_norm": 0.24682271480560303,
      "learning_rate": 8.515491480108975e-06,
      "loss": 0.0329,
      "step": 33935
    },
    {
      "epoch": 1.7229300979745164,
      "grad_norm": 0.22861187160015106,
      "learning_rate": 8.513799346836558e-06,
      "loss": 0.044,
      "step": 33940
    },
    {
      "epoch": 1.723183917965379,
      "grad_norm": 0.4073830246925354,
      "learning_rate": 8.512107213564142e-06,
      "loss": 0.0445,
      "step": 33945
    },
    {
      "epoch": 1.7234377379562416,
      "grad_norm": 0.2965492308139801,
      "learning_rate": 8.510415080291724e-06,
      "loss": 0.0381,
      "step": 33950
    },
    {
      "epoch": 1.723691557947104,
      "grad_norm": 0.2638970911502838,
      "learning_rate": 8.508722947019307e-06,
      "loss": 0.0382,
      "step": 33955
    },
    {
      "epoch": 1.7239453779379663,
      "grad_norm": 0.35170966386795044,
      "learning_rate": 8.507030813746892e-06,
      "loss": 0.0375,
      "step": 33960
    },
    {
      "epoch": 1.724199197928829,
      "grad_norm": 0.3397308588027954,
      "learning_rate": 8.505338680474476e-06,
      "loss": 0.034,
      "step": 33965
    },
    {
      "epoch": 1.7244530179196913,
      "grad_norm": 0.47272270917892456,
      "learning_rate": 8.503646547202058e-06,
      "loss": 0.0417,
      "step": 33970
    },
    {
      "epoch": 1.7247068379105537,
      "grad_norm": 0.3830793499946594,
      "learning_rate": 8.501954413929641e-06,
      "loss": 0.0333,
      "step": 33975
    },
    {
      "epoch": 1.7249606579014163,
      "grad_norm": 0.5919451713562012,
      "learning_rate": 8.500262280657225e-06,
      "loss": 0.0412,
      "step": 33980
    },
    {
      "epoch": 1.7252144778922789,
      "grad_norm": 0.3923093378543854,
      "learning_rate": 8.498570147384808e-06,
      "loss": 0.0414,
      "step": 33985
    },
    {
      "epoch": 1.7254682978831413,
      "grad_norm": 0.2887297570705414,
      "learning_rate": 8.496878014112392e-06,
      "loss": 0.0359,
      "step": 33990
    },
    {
      "epoch": 1.7257221178740036,
      "grad_norm": 0.26516517996788025,
      "learning_rate": 8.495185880839975e-06,
      "loss": 0.0345,
      "step": 33995
    },
    {
      "epoch": 1.7259759378648662,
      "grad_norm": 0.359676331281662,
      "learning_rate": 8.493493747567559e-06,
      "loss": 0.0428,
      "step": 34000
    },
    {
      "epoch": 1.7262297578557289,
      "grad_norm": 0.5799069404602051,
      "learning_rate": 8.491801614295143e-06,
      "loss": 0.037,
      "step": 34005
    },
    {
      "epoch": 1.7264835778465912,
      "grad_norm": 0.23710906505584717,
      "learning_rate": 8.490109481022726e-06,
      "loss": 0.0419,
      "step": 34010
    },
    {
      "epoch": 1.7267373978374536,
      "grad_norm": 0.2692362368106842,
      "learning_rate": 8.48841734775031e-06,
      "loss": 0.0377,
      "step": 34015
    },
    {
      "epoch": 1.7269912178283162,
      "grad_norm": 0.6800029873847961,
      "learning_rate": 8.486725214477893e-06,
      "loss": 0.0404,
      "step": 34020
    },
    {
      "epoch": 1.7272450378191786,
      "grad_norm": 0.37395069003105164,
      "learning_rate": 8.485033081205477e-06,
      "loss": 0.0511,
      "step": 34025
    },
    {
      "epoch": 1.727498857810041,
      "grad_norm": 0.5215486884117126,
      "learning_rate": 8.48334094793306e-06,
      "loss": 0.0412,
      "step": 34030
    },
    {
      "epoch": 1.7277526778009036,
      "grad_norm": 0.30943822860717773,
      "learning_rate": 8.481648814660644e-06,
      "loss": 0.0424,
      "step": 34035
    },
    {
      "epoch": 1.7280064977917662,
      "grad_norm": 0.32664647698402405,
      "learning_rate": 8.479956681388227e-06,
      "loss": 0.043,
      "step": 34040
    },
    {
      "epoch": 1.7282603177826286,
      "grad_norm": 2.2540807723999023,
      "learning_rate": 8.47826454811581e-06,
      "loss": 0.0425,
      "step": 34045
    },
    {
      "epoch": 1.728514137773491,
      "grad_norm": 0.43366777896881104,
      "learning_rate": 8.476572414843394e-06,
      "loss": 0.0347,
      "step": 34050
    },
    {
      "epoch": 1.7287679577643535,
      "grad_norm": 0.5059844255447388,
      "learning_rate": 8.474880281570978e-06,
      "loss": 0.0385,
      "step": 34055
    },
    {
      "epoch": 1.7290217777552161,
      "grad_norm": 0.2656039297580719,
      "learning_rate": 8.473188148298562e-06,
      "loss": 0.0363,
      "step": 34060
    },
    {
      "epoch": 1.7292755977460785,
      "grad_norm": 0.3212661147117615,
      "learning_rate": 8.471496015026143e-06,
      "loss": 0.049,
      "step": 34065
    },
    {
      "epoch": 1.729529417736941,
      "grad_norm": 0.6338233947753906,
      "learning_rate": 8.469803881753727e-06,
      "loss": 0.0459,
      "step": 34070
    },
    {
      "epoch": 1.7297832377278035,
      "grad_norm": 0.3562254309654236,
      "learning_rate": 8.46811174848131e-06,
      "loss": 0.0384,
      "step": 34075
    },
    {
      "epoch": 1.730037057718666,
      "grad_norm": 0.25323736667633057,
      "learning_rate": 8.466419615208894e-06,
      "loss": 0.0377,
      "step": 34080
    },
    {
      "epoch": 1.7302908777095283,
      "grad_norm": 0.35645341873168945,
      "learning_rate": 8.464727481936478e-06,
      "loss": 0.0425,
      "step": 34085
    },
    {
      "epoch": 1.7305446977003909,
      "grad_norm": 0.3992117941379547,
      "learning_rate": 8.463035348664061e-06,
      "loss": 0.0406,
      "step": 34090
    },
    {
      "epoch": 1.7307985176912535,
      "grad_norm": 0.40293341875076294,
      "learning_rate": 8.461343215391645e-06,
      "loss": 0.0504,
      "step": 34095
    },
    {
      "epoch": 1.7310523376821159,
      "grad_norm": 0.27027973532676697,
      "learning_rate": 8.459651082119228e-06,
      "loss": 0.0476,
      "step": 34100
    },
    {
      "epoch": 1.7313061576729782,
      "grad_norm": 0.44675561785697937,
      "learning_rate": 8.457958948846812e-06,
      "loss": 0.0453,
      "step": 34105
    },
    {
      "epoch": 1.7315599776638408,
      "grad_norm": 0.2727741301059723,
      "learning_rate": 8.456266815574395e-06,
      "loss": 0.036,
      "step": 34110
    },
    {
      "epoch": 1.7318137976547034,
      "grad_norm": 0.3078666627407074,
      "learning_rate": 8.454574682301979e-06,
      "loss": 0.035,
      "step": 34115
    },
    {
      "epoch": 1.7320676176455656,
      "grad_norm": 0.2535753846168518,
      "learning_rate": 8.452882549029562e-06,
      "loss": 0.039,
      "step": 34120
    },
    {
      "epoch": 1.7323214376364282,
      "grad_norm": 0.3663230538368225,
      "learning_rate": 8.451190415757146e-06,
      "loss": 0.0383,
      "step": 34125
    },
    {
      "epoch": 1.7325752576272908,
      "grad_norm": 0.34004485607147217,
      "learning_rate": 8.44949828248473e-06,
      "loss": 0.033,
      "step": 34130
    },
    {
      "epoch": 1.7328290776181532,
      "grad_norm": 0.25209248065948486,
      "learning_rate": 8.447806149212313e-06,
      "loss": 0.0359,
      "step": 34135
    },
    {
      "epoch": 1.7330828976090156,
      "grad_norm": 0.351085364818573,
      "learning_rate": 8.446114015939897e-06,
      "loss": 0.041,
      "step": 34140
    },
    {
      "epoch": 1.7333367175998782,
      "grad_norm": 0.28116995096206665,
      "learning_rate": 8.44442188266748e-06,
      "loss": 0.0351,
      "step": 34145
    },
    {
      "epoch": 1.7335905375907408,
      "grad_norm": 0.4009757936000824,
      "learning_rate": 8.442729749395064e-06,
      "loss": 0.0397,
      "step": 34150
    },
    {
      "epoch": 1.7338443575816032,
      "grad_norm": 0.5399395227432251,
      "learning_rate": 8.441037616122647e-06,
      "loss": 0.0443,
      "step": 34155
    },
    {
      "epoch": 1.7340981775724655,
      "grad_norm": 0.28884685039520264,
      "learning_rate": 8.439345482850229e-06,
      "loss": 0.0359,
      "step": 34160
    },
    {
      "epoch": 1.7343519975633281,
      "grad_norm": 0.3574964702129364,
      "learning_rate": 8.437653349577813e-06,
      "loss": 0.0439,
      "step": 34165
    },
    {
      "epoch": 1.7346058175541905,
      "grad_norm": 0.28065600991249084,
      "learning_rate": 8.435961216305398e-06,
      "loss": 0.0398,
      "step": 34170
    },
    {
      "epoch": 1.734859637545053,
      "grad_norm": 0.3115881085395813,
      "learning_rate": 8.43426908303298e-06,
      "loss": 0.0366,
      "step": 34175
    },
    {
      "epoch": 1.7351134575359155,
      "grad_norm": 0.3346794545650482,
      "learning_rate": 8.432576949760563e-06,
      "loss": 0.0398,
      "step": 34180
    },
    {
      "epoch": 1.735367277526778,
      "grad_norm": 0.3550897240638733,
      "learning_rate": 8.430884816488147e-06,
      "loss": 0.0385,
      "step": 34185
    },
    {
      "epoch": 1.7356210975176405,
      "grad_norm": 0.2994544208049774,
      "learning_rate": 8.42919268321573e-06,
      "loss": 0.033,
      "step": 34190
    },
    {
      "epoch": 1.7358749175085029,
      "grad_norm": 0.2627989947795868,
      "learning_rate": 8.427500549943314e-06,
      "loss": 0.0326,
      "step": 34195
    },
    {
      "epoch": 1.7361287374993655,
      "grad_norm": 0.3173164427280426,
      "learning_rate": 8.425808416670897e-06,
      "loss": 0.0393,
      "step": 34200
    },
    {
      "epoch": 1.736382557490228,
      "grad_norm": 0.40492770075798035,
      "learning_rate": 8.424116283398481e-06,
      "loss": 0.0412,
      "step": 34205
    },
    {
      "epoch": 1.7366363774810905,
      "grad_norm": 0.5181750059127808,
      "learning_rate": 8.422424150126065e-06,
      "loss": 0.0397,
      "step": 34210
    },
    {
      "epoch": 1.7368901974719528,
      "grad_norm": 0.3461681008338928,
      "learning_rate": 8.420732016853648e-06,
      "loss": 0.0413,
      "step": 34215
    },
    {
      "epoch": 1.7371440174628154,
      "grad_norm": 0.4116683304309845,
      "learning_rate": 8.419039883581232e-06,
      "loss": 0.0383,
      "step": 34220
    },
    {
      "epoch": 1.7373978374536778,
      "grad_norm": 0.3937915861606598,
      "learning_rate": 8.417347750308815e-06,
      "loss": 0.0503,
      "step": 34225
    },
    {
      "epoch": 1.7376516574445402,
      "grad_norm": 0.2615407109260559,
      "learning_rate": 8.415655617036399e-06,
      "loss": 0.0401,
      "step": 34230
    },
    {
      "epoch": 1.7379054774354028,
      "grad_norm": 0.3176824152469635,
      "learning_rate": 8.413963483763982e-06,
      "loss": 0.0414,
      "step": 34235
    },
    {
      "epoch": 1.7381592974262654,
      "grad_norm": 0.3036479353904724,
      "learning_rate": 8.412271350491566e-06,
      "loss": 0.0442,
      "step": 34240
    },
    {
      "epoch": 1.7384131174171278,
      "grad_norm": 0.3770005404949188,
      "learning_rate": 8.41057921721915e-06,
      "loss": 0.0486,
      "step": 34245
    },
    {
      "epoch": 1.7386669374079902,
      "grad_norm": 0.30073773860931396,
      "learning_rate": 8.408887083946731e-06,
      "loss": 0.0381,
      "step": 34250
    },
    {
      "epoch": 1.7389207573988528,
      "grad_norm": 0.38862597942352295,
      "learning_rate": 8.407194950674315e-06,
      "loss": 0.0328,
      "step": 34255
    },
    {
      "epoch": 1.7391745773897154,
      "grad_norm": 0.3898008167743683,
      "learning_rate": 8.4055028174019e-06,
      "loss": 0.0441,
      "step": 34260
    },
    {
      "epoch": 1.7394283973805775,
      "grad_norm": 0.3264569342136383,
      "learning_rate": 8.403810684129484e-06,
      "loss": 0.041,
      "step": 34265
    },
    {
      "epoch": 1.7396822173714401,
      "grad_norm": 0.3390885293483734,
      "learning_rate": 8.402118550857065e-06,
      "loss": 0.0515,
      "step": 34270
    },
    {
      "epoch": 1.7399360373623027,
      "grad_norm": 0.371646910905838,
      "learning_rate": 8.400426417584649e-06,
      "loss": 0.0428,
      "step": 34275
    },
    {
      "epoch": 1.7401898573531651,
      "grad_norm": 0.46684572100639343,
      "learning_rate": 8.398734284312232e-06,
      "loss": 0.042,
      "step": 34280
    },
    {
      "epoch": 1.7404436773440275,
      "grad_norm": 0.2889266610145569,
      "learning_rate": 8.397042151039818e-06,
      "loss": 0.0357,
      "step": 34285
    },
    {
      "epoch": 1.74069749733489,
      "grad_norm": 0.27424633502960205,
      "learning_rate": 8.3953500177674e-06,
      "loss": 0.0359,
      "step": 34290
    },
    {
      "epoch": 1.7409513173257527,
      "grad_norm": 0.2002898007631302,
      "learning_rate": 8.393657884494983e-06,
      "loss": 0.036,
      "step": 34295
    },
    {
      "epoch": 1.741205137316615,
      "grad_norm": 0.3924329876899719,
      "learning_rate": 8.391965751222567e-06,
      "loss": 0.0358,
      "step": 34300
    },
    {
      "epoch": 1.7414589573074775,
      "grad_norm": 0.25661349296569824,
      "learning_rate": 8.39027361795015e-06,
      "loss": 0.0328,
      "step": 34305
    },
    {
      "epoch": 1.74171277729834,
      "grad_norm": 0.19355782866477966,
      "learning_rate": 8.388581484677734e-06,
      "loss": 0.0312,
      "step": 34310
    },
    {
      "epoch": 1.7419665972892024,
      "grad_norm": 0.32729625701904297,
      "learning_rate": 8.386889351405317e-06,
      "loss": 0.0468,
      "step": 34315
    },
    {
      "epoch": 1.7422204172800648,
      "grad_norm": 0.36025574803352356,
      "learning_rate": 8.3851972181329e-06,
      "loss": 0.0428,
      "step": 34320
    },
    {
      "epoch": 1.7424742372709274,
      "grad_norm": 0.39251708984375,
      "learning_rate": 8.383505084860484e-06,
      "loss": 0.0366,
      "step": 34325
    },
    {
      "epoch": 1.74272805726179,
      "grad_norm": 0.5001270771026611,
      "learning_rate": 8.381812951588068e-06,
      "loss": 0.0404,
      "step": 34330
    },
    {
      "epoch": 1.7429818772526524,
      "grad_norm": 0.36959490180015564,
      "learning_rate": 8.380120818315651e-06,
      "loss": 0.037,
      "step": 34335
    },
    {
      "epoch": 1.7432356972435148,
      "grad_norm": 0.3076193928718567,
      "learning_rate": 8.378428685043235e-06,
      "loss": 0.0389,
      "step": 34340
    },
    {
      "epoch": 1.7434895172343774,
      "grad_norm": 0.37228232622146606,
      "learning_rate": 8.376736551770819e-06,
      "loss": 0.0495,
      "step": 34345
    },
    {
      "epoch": 1.74374333722524,
      "grad_norm": 0.3191566467285156,
      "learning_rate": 8.375044418498402e-06,
      "loss": 0.0344,
      "step": 34350
    },
    {
      "epoch": 1.7439971572161024,
      "grad_norm": 0.3981260061264038,
      "learning_rate": 8.373352285225986e-06,
      "loss": 0.0434,
      "step": 34355
    },
    {
      "epoch": 1.7442509772069648,
      "grad_norm": 0.310577392578125,
      "learning_rate": 8.37166015195357e-06,
      "loss": 0.0335,
      "step": 34360
    },
    {
      "epoch": 1.7445047971978274,
      "grad_norm": 0.4347289204597473,
      "learning_rate": 8.369968018681151e-06,
      "loss": 0.0336,
      "step": 34365
    },
    {
      "epoch": 1.7447586171886897,
      "grad_norm": 0.3463854491710663,
      "learning_rate": 8.368275885408735e-06,
      "loss": 0.0437,
      "step": 34370
    },
    {
      "epoch": 1.7450124371795521,
      "grad_norm": 0.32871583104133606,
      "learning_rate": 8.36658375213632e-06,
      "loss": 0.04,
      "step": 34375
    },
    {
      "epoch": 1.7452662571704147,
      "grad_norm": 0.5799885988235474,
      "learning_rate": 8.364891618863903e-06,
      "loss": 0.0471,
      "step": 34380
    },
    {
      "epoch": 1.7455200771612773,
      "grad_norm": 0.49882349371910095,
      "learning_rate": 8.363199485591485e-06,
      "loss": 0.0408,
      "step": 34385
    },
    {
      "epoch": 1.7457738971521397,
      "grad_norm": 0.29698601365089417,
      "learning_rate": 8.361507352319069e-06,
      "loss": 0.0415,
      "step": 34390
    },
    {
      "epoch": 1.746027717143002,
      "grad_norm": 0.3538743555545807,
      "learning_rate": 8.359815219046652e-06,
      "loss": 0.0442,
      "step": 34395
    },
    {
      "epoch": 1.7462815371338647,
      "grad_norm": 0.20709893107414246,
      "learning_rate": 8.358123085774238e-06,
      "loss": 0.0355,
      "step": 34400
    },
    {
      "epoch": 1.7465353571247273,
      "grad_norm": 0.27079060673713684,
      "learning_rate": 8.35643095250182e-06,
      "loss": 0.0468,
      "step": 34405
    },
    {
      "epoch": 1.7467891771155897,
      "grad_norm": 0.37064772844314575,
      "learning_rate": 8.354738819229403e-06,
      "loss": 0.0377,
      "step": 34410
    },
    {
      "epoch": 1.747042997106452,
      "grad_norm": 0.38094353675842285,
      "learning_rate": 8.353046685956986e-06,
      "loss": 0.0427,
      "step": 34415
    },
    {
      "epoch": 1.7472968170973147,
      "grad_norm": 0.3148980140686035,
      "learning_rate": 8.35135455268457e-06,
      "loss": 0.0339,
      "step": 34420
    },
    {
      "epoch": 1.747550637088177,
      "grad_norm": 0.49832138419151306,
      "learning_rate": 8.349662419412154e-06,
      "loss": 0.0409,
      "step": 34425
    },
    {
      "epoch": 1.7478044570790394,
      "grad_norm": 0.3183405101299286,
      "learning_rate": 8.347970286139737e-06,
      "loss": 0.0418,
      "step": 34430
    },
    {
      "epoch": 1.748058277069902,
      "grad_norm": 0.3183467984199524,
      "learning_rate": 8.34627815286732e-06,
      "loss": 0.0352,
      "step": 34435
    },
    {
      "epoch": 1.7483120970607646,
      "grad_norm": 0.38760673999786377,
      "learning_rate": 8.344586019594904e-06,
      "loss": 0.0423,
      "step": 34440
    },
    {
      "epoch": 1.748565917051627,
      "grad_norm": 0.25345578789711,
      "learning_rate": 8.342893886322488e-06,
      "loss": 0.0322,
      "step": 34445
    },
    {
      "epoch": 1.7488197370424894,
      "grad_norm": 0.3874957859516144,
      "learning_rate": 8.341201753050071e-06,
      "loss": 0.0399,
      "step": 34450
    },
    {
      "epoch": 1.749073557033352,
      "grad_norm": 0.31035029888153076,
      "learning_rate": 8.339509619777655e-06,
      "loss": 0.0429,
      "step": 34455
    },
    {
      "epoch": 1.7493273770242146,
      "grad_norm": 0.5256080627441406,
      "learning_rate": 8.337817486505237e-06,
      "loss": 0.0369,
      "step": 34460
    },
    {
      "epoch": 1.7495811970150768,
      "grad_norm": 0.6625921130180359,
      "learning_rate": 8.336125353232822e-06,
      "loss": 0.0367,
      "step": 34465
    },
    {
      "epoch": 1.7498350170059394,
      "grad_norm": 0.2330664098262787,
      "learning_rate": 8.334433219960405e-06,
      "loss": 0.0389,
      "step": 34470
    },
    {
      "epoch": 1.750088836996802,
      "grad_norm": 0.6050800085067749,
      "learning_rate": 8.332741086687989e-06,
      "loss": 0.0381,
      "step": 34475
    },
    {
      "epoch": 1.7503426569876643,
      "grad_norm": 0.3633989691734314,
      "learning_rate": 8.33104895341557e-06,
      "loss": 0.0333,
      "step": 34480
    },
    {
      "epoch": 1.7505964769785267,
      "grad_norm": 0.25826677680015564,
      "learning_rate": 8.329356820143154e-06,
      "loss": 0.0405,
      "step": 34485
    },
    {
      "epoch": 1.7508502969693893,
      "grad_norm": 0.2678346037864685,
      "learning_rate": 8.32766468687074e-06,
      "loss": 0.041,
      "step": 34490
    },
    {
      "epoch": 1.751104116960252,
      "grad_norm": 0.8798556923866272,
      "learning_rate": 8.325972553598321e-06,
      "loss": 0.0472,
      "step": 34495
    },
    {
      "epoch": 1.7513579369511143,
      "grad_norm": 0.4030362367630005,
      "learning_rate": 8.324280420325905e-06,
      "loss": 0.0423,
      "step": 34500
    },
    {
      "epoch": 1.7516117569419767,
      "grad_norm": 0.5761476159095764,
      "learning_rate": 8.322588287053489e-06,
      "loss": 0.0368,
      "step": 34505
    },
    {
      "epoch": 1.7518655769328393,
      "grad_norm": 0.2705753743648529,
      "learning_rate": 8.320896153781072e-06,
      "loss": 0.038,
      "step": 34510
    },
    {
      "epoch": 1.7521193969237017,
      "grad_norm": 0.3345622420310974,
      "learning_rate": 8.319204020508656e-06,
      "loss": 0.0366,
      "step": 34515
    },
    {
      "epoch": 1.752373216914564,
      "grad_norm": 0.2705942392349243,
      "learning_rate": 8.31751188723624e-06,
      "loss": 0.0358,
      "step": 34520
    },
    {
      "epoch": 1.7526270369054266,
      "grad_norm": 0.28924059867858887,
      "learning_rate": 8.315819753963823e-06,
      "loss": 0.0347,
      "step": 34525
    },
    {
      "epoch": 1.7528808568962893,
      "grad_norm": 0.2418852597475052,
      "learning_rate": 8.314127620691406e-06,
      "loss": 0.0324,
      "step": 34530
    },
    {
      "epoch": 1.7531346768871516,
      "grad_norm": 0.399440735578537,
      "learning_rate": 8.31243548741899e-06,
      "loss": 0.0415,
      "step": 34535
    },
    {
      "epoch": 1.753388496878014,
      "grad_norm": 0.37920889258384705,
      "learning_rate": 8.310743354146573e-06,
      "loss": 0.0381,
      "step": 34540
    },
    {
      "epoch": 1.7536423168688766,
      "grad_norm": 0.37504783272743225,
      "learning_rate": 8.309051220874157e-06,
      "loss": 0.0441,
      "step": 34545
    },
    {
      "epoch": 1.7538961368597392,
      "grad_norm": 0.3581880033016205,
      "learning_rate": 8.30735908760174e-06,
      "loss": 0.0376,
      "step": 34550
    },
    {
      "epoch": 1.7541499568506016,
      "grad_norm": 0.4461826980113983,
      "learning_rate": 8.305666954329324e-06,
      "loss": 0.0412,
      "step": 34555
    },
    {
      "epoch": 1.754403776841464,
      "grad_norm": 0.29656851291656494,
      "learning_rate": 8.303974821056908e-06,
      "loss": 0.0315,
      "step": 34560
    },
    {
      "epoch": 1.7546575968323266,
      "grad_norm": 0.3078388273715973,
      "learning_rate": 8.302282687784491e-06,
      "loss": 0.0354,
      "step": 34565
    },
    {
      "epoch": 1.754911416823189,
      "grad_norm": 0.4501648247241974,
      "learning_rate": 8.300590554512073e-06,
      "loss": 0.0379,
      "step": 34570
    },
    {
      "epoch": 1.7551652368140513,
      "grad_norm": 0.6488507986068726,
      "learning_rate": 8.298898421239656e-06,
      "loss": 0.0429,
      "step": 34575
    },
    {
      "epoch": 1.755419056804914,
      "grad_norm": 0.2891724407672882,
      "learning_rate": 8.297206287967242e-06,
      "loss": 0.0447,
      "step": 34580
    },
    {
      "epoch": 1.7556728767957765,
      "grad_norm": 0.4944934546947479,
      "learning_rate": 8.295514154694825e-06,
      "loss": 0.032,
      "step": 34585
    },
    {
      "epoch": 1.755926696786639,
      "grad_norm": 0.3709994852542877,
      "learning_rate": 8.293822021422407e-06,
      "loss": 0.0426,
      "step": 34590
    },
    {
      "epoch": 1.7561805167775013,
      "grad_norm": 0.552474856376648,
      "learning_rate": 8.29212988814999e-06,
      "loss": 0.0358,
      "step": 34595
    },
    {
      "epoch": 1.756434336768364,
      "grad_norm": 0.3508904278278351,
      "learning_rate": 8.290437754877574e-06,
      "loss": 0.0324,
      "step": 34600
    },
    {
      "epoch": 1.7566881567592265,
      "grad_norm": 0.3192765712738037,
      "learning_rate": 8.28874562160516e-06,
      "loss": 0.038,
      "step": 34605
    },
    {
      "epoch": 1.7569419767500887,
      "grad_norm": 0.3167833983898163,
      "learning_rate": 8.287053488332741e-06,
      "loss": 0.0431,
      "step": 34610
    },
    {
      "epoch": 1.7571957967409513,
      "grad_norm": 0.33134904503822327,
      "learning_rate": 8.285361355060325e-06,
      "loss": 0.0391,
      "step": 34615
    },
    {
      "epoch": 1.7574496167318139,
      "grad_norm": 0.5783857107162476,
      "learning_rate": 8.283669221787908e-06,
      "loss": 0.0446,
      "step": 34620
    },
    {
      "epoch": 1.7577034367226763,
      "grad_norm": 0.24570982158184052,
      "learning_rate": 8.281977088515492e-06,
      "loss": 0.0376,
      "step": 34625
    },
    {
      "epoch": 1.7579572567135386,
      "grad_norm": 0.29740071296691895,
      "learning_rate": 8.280284955243075e-06,
      "loss": 0.0436,
      "step": 34630
    },
    {
      "epoch": 1.7582110767044012,
      "grad_norm": 0.4562722444534302,
      "learning_rate": 8.278592821970659e-06,
      "loss": 0.0445,
      "step": 34635
    },
    {
      "epoch": 1.7584648966952638,
      "grad_norm": 0.5596370100975037,
      "learning_rate": 8.276900688698243e-06,
      "loss": 0.0402,
      "step": 34640
    },
    {
      "epoch": 1.7587187166861262,
      "grad_norm": 0.38506823778152466,
      "learning_rate": 8.275208555425826e-06,
      "loss": 0.0374,
      "step": 34645
    },
    {
      "epoch": 1.7589725366769886,
      "grad_norm": 0.2869029641151428,
      "learning_rate": 8.27351642215341e-06,
      "loss": 0.0391,
      "step": 34650
    },
    {
      "epoch": 1.7592263566678512,
      "grad_norm": 0.5008401274681091,
      "learning_rate": 8.271824288880993e-06,
      "loss": 0.0458,
      "step": 34655
    },
    {
      "epoch": 1.7594801766587136,
      "grad_norm": 0.3160930871963501,
      "learning_rate": 8.270132155608577e-06,
      "loss": 0.0399,
      "step": 34660
    },
    {
      "epoch": 1.759733996649576,
      "grad_norm": 0.32366958260536194,
      "learning_rate": 8.268440022336159e-06,
      "loss": 0.0427,
      "step": 34665
    },
    {
      "epoch": 1.7599878166404386,
      "grad_norm": 0.25260210037231445,
      "learning_rate": 8.266747889063744e-06,
      "loss": 0.0413,
      "step": 34670
    },
    {
      "epoch": 1.7602416366313012,
      "grad_norm": 0.33976075053215027,
      "learning_rate": 8.265055755791327e-06,
      "loss": 0.0402,
      "step": 34675
    },
    {
      "epoch": 1.7604954566221636,
      "grad_norm": 0.47365057468414307,
      "learning_rate": 8.263363622518911e-06,
      "loss": 0.0412,
      "step": 34680
    },
    {
      "epoch": 1.760749276613026,
      "grad_norm": 0.28111734986305237,
      "learning_rate": 8.261671489246493e-06,
      "loss": 0.0363,
      "step": 34685
    },
    {
      "epoch": 1.7610030966038885,
      "grad_norm": 0.220921590924263,
      "learning_rate": 8.259979355974076e-06,
      "loss": 0.0368,
      "step": 34690
    },
    {
      "epoch": 1.7612569165947511,
      "grad_norm": 0.3260694146156311,
      "learning_rate": 8.258287222701662e-06,
      "loss": 0.0401,
      "step": 34695
    },
    {
      "epoch": 1.7615107365856135,
      "grad_norm": 0.32202768325805664,
      "learning_rate": 8.256595089429245e-06,
      "loss": 0.0422,
      "step": 34700
    },
    {
      "epoch": 1.761764556576476,
      "grad_norm": 0.3090651333332062,
      "learning_rate": 8.254902956156827e-06,
      "loss": 0.0417,
      "step": 34705
    },
    {
      "epoch": 1.7620183765673385,
      "grad_norm": 0.3767678737640381,
      "learning_rate": 8.25321082288441e-06,
      "loss": 0.043,
      "step": 34710
    },
    {
      "epoch": 1.7622721965582009,
      "grad_norm": 0.4838859736919403,
      "learning_rate": 8.251518689611994e-06,
      "loss": 0.036,
      "step": 34715
    },
    {
      "epoch": 1.7625260165490633,
      "grad_norm": 0.4400741755962372,
      "learning_rate": 8.24982655633958e-06,
      "loss": 0.0347,
      "step": 34720
    },
    {
      "epoch": 1.7627798365399259,
      "grad_norm": 0.31652846932411194,
      "learning_rate": 8.248134423067161e-06,
      "loss": 0.0543,
      "step": 34725
    },
    {
      "epoch": 1.7630336565307885,
      "grad_norm": 0.46276259422302246,
      "learning_rate": 8.246442289794745e-06,
      "loss": 0.0405,
      "step": 34730
    },
    {
      "epoch": 1.7632874765216509,
      "grad_norm": 0.3353394567966461,
      "learning_rate": 8.244750156522328e-06,
      "loss": 0.0348,
      "step": 34735
    },
    {
      "epoch": 1.7635412965125132,
      "grad_norm": 0.7226163148880005,
      "learning_rate": 8.243058023249912e-06,
      "loss": 0.0379,
      "step": 34740
    },
    {
      "epoch": 1.7637951165033758,
      "grad_norm": 0.3163387179374695,
      "learning_rate": 8.241365889977495e-06,
      "loss": 0.0415,
      "step": 34745
    },
    {
      "epoch": 1.7640489364942384,
      "grad_norm": 0.36522141098976135,
      "learning_rate": 8.239673756705079e-06,
      "loss": 0.0397,
      "step": 34750
    },
    {
      "epoch": 1.7643027564851008,
      "grad_norm": 0.4560871124267578,
      "learning_rate": 8.237981623432662e-06,
      "loss": 0.0454,
      "step": 34755
    },
    {
      "epoch": 1.7645565764759632,
      "grad_norm": 0.6275542974472046,
      "learning_rate": 8.236289490160246e-06,
      "loss": 0.0401,
      "step": 34760
    },
    {
      "epoch": 1.7648103964668258,
      "grad_norm": 0.3421061336994171,
      "learning_rate": 8.23459735688783e-06,
      "loss": 0.0354,
      "step": 34765
    },
    {
      "epoch": 1.7650642164576882,
      "grad_norm": 0.2922126054763794,
      "learning_rate": 8.232905223615413e-06,
      "loss": 0.0379,
      "step": 34770
    },
    {
      "epoch": 1.7653180364485506,
      "grad_norm": 0.37141862511634827,
      "learning_rate": 8.231213090342997e-06,
      "loss": 0.0377,
      "step": 34775
    },
    {
      "epoch": 1.7655718564394132,
      "grad_norm": 0.2555692493915558,
      "learning_rate": 8.229520957070578e-06,
      "loss": 0.0321,
      "step": 34780
    },
    {
      "epoch": 1.7658256764302758,
      "grad_norm": 0.47595980763435364,
      "learning_rate": 8.227828823798164e-06,
      "loss": 0.046,
      "step": 34785
    },
    {
      "epoch": 1.7660794964211382,
      "grad_norm": 0.31954678893089294,
      "learning_rate": 8.226136690525747e-06,
      "loss": 0.0324,
      "step": 34790
    },
    {
      "epoch": 1.7663333164120005,
      "grad_norm": 0.30136197805404663,
      "learning_rate": 8.22444455725333e-06,
      "loss": 0.0454,
      "step": 34795
    },
    {
      "epoch": 1.7665871364028631,
      "grad_norm": 0.41031014919281006,
      "learning_rate": 8.222752423980913e-06,
      "loss": 0.0378,
      "step": 34800
    },
    {
      "epoch": 1.7668409563937255,
      "grad_norm": 0.2514888048171997,
      "learning_rate": 8.221060290708496e-06,
      "loss": 0.0348,
      "step": 34805
    },
    {
      "epoch": 1.767094776384588,
      "grad_norm": 0.34350481629371643,
      "learning_rate": 8.219368157436081e-06,
      "loss": 0.0353,
      "step": 34810
    },
    {
      "epoch": 1.7673485963754505,
      "grad_norm": 0.27384987473487854,
      "learning_rate": 8.217676024163663e-06,
      "loss": 0.0332,
      "step": 34815
    },
    {
      "epoch": 1.767602416366313,
      "grad_norm": 0.40239375829696655,
      "learning_rate": 8.215983890891247e-06,
      "loss": 0.0414,
      "step": 34820
    },
    {
      "epoch": 1.7678562363571755,
      "grad_norm": 0.31507280468940735,
      "learning_rate": 8.21429175761883e-06,
      "loss": 0.041,
      "step": 34825
    },
    {
      "epoch": 1.7681100563480379,
      "grad_norm": 0.5452533960342407,
      "learning_rate": 8.212599624346414e-06,
      "loss": 0.0472,
      "step": 34830
    },
    {
      "epoch": 1.7683638763389005,
      "grad_norm": 0.2969473600387573,
      "learning_rate": 8.210907491073997e-06,
      "loss": 0.0356,
      "step": 34835
    },
    {
      "epoch": 1.768617696329763,
      "grad_norm": 0.4441262483596802,
      "learning_rate": 8.209215357801581e-06,
      "loss": 0.0414,
      "step": 34840
    },
    {
      "epoch": 1.7688715163206254,
      "grad_norm": 0.3782421946525574,
      "learning_rate": 8.207523224529165e-06,
      "loss": 0.0413,
      "step": 34845
    },
    {
      "epoch": 1.7691253363114878,
      "grad_norm": 0.3684667646884918,
      "learning_rate": 8.205831091256748e-06,
      "loss": 0.0388,
      "step": 34850
    },
    {
      "epoch": 1.7693791563023504,
      "grad_norm": 0.3120320439338684,
      "learning_rate": 8.204138957984332e-06,
      "loss": 0.0404,
      "step": 34855
    },
    {
      "epoch": 1.7696329762932128,
      "grad_norm": 0.46274301409721375,
      "learning_rate": 8.202446824711915e-06,
      "loss": 0.0383,
      "step": 34860
    },
    {
      "epoch": 1.7698867962840752,
      "grad_norm": 0.3160901367664337,
      "learning_rate": 8.200754691439499e-06,
      "loss": 0.0349,
      "step": 34865
    },
    {
      "epoch": 1.7701406162749378,
      "grad_norm": 0.3432873487472534,
      "learning_rate": 8.199062558167082e-06,
      "loss": 0.0451,
      "step": 34870
    },
    {
      "epoch": 1.7703944362658004,
      "grad_norm": 0.3663712739944458,
      "learning_rate": 8.197370424894666e-06,
      "loss": 0.037,
      "step": 34875
    },
    {
      "epoch": 1.7706482562566628,
      "grad_norm": 0.29719987511634827,
      "learning_rate": 8.19567829162225e-06,
      "loss": 0.038,
      "step": 34880
    },
    {
      "epoch": 1.7709020762475252,
      "grad_norm": 0.3122144639492035,
      "learning_rate": 8.193986158349833e-06,
      "loss": 0.0416,
      "step": 34885
    },
    {
      "epoch": 1.7711558962383878,
      "grad_norm": 0.3486882150173187,
      "learning_rate": 8.192294025077415e-06,
      "loss": 0.0453,
      "step": 34890
    },
    {
      "epoch": 1.7714097162292504,
      "grad_norm": 0.2783128023147583,
      "learning_rate": 8.190601891804998e-06,
      "loss": 0.0503,
      "step": 34895
    },
    {
      "epoch": 1.7716635362201127,
      "grad_norm": 0.36007043719291687,
      "learning_rate": 8.188909758532583e-06,
      "loss": 0.0362,
      "step": 34900
    },
    {
      "epoch": 1.7719173562109751,
      "grad_norm": 0.7849182486534119,
      "learning_rate": 8.187217625260167e-06,
      "loss": 0.0401,
      "step": 34905
    },
    {
      "epoch": 1.7721711762018377,
      "grad_norm": 0.35825115442276,
      "learning_rate": 8.185525491987749e-06,
      "loss": 0.0429,
      "step": 34910
    },
    {
      "epoch": 1.7724249961927,
      "grad_norm": 0.5037504434585571,
      "learning_rate": 8.183833358715332e-06,
      "loss": 0.0364,
      "step": 34915
    },
    {
      "epoch": 1.7726788161835625,
      "grad_norm": 0.47052305936813354,
      "learning_rate": 8.182141225442916e-06,
      "loss": 0.0427,
      "step": 34920
    },
    {
      "epoch": 1.772932636174425,
      "grad_norm": 0.45571646094322205,
      "learning_rate": 8.180449092170501e-06,
      "loss": 0.0461,
      "step": 34925
    },
    {
      "epoch": 1.7731864561652877,
      "grad_norm": 0.32976770401000977,
      "learning_rate": 8.178756958898083e-06,
      "loss": 0.0397,
      "step": 34930
    },
    {
      "epoch": 1.77344027615615,
      "grad_norm": 0.369817852973938,
      "learning_rate": 8.177064825625667e-06,
      "loss": 0.0477,
      "step": 34935
    },
    {
      "epoch": 1.7736940961470125,
      "grad_norm": 0.31425940990448,
      "learning_rate": 8.17537269235325e-06,
      "loss": 0.0397,
      "step": 34940
    },
    {
      "epoch": 1.773947916137875,
      "grad_norm": 0.6579198837280273,
      "learning_rate": 8.173680559080834e-06,
      "loss": 0.0482,
      "step": 34945
    },
    {
      "epoch": 1.7742017361287377,
      "grad_norm": 0.3199804127216339,
      "learning_rate": 8.171988425808417e-06,
      "loss": 0.0355,
      "step": 34950
    },
    {
      "epoch": 1.7744555561195998,
      "grad_norm": 0.3551079332828522,
      "learning_rate": 8.170296292536e-06,
      "loss": 0.0411,
      "step": 34955
    },
    {
      "epoch": 1.7747093761104624,
      "grad_norm": 0.4659683108329773,
      "learning_rate": 8.168604159263584e-06,
      "loss": 0.0335,
      "step": 34960
    },
    {
      "epoch": 1.774963196101325,
      "grad_norm": 0.28088241815567017,
      "learning_rate": 8.166912025991168e-06,
      "loss": 0.0415,
      "step": 34965
    },
    {
      "epoch": 1.7752170160921874,
      "grad_norm": 0.37877973914146423,
      "learning_rate": 8.165219892718751e-06,
      "loss": 0.0426,
      "step": 34970
    },
    {
      "epoch": 1.7754708360830498,
      "grad_norm": 0.8946444392204285,
      "learning_rate": 8.163527759446335e-06,
      "loss": 0.0396,
      "step": 34975
    },
    {
      "epoch": 1.7757246560739124,
      "grad_norm": 0.29175713658332825,
      "learning_rate": 8.161835626173919e-06,
      "loss": 0.0315,
      "step": 34980
    },
    {
      "epoch": 1.775978476064775,
      "grad_norm": 0.7535488605499268,
      "learning_rate": 8.1601434929015e-06,
      "loss": 0.0439,
      "step": 34985
    },
    {
      "epoch": 1.7762322960556374,
      "grad_norm": 0.39257892966270447,
      "learning_rate": 8.158451359629086e-06,
      "loss": 0.0476,
      "step": 34990
    },
    {
      "epoch": 1.7764861160464998,
      "grad_norm": 0.4855389893054962,
      "learning_rate": 8.156759226356669e-06,
      "loss": 0.0364,
      "step": 34995
    },
    {
      "epoch": 1.7767399360373624,
      "grad_norm": 0.3152945935726166,
      "learning_rate": 8.155067093084253e-06,
      "loss": 0.0389,
      "step": 35000
    },
    {
      "epoch": 1.7769937560282247,
      "grad_norm": 0.34567198157310486,
      "learning_rate": 8.153374959811835e-06,
      "loss": 0.0413,
      "step": 35005
    },
    {
      "epoch": 1.7772475760190871,
      "grad_norm": 0.36939743161201477,
      "learning_rate": 8.151682826539418e-06,
      "loss": 0.0448,
      "step": 35010
    },
    {
      "epoch": 1.7775013960099497,
      "grad_norm": 0.3870468735694885,
      "learning_rate": 8.149990693267003e-06,
      "loss": 0.037,
      "step": 35015
    },
    {
      "epoch": 1.7777552160008123,
      "grad_norm": 0.3341517150402069,
      "learning_rate": 8.148298559994587e-06,
      "loss": 0.0411,
      "step": 35020
    },
    {
      "epoch": 1.7780090359916747,
      "grad_norm": 0.5142117738723755,
      "learning_rate": 8.146606426722169e-06,
      "loss": 0.0442,
      "step": 35025
    },
    {
      "epoch": 1.778262855982537,
      "grad_norm": 0.28023678064346313,
      "learning_rate": 8.144914293449752e-06,
      "loss": 0.0456,
      "step": 35030
    },
    {
      "epoch": 1.7785166759733997,
      "grad_norm": 0.6822634935379028,
      "learning_rate": 8.143222160177336e-06,
      "loss": 0.041,
      "step": 35035
    },
    {
      "epoch": 1.7787704959642623,
      "grad_norm": 0.33477500081062317,
      "learning_rate": 8.141530026904921e-06,
      "loss": 0.0323,
      "step": 35040
    },
    {
      "epoch": 1.7790243159551247,
      "grad_norm": 0.17918257415294647,
      "learning_rate": 8.139837893632503e-06,
      "loss": 0.0325,
      "step": 35045
    },
    {
      "epoch": 1.779278135945987,
      "grad_norm": 0.5592842102050781,
      "learning_rate": 8.138145760360086e-06,
      "loss": 0.0378,
      "step": 35050
    },
    {
      "epoch": 1.7795319559368497,
      "grad_norm": 0.2886914610862732,
      "learning_rate": 8.13645362708767e-06,
      "loss": 0.0454,
      "step": 35055
    },
    {
      "epoch": 1.779785775927712,
      "grad_norm": 0.5471748113632202,
      "learning_rate": 8.134761493815254e-06,
      "loss": 0.0363,
      "step": 35060
    },
    {
      "epoch": 1.7800395959185744,
      "grad_norm": 0.347551554441452,
      "learning_rate": 8.133069360542837e-06,
      "loss": 0.0323,
      "step": 35065
    },
    {
      "epoch": 1.780293415909437,
      "grad_norm": 0.38865914940834045,
      "learning_rate": 8.13137722727042e-06,
      "loss": 0.0343,
      "step": 35070
    },
    {
      "epoch": 1.7805472359002996,
      "grad_norm": 0.5800191760063171,
      "learning_rate": 8.129685093998004e-06,
      "loss": 0.0358,
      "step": 35075
    },
    {
      "epoch": 1.780801055891162,
      "grad_norm": 0.4911668300628662,
      "learning_rate": 8.127992960725588e-06,
      "loss": 0.045,
      "step": 35080
    },
    {
      "epoch": 1.7810548758820244,
      "grad_norm": 0.3816292881965637,
      "learning_rate": 8.126300827453171e-06,
      "loss": 0.0387,
      "step": 35085
    },
    {
      "epoch": 1.781308695872887,
      "grad_norm": 0.3930221498012543,
      "learning_rate": 8.124608694180755e-06,
      "loss": 0.0391,
      "step": 35090
    },
    {
      "epoch": 1.7815625158637496,
      "grad_norm": 0.7621637582778931,
      "learning_rate": 8.122916560908338e-06,
      "loss": 0.0381,
      "step": 35095
    },
    {
      "epoch": 1.7818163358546117,
      "grad_norm": 0.2887486517429352,
      "learning_rate": 8.12122442763592e-06,
      "loss": 0.0387,
      "step": 35100
    },
    {
      "epoch": 1.7820701558454743,
      "grad_norm": 0.46208956837654114,
      "learning_rate": 8.119532294363505e-06,
      "loss": 0.0422,
      "step": 35105
    },
    {
      "epoch": 1.782323975836337,
      "grad_norm": 0.34289759397506714,
      "learning_rate": 8.117840161091089e-06,
      "loss": 0.0365,
      "step": 35110
    },
    {
      "epoch": 1.7825777958271993,
      "grad_norm": 0.31571081280708313,
      "learning_rate": 8.116148027818673e-06,
      "loss": 0.0305,
      "step": 35115
    },
    {
      "epoch": 1.7828316158180617,
      "grad_norm": 0.24657775461673737,
      "learning_rate": 8.114455894546254e-06,
      "loss": 0.042,
      "step": 35120
    },
    {
      "epoch": 1.7830854358089243,
      "grad_norm": 0.22031854093074799,
      "learning_rate": 8.112763761273838e-06,
      "loss": 0.035,
      "step": 35125
    },
    {
      "epoch": 1.783339255799787,
      "grad_norm": 0.28018301725387573,
      "learning_rate": 8.111071628001423e-06,
      "loss": 0.0338,
      "step": 35130
    },
    {
      "epoch": 1.7835930757906493,
      "grad_norm": 0.45661869645118713,
      "learning_rate": 8.109379494729005e-06,
      "loss": 0.04,
      "step": 35135
    },
    {
      "epoch": 1.7838468957815117,
      "grad_norm": 0.5636528730392456,
      "learning_rate": 8.107687361456589e-06,
      "loss": 0.0386,
      "step": 35140
    },
    {
      "epoch": 1.7841007157723743,
      "grad_norm": 0.2825608253479004,
      "learning_rate": 8.105995228184172e-06,
      "loss": 0.0397,
      "step": 35145
    },
    {
      "epoch": 1.7843545357632367,
      "grad_norm": 0.3992137312889099,
      "learning_rate": 8.104303094911756e-06,
      "loss": 0.0468,
      "step": 35150
    },
    {
      "epoch": 1.784608355754099,
      "grad_norm": 0.22503553330898285,
      "learning_rate": 8.10261096163934e-06,
      "loss": 0.0354,
      "step": 35155
    },
    {
      "epoch": 1.7848621757449616,
      "grad_norm": 0.36107516288757324,
      "learning_rate": 8.100918828366923e-06,
      "loss": 0.0435,
      "step": 35160
    },
    {
      "epoch": 1.7851159957358242,
      "grad_norm": 0.29463401436805725,
      "learning_rate": 8.099226695094506e-06,
      "loss": 0.0403,
      "step": 35165
    },
    {
      "epoch": 1.7853698157266866,
      "grad_norm": 0.2965399920940399,
      "learning_rate": 8.09753456182209e-06,
      "loss": 0.041,
      "step": 35170
    },
    {
      "epoch": 1.785623635717549,
      "grad_norm": 0.5433416962623596,
      "learning_rate": 8.095842428549673e-06,
      "loss": 0.0434,
      "step": 35175
    },
    {
      "epoch": 1.7858774557084116,
      "grad_norm": 0.29908525943756104,
      "learning_rate": 8.094150295277257e-06,
      "loss": 0.0339,
      "step": 35180
    },
    {
      "epoch": 1.7861312756992742,
      "grad_norm": 0.2853911221027374,
      "learning_rate": 8.09245816200484e-06,
      "loss": 0.0397,
      "step": 35185
    },
    {
      "epoch": 1.7863850956901366,
      "grad_norm": 0.46170076727867126,
      "learning_rate": 8.090766028732424e-06,
      "loss": 0.0401,
      "step": 35190
    },
    {
      "epoch": 1.786638915680999,
      "grad_norm": 0.4234716296195984,
      "learning_rate": 8.089073895460008e-06,
      "loss": 0.0353,
      "step": 35195
    },
    {
      "epoch": 1.7868927356718616,
      "grad_norm": 0.46433013677597046,
      "learning_rate": 8.087381762187591e-06,
      "loss": 0.0373,
      "step": 35200
    },
    {
      "epoch": 1.787146555662724,
      "grad_norm": 0.33198902010917664,
      "learning_rate": 8.085689628915175e-06,
      "loss": 0.0353,
      "step": 35205
    },
    {
      "epoch": 1.7874003756535863,
      "grad_norm": 0.7026520371437073,
      "learning_rate": 8.083997495642756e-06,
      "loss": 0.0425,
      "step": 35210
    },
    {
      "epoch": 1.787654195644449,
      "grad_norm": 0.32603394985198975,
      "learning_rate": 8.08230536237034e-06,
      "loss": 0.0376,
      "step": 35215
    },
    {
      "epoch": 1.7879080156353115,
      "grad_norm": 0.3134501874446869,
      "learning_rate": 8.080613229097925e-06,
      "loss": 0.0419,
      "step": 35220
    },
    {
      "epoch": 1.788161835626174,
      "grad_norm": 0.3281780779361725,
      "learning_rate": 8.078921095825509e-06,
      "loss": 0.0538,
      "step": 35225
    },
    {
      "epoch": 1.7884156556170363,
      "grad_norm": 0.3749331831932068,
      "learning_rate": 8.07722896255309e-06,
      "loss": 0.0365,
      "step": 35230
    },
    {
      "epoch": 1.788669475607899,
      "grad_norm": 0.7166233062744141,
      "learning_rate": 8.075536829280674e-06,
      "loss": 0.051,
      "step": 35235
    },
    {
      "epoch": 1.7889232955987615,
      "grad_norm": 0.8942742347717285,
      "learning_rate": 8.073844696008258e-06,
      "loss": 0.0376,
      "step": 35240
    },
    {
      "epoch": 1.789177115589624,
      "grad_norm": 0.4451989233493805,
      "learning_rate": 8.072152562735841e-06,
      "loss": 0.0315,
      "step": 35245
    },
    {
      "epoch": 1.7894309355804863,
      "grad_norm": 0.3336634635925293,
      "learning_rate": 8.070460429463425e-06,
      "loss": 0.0406,
      "step": 35250
    },
    {
      "epoch": 1.7896847555713489,
      "grad_norm": 0.28594163060188293,
      "learning_rate": 8.068768296191008e-06,
      "loss": 0.0388,
      "step": 35255
    },
    {
      "epoch": 1.7899385755622113,
      "grad_norm": 0.3234853744506836,
      "learning_rate": 8.067076162918592e-06,
      "loss": 0.0389,
      "step": 35260
    },
    {
      "epoch": 1.7901923955530736,
      "grad_norm": 0.3901161551475525,
      "learning_rate": 8.065384029646175e-06,
      "loss": 0.0363,
      "step": 35265
    },
    {
      "epoch": 1.7904462155439362,
      "grad_norm": 0.30935701727867126,
      "learning_rate": 8.063691896373759e-06,
      "loss": 0.0399,
      "step": 35270
    },
    {
      "epoch": 1.7907000355347988,
      "grad_norm": 0.5069534778594971,
      "learning_rate": 8.061999763101343e-06,
      "loss": 0.0408,
      "step": 35275
    },
    {
      "epoch": 1.7909538555256612,
      "grad_norm": 0.20961259305477142,
      "learning_rate": 8.060307629828926e-06,
      "loss": 0.0329,
      "step": 35280
    },
    {
      "epoch": 1.7912076755165236,
      "grad_norm": 0.32768288254737854,
      "learning_rate": 8.05861549655651e-06,
      "loss": 0.0393,
      "step": 35285
    },
    {
      "epoch": 1.7914614955073862,
      "grad_norm": 0.3863822817802429,
      "learning_rate": 8.056923363284093e-06,
      "loss": 0.0361,
      "step": 35290
    },
    {
      "epoch": 1.7917153154982488,
      "grad_norm": 0.40561339259147644,
      "learning_rate": 8.055231230011677e-06,
      "loss": 0.043,
      "step": 35295
    },
    {
      "epoch": 1.791969135489111,
      "grad_norm": 0.43230199813842773,
      "learning_rate": 8.05353909673926e-06,
      "loss": 0.0435,
      "step": 35300
    },
    {
      "epoch": 1.7922229554799736,
      "grad_norm": 0.3429777920246124,
      "learning_rate": 8.051846963466842e-06,
      "loss": 0.0356,
      "step": 35305
    },
    {
      "epoch": 1.7924767754708362,
      "grad_norm": 0.3415292203426361,
      "learning_rate": 8.050154830194427e-06,
      "loss": 0.0375,
      "step": 35310
    },
    {
      "epoch": 1.7927305954616986,
      "grad_norm": 0.260860413312912,
      "learning_rate": 8.048462696922011e-06,
      "loss": 0.0343,
      "step": 35315
    },
    {
      "epoch": 1.792984415452561,
      "grad_norm": 0.41929513216018677,
      "learning_rate": 8.046770563649594e-06,
      "loss": 0.0417,
      "step": 35320
    },
    {
      "epoch": 1.7932382354434235,
      "grad_norm": 0.4116337299346924,
      "learning_rate": 8.045078430377176e-06,
      "loss": 0.0359,
      "step": 35325
    },
    {
      "epoch": 1.7934920554342861,
      "grad_norm": 0.2833936810493469,
      "learning_rate": 8.04338629710476e-06,
      "loss": 0.0345,
      "step": 35330
    },
    {
      "epoch": 1.7937458754251485,
      "grad_norm": 0.36605337262153625,
      "learning_rate": 8.041694163832343e-06,
      "loss": 0.0442,
      "step": 35335
    },
    {
      "epoch": 1.793999695416011,
      "grad_norm": 0.44311726093292236,
      "learning_rate": 8.040002030559929e-06,
      "loss": 0.0378,
      "step": 35340
    },
    {
      "epoch": 1.7942535154068735,
      "grad_norm": 0.3758402466773987,
      "learning_rate": 8.03830989728751e-06,
      "loss": 0.0516,
      "step": 35345
    },
    {
      "epoch": 1.7945073353977359,
      "grad_norm": 0.38618481159210205,
      "learning_rate": 8.036617764015094e-06,
      "loss": 0.0403,
      "step": 35350
    },
    {
      "epoch": 1.7947611553885983,
      "grad_norm": 0.4462096691131592,
      "learning_rate": 8.034925630742678e-06,
      "loss": 0.0391,
      "step": 35355
    },
    {
      "epoch": 1.7950149753794609,
      "grad_norm": 0.34843987226486206,
      "learning_rate": 8.033233497470261e-06,
      "loss": 0.038,
      "step": 35360
    },
    {
      "epoch": 1.7952687953703235,
      "grad_norm": 0.35366612672805786,
      "learning_rate": 8.031541364197845e-06,
      "loss": 0.0506,
      "step": 35365
    },
    {
      "epoch": 1.7955226153611858,
      "grad_norm": 0.2395736426115036,
      "learning_rate": 8.029849230925428e-06,
      "loss": 0.0443,
      "step": 35370
    },
    {
      "epoch": 1.7957764353520482,
      "grad_norm": 0.4678282141685486,
      "learning_rate": 8.028157097653012e-06,
      "loss": 0.0418,
      "step": 35375
    },
    {
      "epoch": 1.7960302553429108,
      "grad_norm": 0.37620967626571655,
      "learning_rate": 8.026464964380595e-06,
      "loss": 0.0458,
      "step": 35380
    },
    {
      "epoch": 1.7962840753337734,
      "grad_norm": 0.4443618059158325,
      "learning_rate": 8.024772831108179e-06,
      "loss": 0.0382,
      "step": 35385
    },
    {
      "epoch": 1.7965378953246358,
      "grad_norm": 0.22134755551815033,
      "learning_rate": 8.023080697835762e-06,
      "loss": 0.0361,
      "step": 35390
    },
    {
      "epoch": 1.7967917153154982,
      "grad_norm": 0.3631151616573334,
      "learning_rate": 8.021388564563346e-06,
      "loss": 0.0411,
      "step": 35395
    },
    {
      "epoch": 1.7970455353063608,
      "grad_norm": 0.29067274928092957,
      "learning_rate": 8.01969643129093e-06,
      "loss": 0.0439,
      "step": 35400
    },
    {
      "epoch": 1.7972993552972232,
      "grad_norm": 0.3182755410671234,
      "learning_rate": 8.018004298018513e-06,
      "loss": 0.0455,
      "step": 35405
    },
    {
      "epoch": 1.7975531752880856,
      "grad_norm": 0.518097996711731,
      "learning_rate": 8.016312164746097e-06,
      "loss": 0.041,
      "step": 35410
    },
    {
      "epoch": 1.7978069952789482,
      "grad_norm": 0.3362354338169098,
      "learning_rate": 8.01462003147368e-06,
      "loss": 0.0403,
      "step": 35415
    },
    {
      "epoch": 1.7980608152698108,
      "grad_norm": 0.23532569408416748,
      "learning_rate": 8.012927898201262e-06,
      "loss": 0.0395,
      "step": 35420
    },
    {
      "epoch": 1.7983146352606731,
      "grad_norm": 0.2703220546245575,
      "learning_rate": 8.011235764928847e-06,
      "loss": 0.0434,
      "step": 35425
    },
    {
      "epoch": 1.7985684552515355,
      "grad_norm": 0.36935853958129883,
      "learning_rate": 8.00954363165643e-06,
      "loss": 0.041,
      "step": 35430
    },
    {
      "epoch": 1.7988222752423981,
      "grad_norm": 0.34497174620628357,
      "learning_rate": 8.007851498384014e-06,
      "loss": 0.0353,
      "step": 35435
    },
    {
      "epoch": 1.7990760952332607,
      "grad_norm": 0.7554081678390503,
      "learning_rate": 8.006159365111596e-06,
      "loss": 0.0431,
      "step": 35440
    },
    {
      "epoch": 1.7993299152241229,
      "grad_norm": 0.32350003719329834,
      "learning_rate": 8.00446723183918e-06,
      "loss": 0.0336,
      "step": 35445
    },
    {
      "epoch": 1.7995837352149855,
      "grad_norm": 0.27758076786994934,
      "learning_rate": 8.002775098566763e-06,
      "loss": 0.0492,
      "step": 35450
    },
    {
      "epoch": 1.799837555205848,
      "grad_norm": 0.29208967089653015,
      "learning_rate": 8.001082965294347e-06,
      "loss": 0.0403,
      "step": 35455
    },
    {
      "epoch": 1.8000913751967105,
      "grad_norm": 0.5288401246070862,
      "learning_rate": 7.99939083202193e-06,
      "loss": 0.0398,
      "step": 35460
    },
    {
      "epoch": 1.8003451951875729,
      "grad_norm": 0.4076966643333435,
      "learning_rate": 7.997698698749514e-06,
      "loss": 0.0368,
      "step": 35465
    },
    {
      "epoch": 1.8005990151784355,
      "grad_norm": 0.3328183889389038,
      "learning_rate": 7.996006565477097e-06,
      "loss": 0.0359,
      "step": 35470
    },
    {
      "epoch": 1.800852835169298,
      "grad_norm": 0.37678319215774536,
      "learning_rate": 7.994314432204681e-06,
      "loss": 0.0408,
      "step": 35475
    },
    {
      "epoch": 1.8011066551601604,
      "grad_norm": 1.1589629650115967,
      "learning_rate": 7.992622298932265e-06,
      "loss": 0.0345,
      "step": 35480
    },
    {
      "epoch": 1.8013604751510228,
      "grad_norm": 0.41570189595222473,
      "learning_rate": 7.990930165659848e-06,
      "loss": 0.0382,
      "step": 35485
    },
    {
      "epoch": 1.8016142951418854,
      "grad_norm": 0.30130791664123535,
      "learning_rate": 7.989238032387432e-06,
      "loss": 0.0313,
      "step": 35490
    },
    {
      "epoch": 1.8018681151327478,
      "grad_norm": 0.5530520081520081,
      "learning_rate": 7.987545899115015e-06,
      "loss": 0.0452,
      "step": 35495
    },
    {
      "epoch": 1.8021219351236102,
      "grad_norm": 0.33491554856300354,
      "learning_rate": 7.985853765842599e-06,
      "loss": 0.0381,
      "step": 35500
    },
    {
      "epoch": 1.8023757551144728,
      "grad_norm": 0.3501150608062744,
      "learning_rate": 7.984161632570182e-06,
      "loss": 0.0517,
      "step": 35505
    },
    {
      "epoch": 1.8026295751053354,
      "grad_norm": 0.25684893131256104,
      "learning_rate": 7.982469499297766e-06,
      "loss": 0.0497,
      "step": 35510
    },
    {
      "epoch": 1.8028833950961978,
      "grad_norm": 0.29228684306144714,
      "learning_rate": 7.98077736602535e-06,
      "loss": 0.0396,
      "step": 35515
    },
    {
      "epoch": 1.8031372150870602,
      "grad_norm": 0.3270423412322998,
      "learning_rate": 7.979085232752933e-06,
      "loss": 0.0396,
      "step": 35520
    },
    {
      "epoch": 1.8033910350779228,
      "grad_norm": 0.3722864091396332,
      "learning_rate": 7.977393099480516e-06,
      "loss": 0.0492,
      "step": 35525
    },
    {
      "epoch": 1.8036448550687854,
      "grad_norm": 0.424862802028656,
      "learning_rate": 7.9757009662081e-06,
      "loss": 0.0461,
      "step": 35530
    },
    {
      "epoch": 1.8038986750596477,
      "grad_norm": 0.4169599413871765,
      "learning_rate": 7.974008832935682e-06,
      "loss": 0.0377,
      "step": 35535
    },
    {
      "epoch": 1.8041524950505101,
      "grad_norm": 0.30000168085098267,
      "learning_rate": 7.972316699663265e-06,
      "loss": 0.039,
      "step": 35540
    },
    {
      "epoch": 1.8044063150413727,
      "grad_norm": 0.4431754946708679,
      "learning_rate": 7.97062456639085e-06,
      "loss": 0.0388,
      "step": 35545
    },
    {
      "epoch": 1.804660135032235,
      "grad_norm": 0.373129278421402,
      "learning_rate": 7.968932433118432e-06,
      "loss": 0.0357,
      "step": 35550
    },
    {
      "epoch": 1.8049139550230975,
      "grad_norm": 0.44162359833717346,
      "learning_rate": 7.967240299846016e-06,
      "loss": 0.0477,
      "step": 35555
    },
    {
      "epoch": 1.80516777501396,
      "grad_norm": 0.35507842898368835,
      "learning_rate": 7.9655481665736e-06,
      "loss": 0.0372,
      "step": 35560
    },
    {
      "epoch": 1.8054215950048227,
      "grad_norm": 0.32913821935653687,
      "learning_rate": 7.963856033301183e-06,
      "loss": 0.033,
      "step": 35565
    },
    {
      "epoch": 1.805675414995685,
      "grad_norm": 0.3151976466178894,
      "learning_rate": 7.962163900028767e-06,
      "loss": 0.0387,
      "step": 35570
    },
    {
      "epoch": 1.8059292349865474,
      "grad_norm": 0.492038369178772,
      "learning_rate": 7.96047176675635e-06,
      "loss": 0.0351,
      "step": 35575
    },
    {
      "epoch": 1.80618305497741,
      "grad_norm": 0.37101173400878906,
      "learning_rate": 7.958779633483934e-06,
      "loss": 0.0396,
      "step": 35580
    },
    {
      "epoch": 1.8064368749682727,
      "grad_norm": 0.24208274483680725,
      "learning_rate": 7.957087500211517e-06,
      "loss": 0.0334,
      "step": 35585
    },
    {
      "epoch": 1.806690694959135,
      "grad_norm": 0.8557137846946716,
      "learning_rate": 7.9553953669391e-06,
      "loss": 0.0385,
      "step": 35590
    },
    {
      "epoch": 1.8069445149499974,
      "grad_norm": 0.33417654037475586,
      "learning_rate": 7.953703233666684e-06,
      "loss": 0.0427,
      "step": 35595
    },
    {
      "epoch": 1.80719833494086,
      "grad_norm": 0.36135685443878174,
      "learning_rate": 7.952011100394268e-06,
      "loss": 0.0433,
      "step": 35600
    },
    {
      "epoch": 1.8074521549317224,
      "grad_norm": 0.6363187432289124,
      "learning_rate": 7.950318967121851e-06,
      "loss": 0.053,
      "step": 35605
    },
    {
      "epoch": 1.8077059749225848,
      "grad_norm": 0.41718024015426636,
      "learning_rate": 7.948626833849435e-06,
      "loss": 0.0349,
      "step": 35610
    },
    {
      "epoch": 1.8079597949134474,
      "grad_norm": 0.3126942813396454,
      "learning_rate": 7.946934700577019e-06,
      "loss": 0.039,
      "step": 35615
    },
    {
      "epoch": 1.80821361490431,
      "grad_norm": 0.34499695897102356,
      "learning_rate": 7.945242567304602e-06,
      "loss": 0.039,
      "step": 35620
    },
    {
      "epoch": 1.8084674348951724,
      "grad_norm": 0.3745400607585907,
      "learning_rate": 7.943550434032184e-06,
      "loss": 0.0413,
      "step": 35625
    },
    {
      "epoch": 1.8087212548860347,
      "grad_norm": 0.34194043278694153,
      "learning_rate": 7.941858300759767e-06,
      "loss": 0.0402,
      "step": 35630
    },
    {
      "epoch": 1.8089750748768973,
      "grad_norm": 0.3365754187107086,
      "learning_rate": 7.940166167487353e-06,
      "loss": 0.0386,
      "step": 35635
    },
    {
      "epoch": 1.8092288948677597,
      "grad_norm": 0.5197813510894775,
      "learning_rate": 7.938474034214936e-06,
      "loss": 0.0395,
      "step": 35640
    },
    {
      "epoch": 1.809482714858622,
      "grad_norm": 0.35244494676589966,
      "learning_rate": 7.936781900942518e-06,
      "loss": 0.0412,
      "step": 35645
    },
    {
      "epoch": 1.8097365348494847,
      "grad_norm": 0.31359171867370605,
      "learning_rate": 7.935089767670102e-06,
      "loss": 0.0339,
      "step": 35650
    },
    {
      "epoch": 1.8099903548403473,
      "grad_norm": 0.32754194736480713,
      "learning_rate": 7.933397634397685e-06,
      "loss": 0.0362,
      "step": 35655
    },
    {
      "epoch": 1.8102441748312097,
      "grad_norm": 0.6635916233062744,
      "learning_rate": 7.93170550112527e-06,
      "loss": 0.0353,
      "step": 35660
    },
    {
      "epoch": 1.810497994822072,
      "grad_norm": 0.3632308840751648,
      "learning_rate": 7.930013367852852e-06,
      "loss": 0.0379,
      "step": 35665
    },
    {
      "epoch": 1.8107518148129347,
      "grad_norm": 0.2154543250799179,
      "learning_rate": 7.928321234580436e-06,
      "loss": 0.0557,
      "step": 35670
    },
    {
      "epoch": 1.8110056348037973,
      "grad_norm": 0.31016165018081665,
      "learning_rate": 7.92662910130802e-06,
      "loss": 0.0429,
      "step": 35675
    },
    {
      "epoch": 1.8112594547946597,
      "grad_norm": 0.3344568610191345,
      "learning_rate": 7.924936968035603e-06,
      "loss": 0.0385,
      "step": 35680
    },
    {
      "epoch": 1.811513274785522,
      "grad_norm": 0.412442684173584,
      "learning_rate": 7.923244834763186e-06,
      "loss": 0.0372,
      "step": 35685
    },
    {
      "epoch": 1.8117670947763846,
      "grad_norm": 0.24281445145606995,
      "learning_rate": 7.92155270149077e-06,
      "loss": 0.0332,
      "step": 35690
    },
    {
      "epoch": 1.812020914767247,
      "grad_norm": 0.512369692325592,
      "learning_rate": 7.919860568218354e-06,
      "loss": 0.0458,
      "step": 35695
    },
    {
      "epoch": 1.8122747347581094,
      "grad_norm": 0.26524874567985535,
      "learning_rate": 7.918168434945937e-06,
      "loss": 0.0309,
      "step": 35700
    },
    {
      "epoch": 1.812528554748972,
      "grad_norm": 0.31035247445106506,
      "learning_rate": 7.91647630167352e-06,
      "loss": 0.0422,
      "step": 35705
    },
    {
      "epoch": 1.8127823747398346,
      "grad_norm": 0.3776507079601288,
      "learning_rate": 7.914784168401104e-06,
      "loss": 0.0363,
      "step": 35710
    },
    {
      "epoch": 1.813036194730697,
      "grad_norm": 0.3676854074001312,
      "learning_rate": 7.913092035128688e-06,
      "loss": 0.0418,
      "step": 35715
    },
    {
      "epoch": 1.8132900147215594,
      "grad_norm": 0.37066182494163513,
      "learning_rate": 7.91139990185627e-06,
      "loss": 0.041,
      "step": 35720
    },
    {
      "epoch": 1.813543834712422,
      "grad_norm": 0.36492326855659485,
      "learning_rate": 7.909707768583855e-06,
      "loss": 0.0403,
      "step": 35725
    },
    {
      "epoch": 1.8137976547032846,
      "grad_norm": 0.36690789461135864,
      "learning_rate": 7.908015635311438e-06,
      "loss": 0.0295,
      "step": 35730
    },
    {
      "epoch": 1.814051474694147,
      "grad_norm": 0.35419976711273193,
      "learning_rate": 7.906323502039022e-06,
      "loss": 0.0438,
      "step": 35735
    },
    {
      "epoch": 1.8143052946850093,
      "grad_norm": 0.24726921319961548,
      "learning_rate": 7.904631368766604e-06,
      "loss": 0.0355,
      "step": 35740
    },
    {
      "epoch": 1.814559114675872,
      "grad_norm": 0.27474740147590637,
      "learning_rate": 7.902939235494187e-06,
      "loss": 0.0389,
      "step": 35745
    },
    {
      "epoch": 1.8148129346667343,
      "grad_norm": 0.37380725145339966,
      "learning_rate": 7.901247102221773e-06,
      "loss": 0.035,
      "step": 35750
    },
    {
      "epoch": 1.8150667546575967,
      "grad_norm": 0.3170168697834015,
      "learning_rate": 7.899554968949356e-06,
      "loss": 0.0416,
      "step": 35755
    },
    {
      "epoch": 1.8153205746484593,
      "grad_norm": 0.26000136137008667,
      "learning_rate": 7.897862835676938e-06,
      "loss": 0.0311,
      "step": 35760
    },
    {
      "epoch": 1.815574394639322,
      "grad_norm": 0.4417473077774048,
      "learning_rate": 7.896170702404521e-06,
      "loss": 0.0433,
      "step": 35765
    },
    {
      "epoch": 1.8158282146301843,
      "grad_norm": 0.23624826967716217,
      "learning_rate": 7.894478569132105e-06,
      "loss": 0.0384,
      "step": 35770
    },
    {
      "epoch": 1.8160820346210467,
      "grad_norm": 0.4043356776237488,
      "learning_rate": 7.892786435859689e-06,
      "loss": 0.0307,
      "step": 35775
    },
    {
      "epoch": 1.8163358546119093,
      "grad_norm": 0.8513171076774597,
      "learning_rate": 7.891094302587272e-06,
      "loss": 0.0451,
      "step": 35780
    },
    {
      "epoch": 1.8165896746027719,
      "grad_norm": 0.4032558500766754,
      "learning_rate": 7.889402169314856e-06,
      "loss": 0.0392,
      "step": 35785
    },
    {
      "epoch": 1.816843494593634,
      "grad_norm": 0.2610934376716614,
      "learning_rate": 7.88771003604244e-06,
      "loss": 0.0439,
      "step": 35790
    },
    {
      "epoch": 1.8170973145844966,
      "grad_norm": 0.2699238061904907,
      "learning_rate": 7.886017902770023e-06,
      "loss": 0.0443,
      "step": 35795
    },
    {
      "epoch": 1.8173511345753592,
      "grad_norm": 0.5964365005493164,
      "learning_rate": 7.884325769497606e-06,
      "loss": 0.0401,
      "step": 35800
    },
    {
      "epoch": 1.8176049545662216,
      "grad_norm": 0.310382217168808,
      "learning_rate": 7.88263363622519e-06,
      "loss": 0.0417,
      "step": 35805
    },
    {
      "epoch": 1.817858774557084,
      "grad_norm": 0.2789522111415863,
      "learning_rate": 7.880941502952773e-06,
      "loss": 0.044,
      "step": 35810
    },
    {
      "epoch": 1.8181125945479466,
      "grad_norm": 0.2450733631849289,
      "learning_rate": 7.879249369680357e-06,
      "loss": 0.0302,
      "step": 35815
    },
    {
      "epoch": 1.8183664145388092,
      "grad_norm": 0.2876873314380646,
      "learning_rate": 7.87755723640794e-06,
      "loss": 0.045,
      "step": 35820
    },
    {
      "epoch": 1.8186202345296716,
      "grad_norm": 0.3161139190196991,
      "learning_rate": 7.875865103135524e-06,
      "loss": 0.0382,
      "step": 35825
    },
    {
      "epoch": 1.818874054520534,
      "grad_norm": 0.3722044825553894,
      "learning_rate": 7.874172969863108e-06,
      "loss": 0.0405,
      "step": 35830
    },
    {
      "epoch": 1.8191278745113966,
      "grad_norm": 0.3855113685131073,
      "learning_rate": 7.87248083659069e-06,
      "loss": 0.0355,
      "step": 35835
    },
    {
      "epoch": 1.819381694502259,
      "grad_norm": 0.7797011733055115,
      "learning_rate": 7.870788703318275e-06,
      "loss": 0.0452,
      "step": 35840
    },
    {
      "epoch": 1.8196355144931213,
      "grad_norm": 0.27733364701271057,
      "learning_rate": 7.869096570045858e-06,
      "loss": 0.0355,
      "step": 35845
    },
    {
      "epoch": 1.819889334483984,
      "grad_norm": 0.3695673942565918,
      "learning_rate": 7.867404436773442e-06,
      "loss": 0.0453,
      "step": 35850
    },
    {
      "epoch": 1.8201431544748465,
      "grad_norm": 0.30087968707084656,
      "learning_rate": 7.865712303501024e-06,
      "loss": 0.0298,
      "step": 35855
    },
    {
      "epoch": 1.820396974465709,
      "grad_norm": 0.2529532313346863,
      "learning_rate": 7.864020170228607e-06,
      "loss": 0.031,
      "step": 35860
    },
    {
      "epoch": 1.8206507944565713,
      "grad_norm": 0.6073928475379944,
      "learning_rate": 7.862328036956192e-06,
      "loss": 0.0365,
      "step": 35865
    },
    {
      "epoch": 1.820904614447434,
      "grad_norm": 0.5613511204719543,
      "learning_rate": 7.860635903683774e-06,
      "loss": 0.0441,
      "step": 35870
    },
    {
      "epoch": 1.8211584344382965,
      "grad_norm": 0.4455784857273102,
      "learning_rate": 7.858943770411358e-06,
      "loss": 0.0373,
      "step": 35875
    },
    {
      "epoch": 1.8214122544291589,
      "grad_norm": 0.5386607050895691,
      "learning_rate": 7.857251637138941e-06,
      "loss": 0.0334,
      "step": 35880
    },
    {
      "epoch": 1.8216660744200213,
      "grad_norm": 0.49189093708992004,
      "learning_rate": 7.855559503866525e-06,
      "loss": 0.0315,
      "step": 35885
    },
    {
      "epoch": 1.8219198944108839,
      "grad_norm": 0.571233332157135,
      "learning_rate": 7.853867370594108e-06,
      "loss": 0.0336,
      "step": 35890
    },
    {
      "epoch": 1.8221737144017462,
      "grad_norm": 0.2096501588821411,
      "learning_rate": 7.852175237321692e-06,
      "loss": 0.03,
      "step": 35895
    },
    {
      "epoch": 1.8224275343926086,
      "grad_norm": 0.32770031690597534,
      "learning_rate": 7.850483104049275e-06,
      "loss": 0.0378,
      "step": 35900
    },
    {
      "epoch": 1.8226813543834712,
      "grad_norm": 0.22983644902706146,
      "learning_rate": 7.848790970776859e-06,
      "loss": 0.0384,
      "step": 35905
    },
    {
      "epoch": 1.8229351743743338,
      "grad_norm": 0.35651785135269165,
      "learning_rate": 7.847098837504443e-06,
      "loss": 0.0374,
      "step": 35910
    },
    {
      "epoch": 1.8231889943651962,
      "grad_norm": 0.3615643084049225,
      "learning_rate": 7.845406704232026e-06,
      "loss": 0.0469,
      "step": 35915
    },
    {
      "epoch": 1.8234428143560586,
      "grad_norm": 0.4260779023170471,
      "learning_rate": 7.84371457095961e-06,
      "loss": 0.057,
      "step": 35920
    },
    {
      "epoch": 1.8236966343469212,
      "grad_norm": 0.2626129388809204,
      "learning_rate": 7.842022437687193e-06,
      "loss": 0.0389,
      "step": 35925
    },
    {
      "epoch": 1.8239504543377838,
      "grad_norm": 0.2752457559108734,
      "learning_rate": 7.840330304414777e-06,
      "loss": 0.0312,
      "step": 35930
    },
    {
      "epoch": 1.824204274328646,
      "grad_norm": 0.41361814737319946,
      "learning_rate": 7.83863817114236e-06,
      "loss": 0.0349,
      "step": 35935
    },
    {
      "epoch": 1.8244580943195086,
      "grad_norm": 0.39695820212364197,
      "learning_rate": 7.836946037869944e-06,
      "loss": 0.041,
      "step": 35940
    },
    {
      "epoch": 1.8247119143103712,
      "grad_norm": 0.3582683503627777,
      "learning_rate": 7.835253904597526e-06,
      "loss": 0.0376,
      "step": 35945
    },
    {
      "epoch": 1.8249657343012335,
      "grad_norm": 0.581875741481781,
      "learning_rate": 7.83356177132511e-06,
      "loss": 0.0327,
      "step": 35950
    },
    {
      "epoch": 1.825219554292096,
      "grad_norm": 0.2829011380672455,
      "learning_rate": 7.831869638052694e-06,
      "loss": 0.0395,
      "step": 35955
    },
    {
      "epoch": 1.8254733742829585,
      "grad_norm": 0.2991785407066345,
      "learning_rate": 7.830177504780278e-06,
      "loss": 0.0367,
      "step": 35960
    },
    {
      "epoch": 1.8257271942738211,
      "grad_norm": 0.6983319520950317,
      "learning_rate": 7.82848537150786e-06,
      "loss": 0.0401,
      "step": 35965
    },
    {
      "epoch": 1.8259810142646835,
      "grad_norm": 0.32356399297714233,
      "learning_rate": 7.826793238235443e-06,
      "loss": 0.0331,
      "step": 35970
    },
    {
      "epoch": 1.826234834255546,
      "grad_norm": 0.4386630058288574,
      "learning_rate": 7.825101104963027e-06,
      "loss": 0.036,
      "step": 35975
    },
    {
      "epoch": 1.8264886542464085,
      "grad_norm": 0.24106809496879578,
      "learning_rate": 7.823408971690612e-06,
      "loss": 0.0419,
      "step": 35980
    },
    {
      "epoch": 1.8267424742372709,
      "grad_norm": 0.2574845552444458,
      "learning_rate": 7.821716838418194e-06,
      "loss": 0.0384,
      "step": 35985
    },
    {
      "epoch": 1.8269962942281333,
      "grad_norm": 0.3407520055770874,
      "learning_rate": 7.820024705145778e-06,
      "loss": 0.0337,
      "step": 35990
    },
    {
      "epoch": 1.8272501142189959,
      "grad_norm": 0.4200335443019867,
      "learning_rate": 7.818332571873361e-06,
      "loss": 0.0374,
      "step": 35995
    },
    {
      "epoch": 1.8275039342098585,
      "grad_norm": 0.434444397687912,
      "learning_rate": 7.816640438600945e-06,
      "loss": 0.0364,
      "step": 36000
    },
    {
      "epoch": 1.8277577542007208,
      "grad_norm": 0.3582698404788971,
      "learning_rate": 7.814948305328528e-06,
      "loss": 0.0395,
      "step": 36005
    },
    {
      "epoch": 1.8280115741915832,
      "grad_norm": 0.3376217484474182,
      "learning_rate": 7.813256172056112e-06,
      "loss": 0.0421,
      "step": 36010
    },
    {
      "epoch": 1.8282653941824458,
      "grad_norm": 0.7544217109680176,
      "learning_rate": 7.811564038783695e-06,
      "loss": 0.0339,
      "step": 36015
    },
    {
      "epoch": 1.8285192141733084,
      "grad_norm": 0.24290993809700012,
      "learning_rate": 7.809871905511279e-06,
      "loss": 0.0427,
      "step": 36020
    },
    {
      "epoch": 1.8287730341641708,
      "grad_norm": 0.24777421355247498,
      "learning_rate": 7.808179772238862e-06,
      "loss": 0.0395,
      "step": 36025
    },
    {
      "epoch": 1.8290268541550332,
      "grad_norm": 0.23828460276126862,
      "learning_rate": 7.806487638966446e-06,
      "loss": 0.0364,
      "step": 36030
    },
    {
      "epoch": 1.8292806741458958,
      "grad_norm": 0.7653145790100098,
      "learning_rate": 7.80479550569403e-06,
      "loss": 0.0551,
      "step": 36035
    },
    {
      "epoch": 1.8295344941367582,
      "grad_norm": 0.519119143486023,
      "learning_rate": 7.803103372421611e-06,
      "loss": 0.037,
      "step": 36040
    },
    {
      "epoch": 1.8297883141276206,
      "grad_norm": 0.3949947953224182,
      "learning_rate": 7.801411239149197e-06,
      "loss": 0.0353,
      "step": 36045
    },
    {
      "epoch": 1.8300421341184832,
      "grad_norm": 0.3932245969772339,
      "learning_rate": 7.79971910587678e-06,
      "loss": 0.0326,
      "step": 36050
    },
    {
      "epoch": 1.8302959541093458,
      "grad_norm": 0.3780907690525055,
      "learning_rate": 7.798026972604364e-06,
      "loss": 0.0327,
      "step": 36055
    },
    {
      "epoch": 1.8305497741002081,
      "grad_norm": 0.22897011041641235,
      "learning_rate": 7.796334839331946e-06,
      "loss": 0.0416,
      "step": 36060
    },
    {
      "epoch": 1.8308035940910705,
      "grad_norm": 0.33317452669143677,
      "learning_rate": 7.794642706059529e-06,
      "loss": 0.0406,
      "step": 36065
    },
    {
      "epoch": 1.8310574140819331,
      "grad_norm": 0.3200877606868744,
      "learning_rate": 7.792950572787114e-06,
      "loss": 0.0415,
      "step": 36070
    },
    {
      "epoch": 1.8313112340727957,
      "grad_norm": 0.47098788619041443,
      "learning_rate": 7.791258439514698e-06,
      "loss": 0.0519,
      "step": 36075
    },
    {
      "epoch": 1.831565054063658,
      "grad_norm": 0.2071748822927475,
      "learning_rate": 7.78956630624228e-06,
      "loss": 0.0348,
      "step": 36080
    },
    {
      "epoch": 1.8318188740545205,
      "grad_norm": 0.8616563081741333,
      "learning_rate": 7.787874172969863e-06,
      "loss": 0.0374,
      "step": 36085
    },
    {
      "epoch": 1.832072694045383,
      "grad_norm": 0.4214298129081726,
      "learning_rate": 7.786182039697447e-06,
      "loss": 0.0459,
      "step": 36090
    },
    {
      "epoch": 1.8323265140362455,
      "grad_norm": 0.6901262998580933,
      "learning_rate": 7.784489906425032e-06,
      "loss": 0.0355,
      "step": 36095
    },
    {
      "epoch": 1.8325803340271078,
      "grad_norm": 0.7623701095581055,
      "learning_rate": 7.782797773152614e-06,
      "loss": 0.0494,
      "step": 36100
    },
    {
      "epoch": 1.8328341540179705,
      "grad_norm": 0.33716341853141785,
      "learning_rate": 7.781105639880197e-06,
      "loss": 0.0375,
      "step": 36105
    },
    {
      "epoch": 1.833087974008833,
      "grad_norm": 0.35571691393852234,
      "learning_rate": 7.779413506607781e-06,
      "loss": 0.0388,
      "step": 36110
    },
    {
      "epoch": 1.8333417939996954,
      "grad_norm": 0.5281639695167542,
      "learning_rate": 7.777721373335364e-06,
      "loss": 0.0364,
      "step": 36115
    },
    {
      "epoch": 1.8335956139905578,
      "grad_norm": 0.5555434823036194,
      "learning_rate": 7.776029240062948e-06,
      "loss": 0.0345,
      "step": 36120
    },
    {
      "epoch": 1.8338494339814204,
      "grad_norm": 0.5790799260139465,
      "learning_rate": 7.774337106790532e-06,
      "loss": 0.0356,
      "step": 36125
    },
    {
      "epoch": 1.834103253972283,
      "grad_norm": 0.5399708151817322,
      "learning_rate": 7.772644973518115e-06,
      "loss": 0.0362,
      "step": 36130
    },
    {
      "epoch": 1.8343570739631452,
      "grad_norm": 0.27483931183815,
      "learning_rate": 7.770952840245699e-06,
      "loss": 0.0347,
      "step": 36135
    },
    {
      "epoch": 1.8346108939540078,
      "grad_norm": 0.29935479164123535,
      "learning_rate": 7.769260706973282e-06,
      "loss": 0.0402,
      "step": 36140
    },
    {
      "epoch": 1.8348647139448704,
      "grad_norm": 0.46089279651641846,
      "learning_rate": 7.767568573700866e-06,
      "loss": 0.0359,
      "step": 36145
    },
    {
      "epoch": 1.8351185339357328,
      "grad_norm": 0.5341860055923462,
      "learning_rate": 7.76587644042845e-06,
      "loss": 0.0398,
      "step": 36150
    },
    {
      "epoch": 1.8353723539265951,
      "grad_norm": 0.3907197117805481,
      "learning_rate": 7.764184307156031e-06,
      "loss": 0.044,
      "step": 36155
    },
    {
      "epoch": 1.8356261739174577,
      "grad_norm": 0.3691658675670624,
      "learning_rate": 7.762492173883616e-06,
      "loss": 0.0367,
      "step": 36160
    },
    {
      "epoch": 1.8358799939083204,
      "grad_norm": 0.40355217456817627,
      "learning_rate": 7.7608000406112e-06,
      "loss": 0.0396,
      "step": 36165
    },
    {
      "epoch": 1.8361338138991827,
      "grad_norm": 0.38757845759391785,
      "learning_rate": 7.759107907338783e-06,
      "loss": 0.0404,
      "step": 36170
    },
    {
      "epoch": 1.8363876338900451,
      "grad_norm": 0.39395466446876526,
      "learning_rate": 7.757415774066365e-06,
      "loss": 0.0405,
      "step": 36175
    },
    {
      "epoch": 1.8366414538809077,
      "grad_norm": 0.41867950558662415,
      "learning_rate": 7.755723640793949e-06,
      "loss": 0.0343,
      "step": 36180
    },
    {
      "epoch": 1.83689527387177,
      "grad_norm": 0.220166876912117,
      "learning_rate": 7.754031507521534e-06,
      "loss": 0.0381,
      "step": 36185
    },
    {
      "epoch": 1.8371490938626325,
      "grad_norm": 0.34739238023757935,
      "learning_rate": 7.752339374249116e-06,
      "loss": 0.0341,
      "step": 36190
    },
    {
      "epoch": 1.837402913853495,
      "grad_norm": 0.37881341576576233,
      "learning_rate": 7.7506472409767e-06,
      "loss": 0.0377,
      "step": 36195
    },
    {
      "epoch": 1.8376567338443577,
      "grad_norm": 0.7510872483253479,
      "learning_rate": 7.748955107704283e-06,
      "loss": 0.0342,
      "step": 36200
    },
    {
      "epoch": 1.83791055383522,
      "grad_norm": 0.7937986254692078,
      "learning_rate": 7.747262974431867e-06,
      "loss": 0.0396,
      "step": 36205
    },
    {
      "epoch": 1.8381643738260824,
      "grad_norm": 0.5486979484558105,
      "learning_rate": 7.74557084115945e-06,
      "loss": 0.0402,
      "step": 36210
    },
    {
      "epoch": 1.838418193816945,
      "grad_norm": 0.31487876176834106,
      "learning_rate": 7.743878707887034e-06,
      "loss": 0.0542,
      "step": 36215
    },
    {
      "epoch": 1.8386720138078076,
      "grad_norm": 0.39621078968048096,
      "learning_rate": 7.742186574614617e-06,
      "loss": 0.0357,
      "step": 36220
    },
    {
      "epoch": 1.83892583379867,
      "grad_norm": 0.4235934913158417,
      "learning_rate": 7.7404944413422e-06,
      "loss": 0.0446,
      "step": 36225
    },
    {
      "epoch": 1.8391796537895324,
      "grad_norm": 0.30609992146492004,
      "learning_rate": 7.738802308069784e-06,
      "loss": 0.0413,
      "step": 36230
    },
    {
      "epoch": 1.839433473780395,
      "grad_norm": 0.47948646545410156,
      "learning_rate": 7.737110174797368e-06,
      "loss": 0.0413,
      "step": 36235
    },
    {
      "epoch": 1.8396872937712574,
      "grad_norm": 0.34571924805641174,
      "learning_rate": 7.735418041524951e-06,
      "loss": 0.0296,
      "step": 36240
    },
    {
      "epoch": 1.8399411137621198,
      "grad_norm": 0.6670829653739929,
      "learning_rate": 7.733725908252535e-06,
      "loss": 0.0469,
      "step": 36245
    },
    {
      "epoch": 1.8401949337529824,
      "grad_norm": 0.29783716797828674,
      "learning_rate": 7.732033774980119e-06,
      "loss": 0.0395,
      "step": 36250
    },
    {
      "epoch": 1.840448753743845,
      "grad_norm": 0.47688236832618713,
      "learning_rate": 7.730341641707702e-06,
      "loss": 0.0398,
      "step": 36255
    },
    {
      "epoch": 1.8407025737347074,
      "grad_norm": 0.278981477022171,
      "learning_rate": 7.728649508435286e-06,
      "loss": 0.0375,
      "step": 36260
    },
    {
      "epoch": 1.8409563937255697,
      "grad_norm": 0.3316904306411743,
      "learning_rate": 7.726957375162867e-06,
      "loss": 0.0416,
      "step": 36265
    },
    {
      "epoch": 1.8412102137164323,
      "grad_norm": 0.33688676357269287,
      "learning_rate": 7.725265241890451e-06,
      "loss": 0.0347,
      "step": 36270
    },
    {
      "epoch": 1.841464033707295,
      "grad_norm": 0.37452802062034607,
      "learning_rate": 7.723573108618036e-06,
      "loss": 0.0481,
      "step": 36275
    },
    {
      "epoch": 1.841717853698157,
      "grad_norm": 0.41721731424331665,
      "learning_rate": 7.72188097534562e-06,
      "loss": 0.0358,
      "step": 36280
    },
    {
      "epoch": 1.8419716736890197,
      "grad_norm": 0.2928161323070526,
      "learning_rate": 7.720188842073202e-06,
      "loss": 0.035,
      "step": 36285
    },
    {
      "epoch": 1.8422254936798823,
      "grad_norm": 0.4405878186225891,
      "learning_rate": 7.718496708800785e-06,
      "loss": 0.0356,
      "step": 36290
    },
    {
      "epoch": 1.8424793136707447,
      "grad_norm": 0.23799636960029602,
      "learning_rate": 7.716804575528369e-06,
      "loss": 0.0364,
      "step": 36295
    },
    {
      "epoch": 1.842733133661607,
      "grad_norm": 0.3568035960197449,
      "learning_rate": 7.715112442255954e-06,
      "loss": 0.0375,
      "step": 36300
    },
    {
      "epoch": 1.8429869536524697,
      "grad_norm": 0.4410528540611267,
      "learning_rate": 7.713420308983536e-06,
      "loss": 0.0423,
      "step": 36305
    },
    {
      "epoch": 1.8432407736433323,
      "grad_norm": 0.30826106667518616,
      "learning_rate": 7.71172817571112e-06,
      "loss": 0.0357,
      "step": 36310
    },
    {
      "epoch": 1.8434945936341947,
      "grad_norm": 0.9397932887077332,
      "learning_rate": 7.710036042438703e-06,
      "loss": 0.0384,
      "step": 36315
    },
    {
      "epoch": 1.843748413625057,
      "grad_norm": 0.4034275412559509,
      "learning_rate": 7.708343909166286e-06,
      "loss": 0.0408,
      "step": 36320
    },
    {
      "epoch": 1.8440022336159196,
      "grad_norm": 0.29389598965644836,
      "learning_rate": 7.70665177589387e-06,
      "loss": 0.0356,
      "step": 36325
    },
    {
      "epoch": 1.844256053606782,
      "grad_norm": 0.3444618284702301,
      "learning_rate": 7.704959642621454e-06,
      "loss": 0.0347,
      "step": 36330
    },
    {
      "epoch": 1.8445098735976444,
      "grad_norm": 0.49377360939979553,
      "learning_rate": 7.703267509349037e-06,
      "loss": 0.0443,
      "step": 36335
    },
    {
      "epoch": 1.844763693588507,
      "grad_norm": 0.2680582106113434,
      "learning_rate": 7.70157537607662e-06,
      "loss": 0.0426,
      "step": 36340
    },
    {
      "epoch": 1.8450175135793696,
      "grad_norm": 0.316648006439209,
      "learning_rate": 7.699883242804204e-06,
      "loss": 0.037,
      "step": 36345
    },
    {
      "epoch": 1.845271333570232,
      "grad_norm": 0.5903269648551941,
      "learning_rate": 7.698191109531788e-06,
      "loss": 0.0419,
      "step": 36350
    },
    {
      "epoch": 1.8455251535610944,
      "grad_norm": 0.35091760754585266,
      "learning_rate": 7.696498976259371e-06,
      "loss": 0.0401,
      "step": 36355
    },
    {
      "epoch": 1.845778973551957,
      "grad_norm": 0.2766816318035126,
      "learning_rate": 7.694806842986953e-06,
      "loss": 0.0362,
      "step": 36360
    },
    {
      "epoch": 1.8460327935428196,
      "grad_norm": 0.38903939723968506,
      "learning_rate": 7.693114709714538e-06,
      "loss": 0.0435,
      "step": 36365
    },
    {
      "epoch": 1.846286613533682,
      "grad_norm": 0.5219350457191467,
      "learning_rate": 7.691422576442122e-06,
      "loss": 0.0498,
      "step": 36370
    },
    {
      "epoch": 1.8465404335245443,
      "grad_norm": 0.26614704728126526,
      "learning_rate": 7.689730443169705e-06,
      "loss": 0.0383,
      "step": 36375
    },
    {
      "epoch": 1.846794253515407,
      "grad_norm": 0.27631068229675293,
      "learning_rate": 7.688038309897287e-06,
      "loss": 0.0356,
      "step": 36380
    },
    {
      "epoch": 1.8470480735062693,
      "grad_norm": 0.3098543584346771,
      "learning_rate": 7.68634617662487e-06,
      "loss": 0.0387,
      "step": 36385
    },
    {
      "epoch": 1.8473018934971317,
      "grad_norm": 0.4202713966369629,
      "learning_rate": 7.684654043352456e-06,
      "loss": 0.0392,
      "step": 36390
    },
    {
      "epoch": 1.8475557134879943,
      "grad_norm": 0.36861708760261536,
      "learning_rate": 7.68296191008004e-06,
      "loss": 0.0441,
      "step": 36395
    },
    {
      "epoch": 1.847809533478857,
      "grad_norm": 0.41563522815704346,
      "learning_rate": 7.681269776807621e-06,
      "loss": 0.0439,
      "step": 36400
    },
    {
      "epoch": 1.8480633534697193,
      "grad_norm": 0.3617202341556549,
      "learning_rate": 7.679577643535205e-06,
      "loss": 0.0337,
      "step": 36405
    },
    {
      "epoch": 1.8483171734605817,
      "grad_norm": 1.1566804647445679,
      "learning_rate": 7.677885510262789e-06,
      "loss": 0.0361,
      "step": 36410
    },
    {
      "epoch": 1.8485709934514443,
      "grad_norm": 0.33016854524612427,
      "learning_rate": 7.676193376990372e-06,
      "loss": 0.0341,
      "step": 36415
    },
    {
      "epoch": 1.8488248134423069,
      "grad_norm": 0.2854905426502228,
      "learning_rate": 7.674501243717956e-06,
      "loss": 0.0459,
      "step": 36420
    },
    {
      "epoch": 1.8490786334331692,
      "grad_norm": 0.3764021694660187,
      "learning_rate": 7.67280911044554e-06,
      "loss": 0.0426,
      "step": 36425
    },
    {
      "epoch": 1.8493324534240316,
      "grad_norm": 0.42459553480148315,
      "learning_rate": 7.671116977173123e-06,
      "loss": 0.0445,
      "step": 36430
    },
    {
      "epoch": 1.8495862734148942,
      "grad_norm": 0.3755717873573303,
      "learning_rate": 7.669424843900706e-06,
      "loss": 0.0364,
      "step": 36435
    },
    {
      "epoch": 1.8498400934057566,
      "grad_norm": 0.3747606575489044,
      "learning_rate": 7.66773271062829e-06,
      "loss": 0.0388,
      "step": 36440
    },
    {
      "epoch": 1.850093913396619,
      "grad_norm": 0.33696579933166504,
      "learning_rate": 7.666040577355873e-06,
      "loss": 0.0309,
      "step": 36445
    },
    {
      "epoch": 1.8503477333874816,
      "grad_norm": 0.37529999017715454,
      "learning_rate": 7.664348444083457e-06,
      "loss": 0.0383,
      "step": 36450
    },
    {
      "epoch": 1.8506015533783442,
      "grad_norm": 0.325459361076355,
      "learning_rate": 7.66265631081104e-06,
      "loss": 0.0422,
      "step": 36455
    },
    {
      "epoch": 1.8508553733692066,
      "grad_norm": 0.33724257349967957,
      "learning_rate": 7.660964177538624e-06,
      "loss": 0.0386,
      "step": 36460
    },
    {
      "epoch": 1.851109193360069,
      "grad_norm": 0.6354991793632507,
      "learning_rate": 7.659272044266208e-06,
      "loss": 0.0389,
      "step": 36465
    },
    {
      "epoch": 1.8513630133509316,
      "grad_norm": 0.5462630391120911,
      "learning_rate": 7.657579910993791e-06,
      "loss": 0.0371,
      "step": 36470
    },
    {
      "epoch": 1.8516168333417942,
      "grad_norm": 0.3498347997665405,
      "learning_rate": 7.655887777721373e-06,
      "loss": 0.0401,
      "step": 36475
    },
    {
      "epoch": 1.8518706533326563,
      "grad_norm": 0.3156206011772156,
      "learning_rate": 7.654195644448958e-06,
      "loss": 0.0364,
      "step": 36480
    },
    {
      "epoch": 1.852124473323519,
      "grad_norm": 0.3194058835506439,
      "learning_rate": 7.652503511176542e-06,
      "loss": 0.0408,
      "step": 36485
    },
    {
      "epoch": 1.8523782933143815,
      "grad_norm": 0.629120409488678,
      "learning_rate": 7.650811377904125e-06,
      "loss": 0.0372,
      "step": 36490
    },
    {
      "epoch": 1.852632113305244,
      "grad_norm": 0.6228652000427246,
      "learning_rate": 7.649119244631707e-06,
      "loss": 0.0449,
      "step": 36495
    },
    {
      "epoch": 1.8528859332961063,
      "grad_norm": 0.3349374532699585,
      "learning_rate": 7.64742711135929e-06,
      "loss": 0.0402,
      "step": 36500
    },
    {
      "epoch": 1.853139753286969,
      "grad_norm": 0.46309956908226013,
      "learning_rate": 7.645734978086874e-06,
      "loss": 0.0352,
      "step": 36505
    },
    {
      "epoch": 1.8533935732778315,
      "grad_norm": 0.49359285831451416,
      "learning_rate": 7.644042844814458e-06,
      "loss": 0.0374,
      "step": 36510
    },
    {
      "epoch": 1.8536473932686939,
      "grad_norm": 0.3246408700942993,
      "learning_rate": 7.642350711542041e-06,
      "loss": 0.0414,
      "step": 36515
    },
    {
      "epoch": 1.8539012132595563,
      "grad_norm": 0.41470158100128174,
      "learning_rate": 7.640658578269625e-06,
      "loss": 0.0425,
      "step": 36520
    },
    {
      "epoch": 1.8541550332504189,
      "grad_norm": 0.28343310952186584,
      "learning_rate": 7.638966444997208e-06,
      "loss": 0.0439,
      "step": 36525
    },
    {
      "epoch": 1.8544088532412812,
      "grad_norm": 0.3936581015586853,
      "learning_rate": 7.637274311724792e-06,
      "loss": 0.0375,
      "step": 36530
    },
    {
      "epoch": 1.8546626732321436,
      "grad_norm": 0.4883519411087036,
      "learning_rate": 7.635582178452375e-06,
      "loss": 0.0387,
      "step": 36535
    },
    {
      "epoch": 1.8549164932230062,
      "grad_norm": 0.368341863155365,
      "learning_rate": 7.633890045179959e-06,
      "loss": 0.0373,
      "step": 36540
    },
    {
      "epoch": 1.8551703132138688,
      "grad_norm": 0.3583613336086273,
      "learning_rate": 7.632197911907543e-06,
      "loss": 0.0414,
      "step": 36545
    },
    {
      "epoch": 1.8554241332047312,
      "grad_norm": 0.36319252848625183,
      "learning_rate": 7.630505778635126e-06,
      "loss": 0.0394,
      "step": 36550
    },
    {
      "epoch": 1.8556779531955936,
      "grad_norm": 0.5051521062850952,
      "learning_rate": 7.62881364536271e-06,
      "loss": 0.05,
      "step": 36555
    },
    {
      "epoch": 1.8559317731864562,
      "grad_norm": 0.357225239276886,
      "learning_rate": 7.627121512090293e-06,
      "loss": 0.0406,
      "step": 36560
    },
    {
      "epoch": 1.8561855931773188,
      "grad_norm": 0.339259535074234,
      "learning_rate": 7.625429378817877e-06,
      "loss": 0.0478,
      "step": 36565
    },
    {
      "epoch": 1.8564394131681812,
      "grad_norm": 0.36899054050445557,
      "learning_rate": 7.6237372455454594e-06,
      "loss": 0.0323,
      "step": 36570
    },
    {
      "epoch": 1.8566932331590436,
      "grad_norm": 0.3415074944496155,
      "learning_rate": 7.622045112273043e-06,
      "loss": 0.038,
      "step": 36575
    },
    {
      "epoch": 1.8569470531499062,
      "grad_norm": 0.39684244990348816,
      "learning_rate": 7.620352979000627e-06,
      "loss": 0.0445,
      "step": 36580
    },
    {
      "epoch": 1.8572008731407685,
      "grad_norm": 0.33541497588157654,
      "learning_rate": 7.618660845728209e-06,
      "loss": 0.038,
      "step": 36585
    },
    {
      "epoch": 1.857454693131631,
      "grad_norm": 0.37271255254745483,
      "learning_rate": 7.616968712455794e-06,
      "loss": 0.0448,
      "step": 36590
    },
    {
      "epoch": 1.8577085131224935,
      "grad_norm": 0.3162631690502167,
      "learning_rate": 7.615276579183377e-06,
      "loss": 0.0317,
      "step": 36595
    },
    {
      "epoch": 1.8579623331133561,
      "grad_norm": 0.3193531036376953,
      "learning_rate": 7.613584445910961e-06,
      "loss": 0.0359,
      "step": 36600
    },
    {
      "epoch": 1.8582161531042185,
      "grad_norm": 0.4613184332847595,
      "learning_rate": 7.611892312638543e-06,
      "loss": 0.0466,
      "step": 36605
    },
    {
      "epoch": 1.8584699730950809,
      "grad_norm": 0.2512896955013275,
      "learning_rate": 7.610200179366127e-06,
      "loss": 0.0356,
      "step": 36610
    },
    {
      "epoch": 1.8587237930859435,
      "grad_norm": 0.44435155391693115,
      "learning_rate": 7.608508046093711e-06,
      "loss": 0.0355,
      "step": 36615
    },
    {
      "epoch": 1.858977613076806,
      "grad_norm": 0.2987118661403656,
      "learning_rate": 7.606815912821295e-06,
      "loss": 0.0478,
      "step": 36620
    },
    {
      "epoch": 1.8592314330676682,
      "grad_norm": 0.33937567472457886,
      "learning_rate": 7.6051237795488776e-06,
      "loss": 0.0426,
      "step": 36625
    },
    {
      "epoch": 1.8594852530585309,
      "grad_norm": 0.3043089509010315,
      "learning_rate": 7.603431646276461e-06,
      "loss": 0.0376,
      "step": 36630
    },
    {
      "epoch": 1.8597390730493935,
      "grad_norm": 0.5690452456474304,
      "learning_rate": 7.601739513004045e-06,
      "loss": 0.0451,
      "step": 36635
    },
    {
      "epoch": 1.8599928930402558,
      "grad_norm": 0.5804987549781799,
      "learning_rate": 7.600047379731629e-06,
      "loss": 0.0395,
      "step": 36640
    },
    {
      "epoch": 1.8602467130311182,
      "grad_norm": 0.37388941645622253,
      "learning_rate": 7.598355246459212e-06,
      "loss": 0.036,
      "step": 36645
    },
    {
      "epoch": 1.8605005330219808,
      "grad_norm": 0.5256741642951965,
      "learning_rate": 7.596663113186795e-06,
      "loss": 0.0372,
      "step": 36650
    },
    {
      "epoch": 1.8607543530128434,
      "grad_norm": 0.30373093485832214,
      "learning_rate": 7.594970979914379e-06,
      "loss": 0.0331,
      "step": 36655
    },
    {
      "epoch": 1.8610081730037058,
      "grad_norm": 0.3162570893764496,
      "learning_rate": 7.593278846641962e-06,
      "loss": 0.0348,
      "step": 36660
    },
    {
      "epoch": 1.8612619929945682,
      "grad_norm": 0.24273575842380524,
      "learning_rate": 7.591586713369545e-06,
      "loss": 0.0488,
      "step": 36665
    },
    {
      "epoch": 1.8615158129854308,
      "grad_norm": 0.4897520840167999,
      "learning_rate": 7.5898945800971295e-06,
      "loss": 0.0478,
      "step": 36670
    },
    {
      "epoch": 1.8617696329762932,
      "grad_norm": 0.3911951780319214,
      "learning_rate": 7.588202446824713e-06,
      "loss": 0.0327,
      "step": 36675
    },
    {
      "epoch": 1.8620234529671555,
      "grad_norm": 0.47243648767471313,
      "learning_rate": 7.586510313552296e-06,
      "loss": 0.0383,
      "step": 36680
    },
    {
      "epoch": 1.8622772729580181,
      "grad_norm": 0.27408725023269653,
      "learning_rate": 7.584818180279879e-06,
      "loss": 0.0384,
      "step": 36685
    },
    {
      "epoch": 1.8625310929488808,
      "grad_norm": 0.34124070405960083,
      "learning_rate": 7.583126047007463e-06,
      "loss": 0.0363,
      "step": 36690
    },
    {
      "epoch": 1.8627849129397431,
      "grad_norm": 0.49087390303611755,
      "learning_rate": 7.581433913735047e-06,
      "loss": 0.0414,
      "step": 36695
    },
    {
      "epoch": 1.8630387329306055,
      "grad_norm": 0.4342269003391266,
      "learning_rate": 7.579741780462629e-06,
      "loss": 0.0321,
      "step": 36700
    },
    {
      "epoch": 1.8632925529214681,
      "grad_norm": 0.26301121711730957,
      "learning_rate": 7.5780496471902134e-06,
      "loss": 0.0372,
      "step": 36705
    },
    {
      "epoch": 1.8635463729123307,
      "grad_norm": 0.543662428855896,
      "learning_rate": 7.576357513917797e-06,
      "loss": 0.049,
      "step": 36710
    },
    {
      "epoch": 1.863800192903193,
      "grad_norm": 0.3795504570007324,
      "learning_rate": 7.5746653806453805e-06,
      "loss": 0.0369,
      "step": 36715
    },
    {
      "epoch": 1.8640540128940555,
      "grad_norm": 0.28618723154067993,
      "learning_rate": 7.572973247372963e-06,
      "loss": 0.0373,
      "step": 36720
    },
    {
      "epoch": 1.864307832884918,
      "grad_norm": 0.33548107743263245,
      "learning_rate": 7.571281114100547e-06,
      "loss": 0.0357,
      "step": 36725
    },
    {
      "epoch": 1.8645616528757805,
      "grad_norm": 0.2927003502845764,
      "learning_rate": 7.569588980828131e-06,
      "loss": 0.0354,
      "step": 36730
    },
    {
      "epoch": 1.8648154728666428,
      "grad_norm": 0.24474462866783142,
      "learning_rate": 7.567896847555715e-06,
      "loss": 0.0307,
      "step": 36735
    },
    {
      "epoch": 1.8650692928575054,
      "grad_norm": 0.2846127152442932,
      "learning_rate": 7.566204714283297e-06,
      "loss": 0.0343,
      "step": 36740
    },
    {
      "epoch": 1.865323112848368,
      "grad_norm": 0.36875009536743164,
      "learning_rate": 7.564512581010881e-06,
      "loss": 0.0402,
      "step": 36745
    },
    {
      "epoch": 1.8655769328392304,
      "grad_norm": 0.36819466948509216,
      "learning_rate": 7.5628204477384645e-06,
      "loss": 0.0337,
      "step": 36750
    },
    {
      "epoch": 1.8658307528300928,
      "grad_norm": 0.2949831187725067,
      "learning_rate": 7.561128314466047e-06,
      "loss": 0.0385,
      "step": 36755
    },
    {
      "epoch": 1.8660845728209554,
      "grad_norm": 0.41179198026657104,
      "learning_rate": 7.5594361811936316e-06,
      "loss": 0.0316,
      "step": 36760
    },
    {
      "epoch": 1.866338392811818,
      "grad_norm": 0.4023495316505432,
      "learning_rate": 7.557744047921215e-06,
      "loss": 0.0421,
      "step": 36765
    },
    {
      "epoch": 1.8665922128026802,
      "grad_norm": 0.5222672820091248,
      "learning_rate": 7.556051914648799e-06,
      "loss": 0.0393,
      "step": 36770
    },
    {
      "epoch": 1.8668460327935428,
      "grad_norm": 0.3291333317756653,
      "learning_rate": 7.554359781376381e-06,
      "loss": 0.0345,
      "step": 36775
    },
    {
      "epoch": 1.8670998527844054,
      "grad_norm": 0.4046681225299835,
      "learning_rate": 7.552667648103965e-06,
      "loss": 0.04,
      "step": 36780
    },
    {
      "epoch": 1.8673536727752678,
      "grad_norm": 0.2864059507846832,
      "learning_rate": 7.550975514831549e-06,
      "loss": 0.0417,
      "step": 36785
    },
    {
      "epoch": 1.8676074927661301,
      "grad_norm": 0.3820878565311432,
      "learning_rate": 7.549283381559133e-06,
      "loss": 0.0376,
      "step": 36790
    },
    {
      "epoch": 1.8678613127569927,
      "grad_norm": 0.3426235020160675,
      "learning_rate": 7.5475912482867155e-06,
      "loss": 0.0376,
      "step": 36795
    },
    {
      "epoch": 1.8681151327478553,
      "grad_norm": 0.43335989117622375,
      "learning_rate": 7.545899115014299e-06,
      "loss": 0.0337,
      "step": 36800
    },
    {
      "epoch": 1.8683689527387177,
      "grad_norm": 0.5109525918960571,
      "learning_rate": 7.544206981741883e-06,
      "loss": 0.0363,
      "step": 36805
    },
    {
      "epoch": 1.86862277272958,
      "grad_norm": 0.3855689465999603,
      "learning_rate": 7.542514848469466e-06,
      "loss": 0.0343,
      "step": 36810
    },
    {
      "epoch": 1.8688765927204427,
      "grad_norm": 0.584434986114502,
      "learning_rate": 7.540822715197049e-06,
      "loss": 0.0342,
      "step": 36815
    },
    {
      "epoch": 1.869130412711305,
      "grad_norm": 1.0270673036575317,
      "learning_rate": 7.539130581924633e-06,
      "loss": 0.0351,
      "step": 36820
    },
    {
      "epoch": 1.8693842327021675,
      "grad_norm": 0.31905609369277954,
      "learning_rate": 7.537438448652217e-06,
      "loss": 0.0461,
      "step": 36825
    },
    {
      "epoch": 1.86963805269303,
      "grad_norm": 0.244539275765419,
      "learning_rate": 7.5357463153797995e-06,
      "loss": 0.0428,
      "step": 36830
    },
    {
      "epoch": 1.8698918726838927,
      "grad_norm": 0.23592790961265564,
      "learning_rate": 7.534054182107383e-06,
      "loss": 0.0348,
      "step": 36835
    },
    {
      "epoch": 1.870145692674755,
      "grad_norm": 0.28931909799575806,
      "learning_rate": 7.532362048834967e-06,
      "loss": 0.0474,
      "step": 36840
    },
    {
      "epoch": 1.8703995126656174,
      "grad_norm": 1.2063060998916626,
      "learning_rate": 7.530669915562551e-06,
      "loss": 0.0409,
      "step": 36845
    },
    {
      "epoch": 1.87065333265648,
      "grad_norm": 1.4404006004333496,
      "learning_rate": 7.528977782290134e-06,
      "loss": 0.0443,
      "step": 36850
    },
    {
      "epoch": 1.8709071526473426,
      "grad_norm": 0.49859338998794556,
      "learning_rate": 7.527285649017717e-06,
      "loss": 0.0356,
      "step": 36855
    },
    {
      "epoch": 1.871160972638205,
      "grad_norm": 0.23417167365550995,
      "learning_rate": 7.525593515745301e-06,
      "loss": 0.0371,
      "step": 36860
    },
    {
      "epoch": 1.8714147926290674,
      "grad_norm": 0.38218954205513,
      "learning_rate": 7.523901382472884e-06,
      "loss": 0.0391,
      "step": 36865
    },
    {
      "epoch": 1.87166861261993,
      "grad_norm": 0.2729610800743103,
      "learning_rate": 7.522209249200467e-06,
      "loss": 0.0422,
      "step": 36870
    },
    {
      "epoch": 1.8719224326107924,
      "grad_norm": 0.2917572855949402,
      "learning_rate": 7.520517115928051e-06,
      "loss": 0.0465,
      "step": 36875
    },
    {
      "epoch": 1.8721762526016548,
      "grad_norm": 0.4478292167186737,
      "learning_rate": 7.518824982655635e-06,
      "loss": 0.0403,
      "step": 36880
    },
    {
      "epoch": 1.8724300725925174,
      "grad_norm": 0.3274686634540558,
      "learning_rate": 7.5171328493832185e-06,
      "loss": 0.037,
      "step": 36885
    },
    {
      "epoch": 1.87268389258338,
      "grad_norm": 0.19864174723625183,
      "learning_rate": 7.515440716110801e-06,
      "loss": 0.0332,
      "step": 36890
    },
    {
      "epoch": 1.8729377125742424,
      "grad_norm": 0.44155529141426086,
      "learning_rate": 7.513748582838385e-06,
      "loss": 0.0431,
      "step": 36895
    },
    {
      "epoch": 1.8731915325651047,
      "grad_norm": 0.3078289031982422,
      "learning_rate": 7.512056449565968e-06,
      "loss": 0.0346,
      "step": 36900
    },
    {
      "epoch": 1.8734453525559673,
      "grad_norm": 0.3366737365722656,
      "learning_rate": 7.510364316293553e-06,
      "loss": 0.0356,
      "step": 36905
    },
    {
      "epoch": 1.87369917254683,
      "grad_norm": 0.35496756434440613,
      "learning_rate": 7.508672183021135e-06,
      "loss": 0.0482,
      "step": 36910
    },
    {
      "epoch": 1.8739529925376923,
      "grad_norm": 0.3293476700782776,
      "learning_rate": 7.506980049748719e-06,
      "loss": 0.0388,
      "step": 36915
    },
    {
      "epoch": 1.8742068125285547,
      "grad_norm": 0.30580469965934753,
      "learning_rate": 7.5052879164763025e-06,
      "loss": 0.0437,
      "step": 36920
    },
    {
      "epoch": 1.8744606325194173,
      "grad_norm": 0.28655534982681274,
      "learning_rate": 7.503595783203885e-06,
      "loss": 0.0531,
      "step": 36925
    },
    {
      "epoch": 1.8747144525102797,
      "grad_norm": 0.33288678526878357,
      "learning_rate": 7.501903649931469e-06,
      "loss": 0.0331,
      "step": 36930
    },
    {
      "epoch": 1.874968272501142,
      "grad_norm": 0.3044753670692444,
      "learning_rate": 7.500211516659053e-06,
      "loss": 0.0472,
      "step": 36935
    },
    {
      "epoch": 1.8752220924920047,
      "grad_norm": 0.3109078109264374,
      "learning_rate": 7.498519383386637e-06,
      "loss": 0.0414,
      "step": 36940
    },
    {
      "epoch": 1.8754759124828673,
      "grad_norm": 0.40895843505859375,
      "learning_rate": 7.496827250114219e-06,
      "loss": 0.0332,
      "step": 36945
    },
    {
      "epoch": 1.8757297324737296,
      "grad_norm": 0.4430285096168518,
      "learning_rate": 7.495135116841803e-06,
      "loss": 0.0465,
      "step": 36950
    },
    {
      "epoch": 1.875983552464592,
      "grad_norm": 0.48985281586647034,
      "learning_rate": 7.4934429835693864e-06,
      "loss": 0.0374,
      "step": 36955
    },
    {
      "epoch": 1.8762373724554546,
      "grad_norm": 0.3518620729446411,
      "learning_rate": 7.491750850296971e-06,
      "loss": 0.0314,
      "step": 36960
    },
    {
      "epoch": 1.8764911924463172,
      "grad_norm": 0.3940987288951874,
      "learning_rate": 7.4900587170245535e-06,
      "loss": 0.0288,
      "step": 36965
    },
    {
      "epoch": 1.8767450124371794,
      "grad_norm": 0.2915908694267273,
      "learning_rate": 7.488366583752137e-06,
      "loss": 0.0365,
      "step": 36970
    },
    {
      "epoch": 1.876998832428042,
      "grad_norm": 0.2620247006416321,
      "learning_rate": 7.486674450479721e-06,
      "loss": 0.0432,
      "step": 36975
    },
    {
      "epoch": 1.8772526524189046,
      "grad_norm": 0.34094831347465515,
      "learning_rate": 7.484982317207304e-06,
      "loss": 0.0345,
      "step": 36980
    },
    {
      "epoch": 1.877506472409767,
      "grad_norm": 0.2771894037723541,
      "learning_rate": 7.483290183934887e-06,
      "loss": 0.0347,
      "step": 36985
    },
    {
      "epoch": 1.8777602924006294,
      "grad_norm": 0.33156320452690125,
      "learning_rate": 7.48159805066247e-06,
      "loss": 0.0369,
      "step": 36990
    },
    {
      "epoch": 1.878014112391492,
      "grad_norm": 0.5050420761108398,
      "learning_rate": 7.479905917390055e-06,
      "loss": 0.0445,
      "step": 36995
    },
    {
      "epoch": 1.8782679323823546,
      "grad_norm": 0.29895123839378357,
      "learning_rate": 7.4782137841176375e-06,
      "loss": 0.0419,
      "step": 37000
    },
    {
      "epoch": 1.878521752373217,
      "grad_norm": 0.27417564392089844,
      "learning_rate": 7.476521650845221e-06,
      "loss": 0.0425,
      "step": 37005
    },
    {
      "epoch": 1.8787755723640793,
      "grad_norm": 0.5192016959190369,
      "learning_rate": 7.4748295175728046e-06,
      "loss": 0.0447,
      "step": 37010
    },
    {
      "epoch": 1.879029392354942,
      "grad_norm": 0.25949370861053467,
      "learning_rate": 7.473137384300388e-06,
      "loss": 0.025,
      "step": 37015
    },
    {
      "epoch": 1.8792832123458043,
      "grad_norm": 0.44494354724884033,
      "learning_rate": 7.471445251027971e-06,
      "loss": 0.0337,
      "step": 37020
    },
    {
      "epoch": 1.8795370323366667,
      "grad_norm": 0.28168588876724243,
      "learning_rate": 7.469753117755555e-06,
      "loss": 0.0419,
      "step": 37025
    },
    {
      "epoch": 1.8797908523275293,
      "grad_norm": 0.2574255168437958,
      "learning_rate": 7.468060984483139e-06,
      "loss": 0.0363,
      "step": 37030
    },
    {
      "epoch": 1.880044672318392,
      "grad_norm": 0.40726467967033386,
      "learning_rate": 7.466368851210722e-06,
      "loss": 0.0407,
      "step": 37035
    },
    {
      "epoch": 1.8802984923092543,
      "grad_norm": 0.3467763364315033,
      "learning_rate": 7.464676717938305e-06,
      "loss": 0.0385,
      "step": 37040
    },
    {
      "epoch": 1.8805523123001167,
      "grad_norm": 0.5550787448883057,
      "learning_rate": 7.4629845846658885e-06,
      "loss": 0.0402,
      "step": 37045
    },
    {
      "epoch": 1.8808061322909793,
      "grad_norm": 0.28454452753067017,
      "learning_rate": 7.461292451393473e-06,
      "loss": 0.0379,
      "step": 37050
    },
    {
      "epoch": 1.8810599522818419,
      "grad_norm": 0.4483475387096405,
      "learning_rate": 7.4596003181210565e-06,
      "loss": 0.0384,
      "step": 37055
    },
    {
      "epoch": 1.8813137722727042,
      "grad_norm": 0.4282529652118683,
      "learning_rate": 7.457908184848639e-06,
      "loss": 0.0336,
      "step": 37060
    },
    {
      "epoch": 1.8815675922635666,
      "grad_norm": 0.532959520816803,
      "learning_rate": 7.456216051576223e-06,
      "loss": 0.0462,
      "step": 37065
    },
    {
      "epoch": 1.8818214122544292,
      "grad_norm": 0.2977970242500305,
      "learning_rate": 7.454523918303806e-06,
      "loss": 0.0372,
      "step": 37070
    },
    {
      "epoch": 1.8820752322452916,
      "grad_norm": 0.2505723536014557,
      "learning_rate": 7.452831785031389e-06,
      "loss": 0.0375,
      "step": 37075
    },
    {
      "epoch": 1.882329052236154,
      "grad_norm": 0.26401594281196594,
      "learning_rate": 7.4511396517589725e-06,
      "loss": 0.0342,
      "step": 37080
    },
    {
      "epoch": 1.8825828722270166,
      "grad_norm": 0.5874828100204468,
      "learning_rate": 7.449447518486557e-06,
      "loss": 0.0385,
      "step": 37085
    },
    {
      "epoch": 1.8828366922178792,
      "grad_norm": 0.33207279443740845,
      "learning_rate": 7.4477553852141404e-06,
      "loss": 0.0335,
      "step": 37090
    },
    {
      "epoch": 1.8830905122087416,
      "grad_norm": 0.4250926077365875,
      "learning_rate": 7.446063251941723e-06,
      "loss": 0.0387,
      "step": 37095
    },
    {
      "epoch": 1.883344332199604,
      "grad_norm": 0.2815552353858948,
      "learning_rate": 7.444371118669307e-06,
      "loss": 0.0383,
      "step": 37100
    },
    {
      "epoch": 1.8835981521904666,
      "grad_norm": 1.2461798191070557,
      "learning_rate": 7.44267898539689e-06,
      "loss": 0.0413,
      "step": 37105
    },
    {
      "epoch": 1.8838519721813292,
      "grad_norm": 0.4171600341796875,
      "learning_rate": 7.440986852124475e-06,
      "loss": 0.0392,
      "step": 37110
    },
    {
      "epoch": 1.8841057921721913,
      "grad_norm": 0.8906102180480957,
      "learning_rate": 7.439294718852057e-06,
      "loss": 0.0302,
      "step": 37115
    },
    {
      "epoch": 1.884359612163054,
      "grad_norm": 0.6476141214370728,
      "learning_rate": 7.437602585579641e-06,
      "loss": 0.0388,
      "step": 37120
    },
    {
      "epoch": 1.8846134321539165,
      "grad_norm": 0.40726473927497864,
      "learning_rate": 7.435910452307224e-06,
      "loss": 0.0429,
      "step": 37125
    },
    {
      "epoch": 1.884867252144779,
      "grad_norm": 0.47000131011009216,
      "learning_rate": 7.434218319034808e-06,
      "loss": 0.0444,
      "step": 37130
    },
    {
      "epoch": 1.8851210721356413,
      "grad_norm": 0.27233800292015076,
      "learning_rate": 7.432526185762391e-06,
      "loss": 0.0351,
      "step": 37135
    },
    {
      "epoch": 1.8853748921265039,
      "grad_norm": 0.3793398141860962,
      "learning_rate": 7.430834052489975e-06,
      "loss": 0.0375,
      "step": 37140
    },
    {
      "epoch": 1.8856287121173665,
      "grad_norm": 0.5338091850280762,
      "learning_rate": 7.429141919217559e-06,
      "loss": 0.0372,
      "step": 37145
    },
    {
      "epoch": 1.8858825321082289,
      "grad_norm": 0.3071553111076355,
      "learning_rate": 7.427449785945141e-06,
      "loss": 0.0355,
      "step": 37150
    },
    {
      "epoch": 1.8861363520990913,
      "grad_norm": 0.39112332463264465,
      "learning_rate": 7.425757652672725e-06,
      "loss": 0.0372,
      "step": 37155
    },
    {
      "epoch": 1.8863901720899539,
      "grad_norm": 1.7440931797027588,
      "learning_rate": 7.424065519400308e-06,
      "loss": 0.0374,
      "step": 37160
    },
    {
      "epoch": 1.8866439920808162,
      "grad_norm": 0.42379260063171387,
      "learning_rate": 7.422373386127893e-06,
      "loss": 0.0429,
      "step": 37165
    },
    {
      "epoch": 1.8868978120716786,
      "grad_norm": 0.2365400493144989,
      "learning_rate": 7.420681252855475e-06,
      "loss": 0.0399,
      "step": 37170
    },
    {
      "epoch": 1.8871516320625412,
      "grad_norm": 0.29219627380371094,
      "learning_rate": 7.418989119583059e-06,
      "loss": 0.034,
      "step": 37175
    },
    {
      "epoch": 1.8874054520534038,
      "grad_norm": 0.42391523718833923,
      "learning_rate": 7.4172969863106425e-06,
      "loss": 0.0434,
      "step": 37180
    },
    {
      "epoch": 1.8876592720442662,
      "grad_norm": 0.7317461967468262,
      "learning_rate": 7.415604853038226e-06,
      "loss": 0.0409,
      "step": 37185
    },
    {
      "epoch": 1.8879130920351286,
      "grad_norm": 0.3624744713306427,
      "learning_rate": 7.413912719765809e-06,
      "loss": 0.0398,
      "step": 37190
    },
    {
      "epoch": 1.8881669120259912,
      "grad_norm": 0.24236305058002472,
      "learning_rate": 7.412220586493392e-06,
      "loss": 0.03,
      "step": 37195
    },
    {
      "epoch": 1.8884207320168538,
      "grad_norm": 0.3188173174858093,
      "learning_rate": 7.410528453220977e-06,
      "loss": 0.0333,
      "step": 37200
    },
    {
      "epoch": 1.8886745520077162,
      "grad_norm": 0.3659869432449341,
      "learning_rate": 7.40883631994856e-06,
      "loss": 0.0348,
      "step": 37205
    },
    {
      "epoch": 1.8889283719985785,
      "grad_norm": 0.4334191679954529,
      "learning_rate": 7.407144186676143e-06,
      "loss": 0.0283,
      "step": 37210
    },
    {
      "epoch": 1.8891821919894412,
      "grad_norm": 0.2886342406272888,
      "learning_rate": 7.4054520534037265e-06,
      "loss": 0.0384,
      "step": 37215
    },
    {
      "epoch": 1.8894360119803035,
      "grad_norm": 0.5727047324180603,
      "learning_rate": 7.40375992013131e-06,
      "loss": 0.0387,
      "step": 37220
    },
    {
      "epoch": 1.889689831971166,
      "grad_norm": 0.2886624336242676,
      "learning_rate": 7.4020677868588944e-06,
      "loss": 0.0333,
      "step": 37225
    },
    {
      "epoch": 1.8899436519620285,
      "grad_norm": 0.9751167893409729,
      "learning_rate": 7.400375653586477e-06,
      "loss": 0.043,
      "step": 37230
    },
    {
      "epoch": 1.8901974719528911,
      "grad_norm": 0.3085179030895233,
      "learning_rate": 7.398683520314061e-06,
      "loss": 0.0424,
      "step": 37235
    },
    {
      "epoch": 1.8904512919437535,
      "grad_norm": 0.29897645115852356,
      "learning_rate": 7.396991387041644e-06,
      "loss": 0.0327,
      "step": 37240
    },
    {
      "epoch": 1.8907051119346159,
      "grad_norm": 0.346880167722702,
      "learning_rate": 7.395299253769227e-06,
      "loss": 0.0367,
      "step": 37245
    },
    {
      "epoch": 1.8909589319254785,
      "grad_norm": 0.4613039195537567,
      "learning_rate": 7.3936071204968105e-06,
      "loss": 0.0449,
      "step": 37250
    },
    {
      "epoch": 1.891212751916341,
      "grad_norm": 0.3410487473011017,
      "learning_rate": 7.391914987224395e-06,
      "loss": 0.0376,
      "step": 37255
    },
    {
      "epoch": 1.8914665719072035,
      "grad_norm": 0.2915606200695038,
      "learning_rate": 7.390222853951978e-06,
      "loss": 0.0431,
      "step": 37260
    },
    {
      "epoch": 1.8917203918980658,
      "grad_norm": 0.48398151993751526,
      "learning_rate": 7.388530720679561e-06,
      "loss": 0.029,
      "step": 37265
    },
    {
      "epoch": 1.8919742118889284,
      "grad_norm": 0.4111987352371216,
      "learning_rate": 7.386838587407145e-06,
      "loss": 0.0377,
      "step": 37270
    },
    {
      "epoch": 1.8922280318797908,
      "grad_norm": 0.3373020887374878,
      "learning_rate": 7.385146454134728e-06,
      "loss": 0.0384,
      "step": 37275
    },
    {
      "epoch": 1.8924818518706532,
      "grad_norm": 0.8100658655166626,
      "learning_rate": 7.383454320862313e-06,
      "loss": 0.0403,
      "step": 37280
    },
    {
      "epoch": 1.8927356718615158,
      "grad_norm": 0.3214988708496094,
      "learning_rate": 7.3817621875898944e-06,
      "loss": 0.0343,
      "step": 37285
    },
    {
      "epoch": 1.8929894918523784,
      "grad_norm": 0.5179448127746582,
      "learning_rate": 7.380070054317479e-06,
      "loss": 0.0385,
      "step": 37290
    },
    {
      "epoch": 1.8932433118432408,
      "grad_norm": 0.28286972641944885,
      "learning_rate": 7.378377921045062e-06,
      "loss": 0.034,
      "step": 37295
    },
    {
      "epoch": 1.8934971318341032,
      "grad_norm": 0.28703486919403076,
      "learning_rate": 7.376685787772646e-06,
      "loss": 0.0351,
      "step": 37300
    },
    {
      "epoch": 1.8937509518249658,
      "grad_norm": 0.23158909380435944,
      "learning_rate": 7.374993654500229e-06,
      "loss": 0.0448,
      "step": 37305
    },
    {
      "epoch": 1.8940047718158284,
      "grad_norm": 0.29262420535087585,
      "learning_rate": 7.373301521227812e-06,
      "loss": 0.0383,
      "step": 37310
    },
    {
      "epoch": 1.8942585918066905,
      "grad_norm": 0.3622038662433624,
      "learning_rate": 7.3716093879553965e-06,
      "loss": 0.0384,
      "step": 37315
    },
    {
      "epoch": 1.8945124117975531,
      "grad_norm": 0.3091678023338318,
      "learning_rate": 7.369917254682979e-06,
      "loss": 0.029,
      "step": 37320
    },
    {
      "epoch": 1.8947662317884157,
      "grad_norm": 0.4038952887058258,
      "learning_rate": 7.368225121410563e-06,
      "loss": 0.0333,
      "step": 37325
    },
    {
      "epoch": 1.8950200517792781,
      "grad_norm": 0.45298081636428833,
      "learning_rate": 7.366532988138146e-06,
      "loss": 0.0381,
      "step": 37330
    },
    {
      "epoch": 1.8952738717701405,
      "grad_norm": 0.3163958191871643,
      "learning_rate": 7.36484085486573e-06,
      "loss": 0.0391,
      "step": 37335
    },
    {
      "epoch": 1.895527691761003,
      "grad_norm": 0.34478557109832764,
      "learning_rate": 7.363148721593313e-06,
      "loss": 0.0322,
      "step": 37340
    },
    {
      "epoch": 1.8957815117518657,
      "grad_norm": 1.6724156141281128,
      "learning_rate": 7.361456588320897e-06,
      "loss": 0.0409,
      "step": 37345
    },
    {
      "epoch": 1.896035331742728,
      "grad_norm": 0.20028600096702576,
      "learning_rate": 7.3597644550484805e-06,
      "loss": 0.0356,
      "step": 37350
    },
    {
      "epoch": 1.8962891517335905,
      "grad_norm": 0.31665441393852234,
      "learning_rate": 7.358072321776064e-06,
      "loss": 0.0409,
      "step": 37355
    },
    {
      "epoch": 1.896542971724453,
      "grad_norm": 0.2620135247707367,
      "learning_rate": 7.356380188503647e-06,
      "loss": 0.0352,
      "step": 37360
    },
    {
      "epoch": 1.8967967917153155,
      "grad_norm": 0.39468491077423096,
      "learning_rate": 7.35468805523123e-06,
      "loss": 0.0369,
      "step": 37365
    },
    {
      "epoch": 1.8970506117061778,
      "grad_norm": 0.2632450759410858,
      "learning_rate": 7.352995921958815e-06,
      "loss": 0.0325,
      "step": 37370
    },
    {
      "epoch": 1.8973044316970404,
      "grad_norm": 0.32689785957336426,
      "learning_rate": 7.351303788686398e-06,
      "loss": 0.0406,
      "step": 37375
    },
    {
      "epoch": 1.897558251687903,
      "grad_norm": 0.4007929563522339,
      "learning_rate": 7.349611655413981e-06,
      "loss": 0.0432,
      "step": 37380
    },
    {
      "epoch": 1.8978120716787654,
      "grad_norm": 0.28323855996131897,
      "learning_rate": 7.3479195221415645e-06,
      "loss": 0.0336,
      "step": 37385
    },
    {
      "epoch": 1.8980658916696278,
      "grad_norm": 0.3580118417739868,
      "learning_rate": 7.346227388869148e-06,
      "loss": 0.04,
      "step": 37390
    },
    {
      "epoch": 1.8983197116604904,
      "grad_norm": 0.31773775815963745,
      "learning_rate": 7.344535255596731e-06,
      "loss": 0.0442,
      "step": 37395
    },
    {
      "epoch": 1.898573531651353,
      "grad_norm": 0.5973715782165527,
      "learning_rate": 7.342843122324314e-06,
      "loss": 0.0448,
      "step": 37400
    },
    {
      "epoch": 1.8988273516422154,
      "grad_norm": 0.30234646797180176,
      "learning_rate": 7.341150989051899e-06,
      "loss": 0.0361,
      "step": 37405
    },
    {
      "epoch": 1.8990811716330778,
      "grad_norm": 0.2853446900844574,
      "learning_rate": 7.339458855779482e-06,
      "loss": 0.0339,
      "step": 37410
    },
    {
      "epoch": 1.8993349916239404,
      "grad_norm": 0.4864497482776642,
      "learning_rate": 7.337766722507065e-06,
      "loss": 0.0444,
      "step": 37415
    },
    {
      "epoch": 1.8995888116148028,
      "grad_norm": 0.35563915967941284,
      "learning_rate": 7.3360745892346484e-06,
      "loss": 0.041,
      "step": 37420
    },
    {
      "epoch": 1.8998426316056651,
      "grad_norm": 0.2955119013786316,
      "learning_rate": 7.334382455962232e-06,
      "loss": 0.0385,
      "step": 37425
    },
    {
      "epoch": 1.9000964515965277,
      "grad_norm": 0.23126500844955444,
      "learning_rate": 7.332690322689816e-06,
      "loss": 0.0299,
      "step": 37430
    },
    {
      "epoch": 1.9003502715873903,
      "grad_norm": 0.2585117220878601,
      "learning_rate": 7.330998189417399e-06,
      "loss": 0.0465,
      "step": 37435
    },
    {
      "epoch": 1.9006040915782527,
      "grad_norm": 0.36390066146850586,
      "learning_rate": 7.329306056144983e-06,
      "loss": 0.037,
      "step": 37440
    },
    {
      "epoch": 1.900857911569115,
      "grad_norm": 0.32259151339530945,
      "learning_rate": 7.327613922872566e-06,
      "loss": 0.0421,
      "step": 37445
    },
    {
      "epoch": 1.9011117315599777,
      "grad_norm": 0.2413652539253235,
      "learning_rate": 7.32592178960015e-06,
      "loss": 0.0296,
      "step": 37450
    },
    {
      "epoch": 1.9013655515508403,
      "grad_norm": 0.25002026557922363,
      "learning_rate": 7.324229656327732e-06,
      "loss": 0.0381,
      "step": 37455
    },
    {
      "epoch": 1.9016193715417025,
      "grad_norm": 0.5289567112922668,
      "learning_rate": 7.322537523055317e-06,
      "loss": 0.0359,
      "step": 37460
    },
    {
      "epoch": 1.901873191532565,
      "grad_norm": 0.2889693081378937,
      "learning_rate": 7.3208453897829e-06,
      "loss": 0.0385,
      "step": 37465
    },
    {
      "epoch": 1.9021270115234277,
      "grad_norm": 0.31875142455101013,
      "learning_rate": 7.319153256510484e-06,
      "loss": 0.0346,
      "step": 37470
    },
    {
      "epoch": 1.90238083151429,
      "grad_norm": 0.5404953360557556,
      "learning_rate": 7.317461123238067e-06,
      "loss": 0.037,
      "step": 37475
    },
    {
      "epoch": 1.9026346515051524,
      "grad_norm": 0.3537548780441284,
      "learning_rate": 7.31576898996565e-06,
      "loss": 0.0361,
      "step": 37480
    },
    {
      "epoch": 1.902888471496015,
      "grad_norm": 0.2968684732913971,
      "learning_rate": 7.314076856693234e-06,
      "loss": 0.0325,
      "step": 37485
    },
    {
      "epoch": 1.9031422914868776,
      "grad_norm": 0.294409841299057,
      "learning_rate": 7.312384723420816e-06,
      "loss": 0.0337,
      "step": 37490
    },
    {
      "epoch": 1.90339611147774,
      "grad_norm": 0.24077461659908295,
      "learning_rate": 7.310692590148401e-06,
      "loss": 0.0409,
      "step": 37495
    },
    {
      "epoch": 1.9036499314686024,
      "grad_norm": 0.3184297978878021,
      "learning_rate": 7.309000456875984e-06,
      "loss": 0.0337,
      "step": 37500
    },
    {
      "epoch": 1.903903751459465,
      "grad_norm": 0.2995074391365051,
      "learning_rate": 7.307308323603568e-06,
      "loss": 0.0387,
      "step": 37505
    },
    {
      "epoch": 1.9041575714503274,
      "grad_norm": 0.587436318397522,
      "learning_rate": 7.3056161903311506e-06,
      "loss": 0.037,
      "step": 37510
    },
    {
      "epoch": 1.9044113914411898,
      "grad_norm": 0.29330456256866455,
      "learning_rate": 7.303924057058734e-06,
      "loss": 0.0388,
      "step": 37515
    },
    {
      "epoch": 1.9046652114320524,
      "grad_norm": 0.24542950093746185,
      "learning_rate": 7.3022319237863185e-06,
      "loss": 0.0416,
      "step": 37520
    },
    {
      "epoch": 1.904919031422915,
      "grad_norm": 0.35747164487838745,
      "learning_rate": 7.300539790513902e-06,
      "loss": 0.0415,
      "step": 37525
    },
    {
      "epoch": 1.9051728514137773,
      "grad_norm": 0.20445160567760468,
      "learning_rate": 7.298847657241485e-06,
      "loss": 0.0381,
      "step": 37530
    },
    {
      "epoch": 1.9054266714046397,
      "grad_norm": 0.8954115509986877,
      "learning_rate": 7.297155523969068e-06,
      "loss": 0.0375,
      "step": 37535
    },
    {
      "epoch": 1.9056804913955023,
      "grad_norm": 0.3642162084579468,
      "learning_rate": 7.295463390696652e-06,
      "loss": 0.0391,
      "step": 37540
    },
    {
      "epoch": 1.905934311386365,
      "grad_norm": 0.2574728727340698,
      "learning_rate": 7.293771257424236e-06,
      "loss": 0.0352,
      "step": 37545
    },
    {
      "epoch": 1.9061881313772273,
      "grad_norm": 0.3355601727962494,
      "learning_rate": 7.292079124151819e-06,
      "loss": 0.0345,
      "step": 37550
    },
    {
      "epoch": 1.9064419513680897,
      "grad_norm": 0.276810884475708,
      "learning_rate": 7.2903869908794025e-06,
      "loss": 0.0401,
      "step": 37555
    },
    {
      "epoch": 1.9066957713589523,
      "grad_norm": 0.31906431913375854,
      "learning_rate": 7.288694857606986e-06,
      "loss": 0.0377,
      "step": 37560
    },
    {
      "epoch": 1.9069495913498147,
      "grad_norm": 0.48261189460754395,
      "learning_rate": 7.287002724334569e-06,
      "loss": 0.0369,
      "step": 37565
    },
    {
      "epoch": 1.907203411340677,
      "grad_norm": 0.34832999110221863,
      "learning_rate": 7.285310591062152e-06,
      "loss": 0.0316,
      "step": 37570
    },
    {
      "epoch": 1.9074572313315397,
      "grad_norm": 0.2636961042881012,
      "learning_rate": 7.283618457789736e-06,
      "loss": 0.0417,
      "step": 37575
    },
    {
      "epoch": 1.9077110513224023,
      "grad_norm": 0.27580955624580383,
      "learning_rate": 7.28192632451732e-06,
      "loss": 0.0423,
      "step": 37580
    },
    {
      "epoch": 1.9079648713132646,
      "grad_norm": 0.2866416573524475,
      "learning_rate": 7.280234191244903e-06,
      "loss": 0.0363,
      "step": 37585
    },
    {
      "epoch": 1.908218691304127,
      "grad_norm": 1.0502089262008667,
      "learning_rate": 7.278542057972486e-06,
      "loss": 0.0338,
      "step": 37590
    },
    {
      "epoch": 1.9084725112949896,
      "grad_norm": 0.6088945865631104,
      "learning_rate": 7.27684992470007e-06,
      "loss": 0.0454,
      "step": 37595
    },
    {
      "epoch": 1.9087263312858522,
      "grad_norm": 0.3789014220237732,
      "learning_rate": 7.2751577914276535e-06,
      "loss": 0.034,
      "step": 37600
    },
    {
      "epoch": 1.9089801512767144,
      "grad_norm": 0.2999269962310791,
      "learning_rate": 7.273465658155236e-06,
      "loss": 0.0372,
      "step": 37605
    },
    {
      "epoch": 1.909233971267577,
      "grad_norm": 0.3169345557689667,
      "learning_rate": 7.271773524882821e-06,
      "loss": 0.0334,
      "step": 37610
    },
    {
      "epoch": 1.9094877912584396,
      "grad_norm": 0.2436423897743225,
      "learning_rate": 7.270081391610404e-06,
      "loss": 0.0354,
      "step": 37615
    },
    {
      "epoch": 1.909741611249302,
      "grad_norm": 0.34217819571495056,
      "learning_rate": 7.268389258337988e-06,
      "loss": 0.0417,
      "step": 37620
    },
    {
      "epoch": 1.9099954312401644,
      "grad_norm": 0.35862669348716736,
      "learning_rate": 7.26669712506557e-06,
      "loss": 0.0357,
      "step": 37625
    },
    {
      "epoch": 1.910249251231027,
      "grad_norm": 0.36769625544548035,
      "learning_rate": 7.265004991793154e-06,
      "loss": 0.031,
      "step": 37630
    },
    {
      "epoch": 1.9105030712218896,
      "grad_norm": 0.2944765090942383,
      "learning_rate": 7.263312858520738e-06,
      "loss": 0.0344,
      "step": 37635
    },
    {
      "epoch": 1.910756891212752,
      "grad_norm": 0.2816435396671295,
      "learning_rate": 7.261620725248321e-06,
      "loss": 0.0334,
      "step": 37640
    },
    {
      "epoch": 1.9110107112036143,
      "grad_norm": 0.369663804769516,
      "learning_rate": 7.2599285919759046e-06,
      "loss": 0.0442,
      "step": 37645
    },
    {
      "epoch": 1.911264531194477,
      "grad_norm": 0.3657083213329315,
      "learning_rate": 7.258236458703488e-06,
      "loss": 0.041,
      "step": 37650
    },
    {
      "epoch": 1.9115183511853393,
      "grad_norm": 0.34254011511802673,
      "learning_rate": 7.256544325431072e-06,
      "loss": 0.0373,
      "step": 37655
    },
    {
      "epoch": 1.9117721711762017,
      "grad_norm": 0.41274362802505493,
      "learning_rate": 7.254852192158654e-06,
      "loss": 0.0418,
      "step": 37660
    },
    {
      "epoch": 1.9120259911670643,
      "grad_norm": 0.33585357666015625,
      "learning_rate": 7.253160058886238e-06,
      "loss": 0.0304,
      "step": 37665
    },
    {
      "epoch": 1.912279811157927,
      "grad_norm": 0.26096150279045105,
      "learning_rate": 7.251467925613822e-06,
      "loss": 0.0365,
      "step": 37670
    },
    {
      "epoch": 1.9125336311487893,
      "grad_norm": 0.4009125828742981,
      "learning_rate": 7.249775792341406e-06,
      "loss": 0.0338,
      "step": 37675
    },
    {
      "epoch": 1.9127874511396517,
      "grad_norm": 0.4178220331668854,
      "learning_rate": 7.2480836590689885e-06,
      "loss": 0.0388,
      "step": 37680
    },
    {
      "epoch": 1.9130412711305143,
      "grad_norm": 0.4047532379627228,
      "learning_rate": 7.246391525796572e-06,
      "loss": 0.0393,
      "step": 37685
    },
    {
      "epoch": 1.9132950911213769,
      "grad_norm": 0.28076454997062683,
      "learning_rate": 7.244699392524156e-06,
      "loss": 0.0364,
      "step": 37690
    },
    {
      "epoch": 1.9135489111122392,
      "grad_norm": 0.4012969136238098,
      "learning_rate": 7.24300725925174e-06,
      "loss": 0.0322,
      "step": 37695
    },
    {
      "epoch": 1.9138027311031016,
      "grad_norm": 0.40359947085380554,
      "learning_rate": 7.241315125979323e-06,
      "loss": 0.0397,
      "step": 37700
    },
    {
      "epoch": 1.9140565510939642,
      "grad_norm": 0.42775213718414307,
      "learning_rate": 7.239622992706906e-06,
      "loss": 0.0362,
      "step": 37705
    },
    {
      "epoch": 1.9143103710848266,
      "grad_norm": 0.2496214509010315,
      "learning_rate": 7.23793085943449e-06,
      "loss": 0.0368,
      "step": 37710
    },
    {
      "epoch": 1.914564191075689,
      "grad_norm": 0.3519715666770935,
      "learning_rate": 7.2362387261620725e-06,
      "loss": 0.042,
      "step": 37715
    },
    {
      "epoch": 1.9148180110665516,
      "grad_norm": 0.30283480882644653,
      "learning_rate": 7.234546592889656e-06,
      "loss": 0.0321,
      "step": 37720
    },
    {
      "epoch": 1.9150718310574142,
      "grad_norm": 0.3165220618247986,
      "learning_rate": 7.2328544596172404e-06,
      "loss": 0.0388,
      "step": 37725
    },
    {
      "epoch": 1.9153256510482766,
      "grad_norm": 0.499083012342453,
      "learning_rate": 7.231162326344824e-06,
      "loss": 0.038,
      "step": 37730
    },
    {
      "epoch": 1.915579471039139,
      "grad_norm": 0.38413795828819275,
      "learning_rate": 7.229470193072407e-06,
      "loss": 0.0499,
      "step": 37735
    },
    {
      "epoch": 1.9158332910300016,
      "grad_norm": 0.5062192678451538,
      "learning_rate": 7.22777805979999e-06,
      "loss": 0.0455,
      "step": 37740
    },
    {
      "epoch": 1.9160871110208642,
      "grad_norm": 0.2117767184972763,
      "learning_rate": 7.226085926527574e-06,
      "loss": 0.0374,
      "step": 37745
    },
    {
      "epoch": 1.9163409310117265,
      "grad_norm": 0.46465274691581726,
      "learning_rate": 7.224393793255158e-06,
      "loss": 0.0408,
      "step": 37750
    },
    {
      "epoch": 1.916594751002589,
      "grad_norm": 0.5478437542915344,
      "learning_rate": 7.22270165998274e-06,
      "loss": 0.036,
      "step": 37755
    },
    {
      "epoch": 1.9168485709934515,
      "grad_norm": 0.30504342913627625,
      "learning_rate": 7.221009526710324e-06,
      "loss": 0.0404,
      "step": 37760
    },
    {
      "epoch": 1.917102390984314,
      "grad_norm": 0.3328668177127838,
      "learning_rate": 7.219317393437908e-06,
      "loss": 0.0428,
      "step": 37765
    },
    {
      "epoch": 1.9173562109751763,
      "grad_norm": 0.276241660118103,
      "learning_rate": 7.2176252601654915e-06,
      "loss": 0.0364,
      "step": 37770
    },
    {
      "epoch": 1.9176100309660389,
      "grad_norm": 0.2813354432582855,
      "learning_rate": 7.215933126893074e-06,
      "loss": 0.0316,
      "step": 37775
    },
    {
      "epoch": 1.9178638509569015,
      "grad_norm": 0.33498886227607727,
      "learning_rate": 7.214240993620658e-06,
      "loss": 0.0321,
      "step": 37780
    },
    {
      "epoch": 1.9181176709477639,
      "grad_norm": 0.5148637890815735,
      "learning_rate": 7.212548860348242e-06,
      "loss": 0.0417,
      "step": 37785
    },
    {
      "epoch": 1.9183714909386262,
      "grad_norm": 0.38031625747680664,
      "learning_rate": 7.210856727075826e-06,
      "loss": 0.0364,
      "step": 37790
    },
    {
      "epoch": 1.9186253109294888,
      "grad_norm": 0.2922528088092804,
      "learning_rate": 7.209164593803408e-06,
      "loss": 0.0414,
      "step": 37795
    },
    {
      "epoch": 1.9188791309203515,
      "grad_norm": 0.27960649132728577,
      "learning_rate": 7.207472460530992e-06,
      "loss": 0.035,
      "step": 37800
    },
    {
      "epoch": 1.9191329509112136,
      "grad_norm": 0.34974753856658936,
      "learning_rate": 7.2057803272585754e-06,
      "loss": 0.0354,
      "step": 37805
    },
    {
      "epoch": 1.9193867709020762,
      "grad_norm": 0.28370293974876404,
      "learning_rate": 7.204088193986158e-06,
      "loss": 0.032,
      "step": 37810
    },
    {
      "epoch": 1.9196405908929388,
      "grad_norm": 0.47823625802993774,
      "learning_rate": 7.2023960607137425e-06,
      "loss": 0.0387,
      "step": 37815
    },
    {
      "epoch": 1.9198944108838012,
      "grad_norm": 0.3467170000076294,
      "learning_rate": 7.200703927441326e-06,
      "loss": 0.0345,
      "step": 37820
    },
    {
      "epoch": 1.9201482308746636,
      "grad_norm": 0.44955742359161377,
      "learning_rate": 7.19901179416891e-06,
      "loss": 0.0415,
      "step": 37825
    },
    {
      "epoch": 1.9204020508655262,
      "grad_norm": 0.4269384741783142,
      "learning_rate": 7.197319660896492e-06,
      "loss": 0.037,
      "step": 37830
    },
    {
      "epoch": 1.9206558708563888,
      "grad_norm": 0.3061307668685913,
      "learning_rate": 7.195627527624076e-06,
      "loss": 0.0335,
      "step": 37835
    },
    {
      "epoch": 1.9209096908472512,
      "grad_norm": 0.34396281838417053,
      "learning_rate": 7.19393539435166e-06,
      "loss": 0.0323,
      "step": 37840
    },
    {
      "epoch": 1.9211635108381135,
      "grad_norm": 0.3059838116168976,
      "learning_rate": 7.192243261079244e-06,
      "loss": 0.0388,
      "step": 37845
    },
    {
      "epoch": 1.9214173308289761,
      "grad_norm": 0.3554377555847168,
      "learning_rate": 7.1905511278068265e-06,
      "loss": 0.0445,
      "step": 37850
    },
    {
      "epoch": 1.9216711508198385,
      "grad_norm": 0.251293420791626,
      "learning_rate": 7.18885899453441e-06,
      "loss": 0.0361,
      "step": 37855
    },
    {
      "epoch": 1.921924970810701,
      "grad_norm": 0.39309418201446533,
      "learning_rate": 7.187166861261994e-06,
      "loss": 0.0464,
      "step": 37860
    },
    {
      "epoch": 1.9221787908015635,
      "grad_norm": 0.5594707727432251,
      "learning_rate": 7.185474727989578e-06,
      "loss": 0.0373,
      "step": 37865
    },
    {
      "epoch": 1.922432610792426,
      "grad_norm": 0.5243818163871765,
      "learning_rate": 7.18378259471716e-06,
      "loss": 0.0343,
      "step": 37870
    },
    {
      "epoch": 1.9226864307832885,
      "grad_norm": 0.49607399106025696,
      "learning_rate": 7.182090461444744e-06,
      "loss": 0.0386,
      "step": 37875
    },
    {
      "epoch": 1.9229402507741509,
      "grad_norm": 0.3219578266143799,
      "learning_rate": 7.180398328172328e-06,
      "loss": 0.0405,
      "step": 37880
    },
    {
      "epoch": 1.9231940707650135,
      "grad_norm": 0.23030859231948853,
      "learning_rate": 7.1787061948999105e-06,
      "loss": 0.0472,
      "step": 37885
    },
    {
      "epoch": 1.923447890755876,
      "grad_norm": 0.3197329640388489,
      "learning_rate": 7.177014061627494e-06,
      "loss": 0.0329,
      "step": 37890
    },
    {
      "epoch": 1.9237017107467385,
      "grad_norm": 0.2778763473033905,
      "learning_rate": 7.1753219283550776e-06,
      "loss": 0.0314,
      "step": 37895
    },
    {
      "epoch": 1.9239555307376008,
      "grad_norm": 0.22582057118415833,
      "learning_rate": 7.173629795082662e-06,
      "loss": 0.0375,
      "step": 37900
    },
    {
      "epoch": 1.9242093507284634,
      "grad_norm": 0.2623133659362793,
      "learning_rate": 7.171937661810245e-06,
      "loss": 0.0367,
      "step": 37905
    },
    {
      "epoch": 1.9244631707193258,
      "grad_norm": 0.30197757482528687,
      "learning_rate": 7.170245528537828e-06,
      "loss": 0.0359,
      "step": 37910
    },
    {
      "epoch": 1.9247169907101882,
      "grad_norm": 0.39974135160446167,
      "learning_rate": 7.168553395265412e-06,
      "loss": 0.0438,
      "step": 37915
    },
    {
      "epoch": 1.9249708107010508,
      "grad_norm": 0.8644199967384338,
      "learning_rate": 7.166861261992995e-06,
      "loss": 0.0375,
      "step": 37920
    },
    {
      "epoch": 1.9252246306919134,
      "grad_norm": 0.33568376302719116,
      "learning_rate": 7.165169128720578e-06,
      "loss": 0.0351,
      "step": 37925
    },
    {
      "epoch": 1.9254784506827758,
      "grad_norm": 0.2705657482147217,
      "learning_rate": 7.163476995448162e-06,
      "loss": 0.0408,
      "step": 37930
    },
    {
      "epoch": 1.9257322706736382,
      "grad_norm": 0.27909040451049805,
      "learning_rate": 7.161784862175746e-06,
      "loss": 0.035,
      "step": 37935
    },
    {
      "epoch": 1.9259860906645008,
      "grad_norm": 1.79963219165802,
      "learning_rate": 7.1600927289033295e-06,
      "loss": 0.0386,
      "step": 37940
    },
    {
      "epoch": 1.9262399106553634,
      "grad_norm": 0.2219754010438919,
      "learning_rate": 7.158400595630912e-06,
      "loss": 0.0331,
      "step": 37945
    },
    {
      "epoch": 1.9264937306462255,
      "grad_norm": 0.5909177660942078,
      "learning_rate": 7.156708462358496e-06,
      "loss": 0.0411,
      "step": 37950
    },
    {
      "epoch": 1.9267475506370881,
      "grad_norm": 0.4270838797092438,
      "learning_rate": 7.15501632908608e-06,
      "loss": 0.0323,
      "step": 37955
    },
    {
      "epoch": 1.9270013706279507,
      "grad_norm": 0.4282520115375519,
      "learning_rate": 7.153324195813662e-06,
      "loss": 0.0381,
      "step": 37960
    },
    {
      "epoch": 1.9272551906188131,
      "grad_norm": 0.2768152952194214,
      "learning_rate": 7.151632062541246e-06,
      "loss": 0.047,
      "step": 37965
    },
    {
      "epoch": 1.9275090106096755,
      "grad_norm": 0.30975621938705444,
      "learning_rate": 7.14993992926883e-06,
      "loss": 0.037,
      "step": 37970
    },
    {
      "epoch": 1.927762830600538,
      "grad_norm": 0.5023228526115417,
      "learning_rate": 7.148247795996413e-06,
      "loss": 0.0396,
      "step": 37975
    },
    {
      "epoch": 1.9280166505914007,
      "grad_norm": 0.4292077124118805,
      "learning_rate": 7.146555662723996e-06,
      "loss": 0.0379,
      "step": 37980
    },
    {
      "epoch": 1.928270470582263,
      "grad_norm": 0.37395018339157104,
      "learning_rate": 7.14486352945158e-06,
      "loss": 0.0314,
      "step": 37985
    },
    {
      "epoch": 1.9285242905731255,
      "grad_norm": 0.24250783026218414,
      "learning_rate": 7.143171396179164e-06,
      "loss": 0.0376,
      "step": 37990
    },
    {
      "epoch": 1.928778110563988,
      "grad_norm": 0.3052719235420227,
      "learning_rate": 7.141479262906748e-06,
      "loss": 0.0367,
      "step": 37995
    },
    {
      "epoch": 1.9290319305548504,
      "grad_norm": 0.34796974062919617,
      "learning_rate": 7.13978712963433e-06,
      "loss": 0.0391,
      "step": 38000
    },
    {
      "epoch": 1.9292857505457128,
      "grad_norm": 0.2718687653541565,
      "learning_rate": 7.138094996361914e-06,
      "loss": 0.0285,
      "step": 38005
    },
    {
      "epoch": 1.9295395705365754,
      "grad_norm": 0.6020230650901794,
      "learning_rate": 7.136402863089497e-06,
      "loss": 0.0329,
      "step": 38010
    },
    {
      "epoch": 1.929793390527438,
      "grad_norm": 0.4138783812522888,
      "learning_rate": 7.134710729817082e-06,
      "loss": 0.037,
      "step": 38015
    },
    {
      "epoch": 1.9300472105183004,
      "grad_norm": 0.31362026929855347,
      "learning_rate": 7.1330185965446645e-06,
      "loss": 0.0294,
      "step": 38020
    },
    {
      "epoch": 1.9303010305091628,
      "grad_norm": 1.2694065570831299,
      "learning_rate": 7.131326463272248e-06,
      "loss": 0.035,
      "step": 38025
    },
    {
      "epoch": 1.9305548505000254,
      "grad_norm": 0.3924097418785095,
      "learning_rate": 7.1296343299998316e-06,
      "loss": 0.0359,
      "step": 38030
    },
    {
      "epoch": 1.930808670490888,
      "grad_norm": 0.3452089726924896,
      "learning_rate": 7.127942196727415e-06,
      "loss": 0.0381,
      "step": 38035
    },
    {
      "epoch": 1.9310624904817504,
      "grad_norm": 0.3343926668167114,
      "learning_rate": 7.126250063454998e-06,
      "loss": 0.0392,
      "step": 38040
    },
    {
      "epoch": 1.9313163104726128,
      "grad_norm": 0.3243504762649536,
      "learning_rate": 7.124557930182582e-06,
      "loss": 0.0421,
      "step": 38045
    },
    {
      "epoch": 1.9315701304634754,
      "grad_norm": 0.30533623695373535,
      "learning_rate": 7.122865796910166e-06,
      "loss": 0.0439,
      "step": 38050
    },
    {
      "epoch": 1.9318239504543377,
      "grad_norm": 0.4469950795173645,
      "learning_rate": 7.1211736636377484e-06,
      "loss": 0.0366,
      "step": 38055
    },
    {
      "epoch": 1.9320777704452001,
      "grad_norm": 0.4529452919960022,
      "learning_rate": 7.119481530365332e-06,
      "loss": 0.0326,
      "step": 38060
    },
    {
      "epoch": 1.9323315904360627,
      "grad_norm": 0.2496260106563568,
      "learning_rate": 7.1177893970929155e-06,
      "loss": 0.0322,
      "step": 38065
    },
    {
      "epoch": 1.9325854104269253,
      "grad_norm": 0.3679202198982239,
      "learning_rate": 7.116097263820499e-06,
      "loss": 0.0371,
      "step": 38070
    },
    {
      "epoch": 1.9328392304177877,
      "grad_norm": 0.527133047580719,
      "learning_rate": 7.114405130548082e-06,
      "loss": 0.0381,
      "step": 38075
    },
    {
      "epoch": 1.93309305040865,
      "grad_norm": 0.7933568358421326,
      "learning_rate": 7.112712997275666e-06,
      "loss": 0.0389,
      "step": 38080
    },
    {
      "epoch": 1.9333468703995127,
      "grad_norm": 0.37088605761528015,
      "learning_rate": 7.11102086400325e-06,
      "loss": 0.0372,
      "step": 38085
    },
    {
      "epoch": 1.9336006903903753,
      "grad_norm": 0.2773051857948303,
      "learning_rate": 7.109328730730833e-06,
      "loss": 0.0344,
      "step": 38090
    },
    {
      "epoch": 1.9338545103812377,
      "grad_norm": 0.4214688837528229,
      "learning_rate": 7.107636597458416e-06,
      "loss": 0.0436,
      "step": 38095
    },
    {
      "epoch": 1.9341083303721,
      "grad_norm": 0.3429591953754425,
      "learning_rate": 7.1059444641859995e-06,
      "loss": 0.038,
      "step": 38100
    },
    {
      "epoch": 1.9343621503629627,
      "grad_norm": 0.6051746606826782,
      "learning_rate": 7.104252330913584e-06,
      "loss": 0.0414,
      "step": 38105
    },
    {
      "epoch": 1.934615970353825,
      "grad_norm": 0.27440497279167175,
      "learning_rate": 7.1025601976411674e-06,
      "loss": 0.0397,
      "step": 38110
    },
    {
      "epoch": 1.9348697903446874,
      "grad_norm": 0.2947607934474945,
      "learning_rate": 7.10086806436875e-06,
      "loss": 0.0413,
      "step": 38115
    },
    {
      "epoch": 1.93512361033555,
      "grad_norm": 0.2716275453567505,
      "learning_rate": 7.099175931096334e-06,
      "loss": 0.0338,
      "step": 38120
    },
    {
      "epoch": 1.9353774303264126,
      "grad_norm": 0.20705972611904144,
      "learning_rate": 7.097483797823917e-06,
      "loss": 0.0317,
      "step": 38125
    },
    {
      "epoch": 1.935631250317275,
      "grad_norm": 0.2767583727836609,
      "learning_rate": 7.0957916645515e-06,
      "loss": 0.029,
      "step": 38130
    },
    {
      "epoch": 1.9358850703081374,
      "grad_norm": 0.31235581636428833,
      "learning_rate": 7.094099531279084e-06,
      "loss": 0.0376,
      "step": 38135
    },
    {
      "epoch": 1.936138890299,
      "grad_norm": 0.26473572850227356,
      "learning_rate": 7.092407398006668e-06,
      "loss": 0.0388,
      "step": 38140
    },
    {
      "epoch": 1.9363927102898626,
      "grad_norm": 0.38204026222229004,
      "learning_rate": 7.090715264734251e-06,
      "loss": 0.0362,
      "step": 38145
    },
    {
      "epoch": 1.9366465302807248,
      "grad_norm": 0.39236733317375183,
      "learning_rate": 7.089023131461834e-06,
      "loss": 0.0344,
      "step": 38150
    },
    {
      "epoch": 1.9369003502715874,
      "grad_norm": 0.27405235171318054,
      "learning_rate": 7.087330998189418e-06,
      "loss": 0.0355,
      "step": 38155
    },
    {
      "epoch": 1.93715417026245,
      "grad_norm": 0.2893589437007904,
      "learning_rate": 7.085638864917001e-06,
      "loss": 0.0405,
      "step": 38160
    },
    {
      "epoch": 1.9374079902533123,
      "grad_norm": 0.3183000981807709,
      "learning_rate": 7.0839467316445856e-06,
      "loss": 0.036,
      "step": 38165
    },
    {
      "epoch": 1.9376618102441747,
      "grad_norm": 0.3648395836353302,
      "learning_rate": 7.082254598372168e-06,
      "loss": 0.0423,
      "step": 38170
    },
    {
      "epoch": 1.9379156302350373,
      "grad_norm": 0.4205324351787567,
      "learning_rate": 7.080562465099752e-06,
      "loss": 0.0311,
      "step": 38175
    },
    {
      "epoch": 1.9381694502259,
      "grad_norm": 0.4777676463127136,
      "learning_rate": 7.078870331827335e-06,
      "loss": 0.0374,
      "step": 38180
    },
    {
      "epoch": 1.9384232702167623,
      "grad_norm": 0.3146344721317291,
      "learning_rate": 7.077178198554919e-06,
      "loss": 0.0326,
      "step": 38185
    },
    {
      "epoch": 1.9386770902076247,
      "grad_norm": 0.38266465067863464,
      "learning_rate": 7.075486065282502e-06,
      "loss": 0.0363,
      "step": 38190
    },
    {
      "epoch": 1.9389309101984873,
      "grad_norm": 0.18472445011138916,
      "learning_rate": 7.073793932010086e-06,
      "loss": 0.0414,
      "step": 38195
    },
    {
      "epoch": 1.9391847301893497,
      "grad_norm": 0.4240174889564514,
      "learning_rate": 7.0721017987376695e-06,
      "loss": 0.0421,
      "step": 38200
    },
    {
      "epoch": 1.939438550180212,
      "grad_norm": 0.3366274833679199,
      "learning_rate": 7.070409665465252e-06,
      "loss": 0.0358,
      "step": 38205
    },
    {
      "epoch": 1.9396923701710747,
      "grad_norm": 0.3867480754852295,
      "learning_rate": 7.068717532192836e-06,
      "loss": 0.0391,
      "step": 38210
    },
    {
      "epoch": 1.9399461901619373,
      "grad_norm": 0.32904496788978577,
      "learning_rate": 7.067025398920419e-06,
      "loss": 0.0323,
      "step": 38215
    },
    {
      "epoch": 1.9402000101527996,
      "grad_norm": 0.3309951424598694,
      "learning_rate": 7.065333265648004e-06,
      "loss": 0.0338,
      "step": 38220
    },
    {
      "epoch": 1.940453830143662,
      "grad_norm": 0.5160368084907532,
      "learning_rate": 7.063641132375586e-06,
      "loss": 0.033,
      "step": 38225
    },
    {
      "epoch": 1.9407076501345246,
      "grad_norm": 0.4503108859062195,
      "learning_rate": 7.06194899910317e-06,
      "loss": 0.0369,
      "step": 38230
    },
    {
      "epoch": 1.9409614701253872,
      "grad_norm": 0.39195284247398376,
      "learning_rate": 7.0602568658307535e-06,
      "loss": 0.0303,
      "step": 38235
    },
    {
      "epoch": 1.9412152901162496,
      "grad_norm": 0.28088903427124023,
      "learning_rate": 7.058564732558337e-06,
      "loss": 0.0302,
      "step": 38240
    },
    {
      "epoch": 1.941469110107112,
      "grad_norm": 0.4266859292984009,
      "learning_rate": 7.05687259928592e-06,
      "loss": 0.0408,
      "step": 38245
    },
    {
      "epoch": 1.9417229300979746,
      "grad_norm": 0.31425270438194275,
      "learning_rate": 7.055180466013503e-06,
      "loss": 0.0326,
      "step": 38250
    },
    {
      "epoch": 1.941976750088837,
      "grad_norm": 0.6080331802368164,
      "learning_rate": 7.053488332741088e-06,
      "loss": 0.0427,
      "step": 38255
    },
    {
      "epoch": 1.9422305700796993,
      "grad_norm": 0.4568061828613281,
      "learning_rate": 7.051796199468671e-06,
      "loss": 0.0406,
      "step": 38260
    },
    {
      "epoch": 1.942484390070562,
      "grad_norm": 0.2914579212665558,
      "learning_rate": 7.050104066196254e-06,
      "loss": 0.0364,
      "step": 38265
    },
    {
      "epoch": 1.9427382100614246,
      "grad_norm": 0.31048232316970825,
      "learning_rate": 7.0484119329238375e-06,
      "loss": 0.036,
      "step": 38270
    },
    {
      "epoch": 1.942992030052287,
      "grad_norm": 0.40617647767066956,
      "learning_rate": 7.046719799651421e-06,
      "loss": 0.0346,
      "step": 38275
    },
    {
      "epoch": 1.9432458500431493,
      "grad_norm": 0.2798089385032654,
      "learning_rate": 7.045027666379004e-06,
      "loss": 0.0327,
      "step": 38280
    },
    {
      "epoch": 1.943499670034012,
      "grad_norm": 1.6191227436065674,
      "learning_rate": 7.043335533106588e-06,
      "loss": 0.0295,
      "step": 38285
    },
    {
      "epoch": 1.9437534900248745,
      "grad_norm": 0.3303340971469879,
      "learning_rate": 7.041643399834172e-06,
      "loss": 0.054,
      "step": 38290
    },
    {
      "epoch": 1.9440073100157367,
      "grad_norm": 0.3839823007583618,
      "learning_rate": 7.039951266561755e-06,
      "loss": 0.0386,
      "step": 38295
    },
    {
      "epoch": 1.9442611300065993,
      "grad_norm": 0.29821422696113586,
      "learning_rate": 7.038259133289338e-06,
      "loss": 0.031,
      "step": 38300
    },
    {
      "epoch": 1.9445149499974619,
      "grad_norm": 0.23630665242671967,
      "learning_rate": 7.0365670000169214e-06,
      "loss": 0.0336,
      "step": 38305
    },
    {
      "epoch": 1.9447687699883243,
      "grad_norm": 0.45323365926742554,
      "learning_rate": 7.034874866744506e-06,
      "loss": 0.0312,
      "step": 38310
    },
    {
      "epoch": 1.9450225899791866,
      "grad_norm": 0.33891797065734863,
      "learning_rate": 7.033182733472089e-06,
      "loss": 0.034,
      "step": 38315
    },
    {
      "epoch": 1.9452764099700492,
      "grad_norm": 0.23988015949726105,
      "learning_rate": 7.031490600199672e-06,
      "loss": 0.0358,
      "step": 38320
    },
    {
      "epoch": 1.9455302299609118,
      "grad_norm": 0.32248857617378235,
      "learning_rate": 7.029798466927256e-06,
      "loss": 0.0375,
      "step": 38325
    },
    {
      "epoch": 1.9457840499517742,
      "grad_norm": 0.34360724687576294,
      "learning_rate": 7.028106333654839e-06,
      "loss": 0.0415,
      "step": 38330
    },
    {
      "epoch": 1.9460378699426366,
      "grad_norm": 0.25492408871650696,
      "learning_rate": 7.0264142003824235e-06,
      "loss": 0.0312,
      "step": 38335
    },
    {
      "epoch": 1.9462916899334992,
      "grad_norm": 0.3143220543861389,
      "learning_rate": 7.024722067110005e-06,
      "loss": 0.0292,
      "step": 38340
    },
    {
      "epoch": 1.9465455099243616,
      "grad_norm": 0.31019699573516846,
      "learning_rate": 7.02302993383759e-06,
      "loss": 0.0309,
      "step": 38345
    },
    {
      "epoch": 1.946799329915224,
      "grad_norm": 0.3752652406692505,
      "learning_rate": 7.021337800565173e-06,
      "loss": 0.0387,
      "step": 38350
    },
    {
      "epoch": 1.9470531499060866,
      "grad_norm": 0.4151340425014496,
      "learning_rate": 7.019645667292757e-06,
      "loss": 0.044,
      "step": 38355
    },
    {
      "epoch": 1.9473069698969492,
      "grad_norm": 0.3452011048793793,
      "learning_rate": 7.0179535340203396e-06,
      "loss": 0.0361,
      "step": 38360
    },
    {
      "epoch": 1.9475607898878116,
      "grad_norm": 0.34209251403808594,
      "learning_rate": 7.016261400747923e-06,
      "loss": 0.044,
      "step": 38365
    },
    {
      "epoch": 1.947814609878674,
      "grad_norm": 0.28460708260536194,
      "learning_rate": 7.0145692674755075e-06,
      "loss": 0.0402,
      "step": 38370
    },
    {
      "epoch": 1.9480684298695365,
      "grad_norm": 0.40208396315574646,
      "learning_rate": 7.01287713420309e-06,
      "loss": 0.046,
      "step": 38375
    },
    {
      "epoch": 1.9483222498603991,
      "grad_norm": 0.4834042489528656,
      "learning_rate": 7.011185000930674e-06,
      "loss": 0.0434,
      "step": 38380
    },
    {
      "epoch": 1.9485760698512615,
      "grad_norm": 0.2941204011440277,
      "learning_rate": 7.009492867658257e-06,
      "loss": 0.0431,
      "step": 38385
    },
    {
      "epoch": 1.948829889842124,
      "grad_norm": 0.34474998712539673,
      "learning_rate": 7.007800734385841e-06,
      "loss": 0.0402,
      "step": 38390
    },
    {
      "epoch": 1.9490837098329865,
      "grad_norm": 0.3636670410633087,
      "learning_rate": 7.0061086011134235e-06,
      "loss": 0.038,
      "step": 38395
    },
    {
      "epoch": 1.949337529823849,
      "grad_norm": 0.2243959754705429,
      "learning_rate": 7.004416467841008e-06,
      "loss": 0.036,
      "step": 38400
    },
    {
      "epoch": 1.9495913498147113,
      "grad_norm": 0.31654590368270874,
      "learning_rate": 7.0027243345685915e-06,
      "loss": 0.0382,
      "step": 38405
    },
    {
      "epoch": 1.9498451698055739,
      "grad_norm": 0.308843195438385,
      "learning_rate": 7.001032201296175e-06,
      "loss": 0.036,
      "step": 38410
    },
    {
      "epoch": 1.9500989897964365,
      "grad_norm": 0.2872798442840576,
      "learning_rate": 6.999340068023758e-06,
      "loss": 0.0336,
      "step": 38415
    },
    {
      "epoch": 1.9503528097872989,
      "grad_norm": 0.42075660824775696,
      "learning_rate": 6.997647934751341e-06,
      "loss": 0.0405,
      "step": 38420
    },
    {
      "epoch": 1.9506066297781612,
      "grad_norm": 0.5195392370223999,
      "learning_rate": 6.995955801478926e-06,
      "loss": 0.0454,
      "step": 38425
    },
    {
      "epoch": 1.9508604497690238,
      "grad_norm": 0.3235759437084198,
      "learning_rate": 6.994263668206509e-06,
      "loss": 0.031,
      "step": 38430
    },
    {
      "epoch": 1.9511142697598864,
      "grad_norm": 0.31546446681022644,
      "learning_rate": 6.992571534934092e-06,
      "loss": 0.031,
      "step": 38435
    },
    {
      "epoch": 1.9513680897507488,
      "grad_norm": 0.4008297026157379,
      "learning_rate": 6.9908794016616754e-06,
      "loss": 0.0281,
      "step": 38440
    },
    {
      "epoch": 1.9516219097416112,
      "grad_norm": 0.3607640862464905,
      "learning_rate": 6.989187268389259e-06,
      "loss": 0.041,
      "step": 38445
    },
    {
      "epoch": 1.9518757297324738,
      "grad_norm": 0.34919342398643494,
      "learning_rate": 6.987495135116842e-06,
      "loss": 0.032,
      "step": 38450
    },
    {
      "epoch": 1.9521295497233362,
      "grad_norm": 0.27828821539878845,
      "learning_rate": 6.985803001844425e-06,
      "loss": 0.0367,
      "step": 38455
    },
    {
      "epoch": 1.9523833697141986,
      "grad_norm": 0.40238264203071594,
      "learning_rate": 6.98411086857201e-06,
      "loss": 0.047,
      "step": 38460
    },
    {
      "epoch": 1.9526371897050612,
      "grad_norm": 0.2481817901134491,
      "learning_rate": 6.982418735299593e-06,
      "loss": 0.037,
      "step": 38465
    },
    {
      "epoch": 1.9528910096959238,
      "grad_norm": 0.3263671100139618,
      "learning_rate": 6.980726602027176e-06,
      "loss": 0.0342,
      "step": 38470
    },
    {
      "epoch": 1.9531448296867862,
      "grad_norm": 0.925449788570404,
      "learning_rate": 6.979034468754759e-06,
      "loss": 0.0351,
      "step": 38475
    },
    {
      "epoch": 1.9533986496776485,
      "grad_norm": 0.2783336341381073,
      "learning_rate": 6.977342335482343e-06,
      "loss": 0.0349,
      "step": 38480
    },
    {
      "epoch": 1.9536524696685111,
      "grad_norm": 0.2663723826408386,
      "learning_rate": 6.975650202209927e-06,
      "loss": 0.0349,
      "step": 38485
    },
    {
      "epoch": 1.9539062896593735,
      "grad_norm": 0.7413889765739441,
      "learning_rate": 6.97395806893751e-06,
      "loss": 0.0414,
      "step": 38490
    },
    {
      "epoch": 1.954160109650236,
      "grad_norm": 0.39749935269355774,
      "learning_rate": 6.972265935665094e-06,
      "loss": 0.0397,
      "step": 38495
    },
    {
      "epoch": 1.9544139296410985,
      "grad_norm": 0.3025566339492798,
      "learning_rate": 6.970573802392677e-06,
      "loss": 0.0372,
      "step": 38500
    },
    {
      "epoch": 1.954667749631961,
      "grad_norm": 0.35124385356903076,
      "learning_rate": 6.968881669120261e-06,
      "loss": 0.036,
      "step": 38505
    },
    {
      "epoch": 1.9549215696228235,
      "grad_norm": 0.4823085367679596,
      "learning_rate": 6.967189535847843e-06,
      "loss": 0.035,
      "step": 38510
    },
    {
      "epoch": 1.9551753896136859,
      "grad_norm": 0.3143712282180786,
      "learning_rate": 6.965497402575428e-06,
      "loss": 0.04,
      "step": 38515
    },
    {
      "epoch": 1.9554292096045485,
      "grad_norm": 0.24416399002075195,
      "learning_rate": 6.963805269303011e-06,
      "loss": 0.0351,
      "step": 38520
    },
    {
      "epoch": 1.955683029595411,
      "grad_norm": 0.3662688732147217,
      "learning_rate": 6.962113136030594e-06,
      "loss": 0.0332,
      "step": 38525
    },
    {
      "epoch": 1.9559368495862735,
      "grad_norm": 0.25640466809272766,
      "learning_rate": 6.9604210027581775e-06,
      "loss": 0.0368,
      "step": 38530
    },
    {
      "epoch": 1.9561906695771358,
      "grad_norm": 0.27777099609375,
      "learning_rate": 6.958728869485761e-06,
      "loss": 0.0268,
      "step": 38535
    },
    {
      "epoch": 1.9564444895679984,
      "grad_norm": 0.7067055106163025,
      "learning_rate": 6.9570367362133455e-06,
      "loss": 0.0328,
      "step": 38540
    },
    {
      "epoch": 1.9566983095588608,
      "grad_norm": 0.3245342969894409,
      "learning_rate": 6.955344602940927e-06,
      "loss": 0.0296,
      "step": 38545
    },
    {
      "epoch": 1.9569521295497232,
      "grad_norm": 0.3933471739292145,
      "learning_rate": 6.953652469668512e-06,
      "loss": 0.0396,
      "step": 38550
    },
    {
      "epoch": 1.9572059495405858,
      "grad_norm": 0.2676612436771393,
      "learning_rate": 6.951960336396095e-06,
      "loss": 0.0323,
      "step": 38555
    },
    {
      "epoch": 1.9574597695314484,
      "grad_norm": 0.3796135485172272,
      "learning_rate": 6.950268203123679e-06,
      "loss": 0.0391,
      "step": 38560
    },
    {
      "epoch": 1.9577135895223108,
      "grad_norm": 0.529050886631012,
      "learning_rate": 6.9485760698512615e-06,
      "loss": 0.0365,
      "step": 38565
    },
    {
      "epoch": 1.9579674095131732,
      "grad_norm": 0.3404449224472046,
      "learning_rate": 6.946883936578845e-06,
      "loss": 0.0326,
      "step": 38570
    },
    {
      "epoch": 1.9582212295040358,
      "grad_norm": 0.30739548802375793,
      "learning_rate": 6.9451918033064294e-06,
      "loss": 0.0415,
      "step": 38575
    },
    {
      "epoch": 1.9584750494948984,
      "grad_norm": 0.2338048368692398,
      "learning_rate": 6.943499670034013e-06,
      "loss": 0.0363,
      "step": 38580
    },
    {
      "epoch": 1.9587288694857607,
      "grad_norm": 0.4167294204235077,
      "learning_rate": 6.941807536761596e-06,
      "loss": 0.0339,
      "step": 38585
    },
    {
      "epoch": 1.9589826894766231,
      "grad_norm": 0.27878156304359436,
      "learning_rate": 6.940115403489179e-06,
      "loss": 0.0286,
      "step": 38590
    },
    {
      "epoch": 1.9592365094674857,
      "grad_norm": 0.3524043560028076,
      "learning_rate": 6.938423270216763e-06,
      "loss": 0.0397,
      "step": 38595
    },
    {
      "epoch": 1.9594903294583481,
      "grad_norm": 0.38988733291625977,
      "learning_rate": 6.936731136944347e-06,
      "loss": 0.0358,
      "step": 38600
    },
    {
      "epoch": 1.9597441494492105,
      "grad_norm": 0.3674320876598358,
      "learning_rate": 6.93503900367193e-06,
      "loss": 0.0395,
      "step": 38605
    },
    {
      "epoch": 1.959997969440073,
      "grad_norm": 0.31990566849708557,
      "learning_rate": 6.933346870399513e-06,
      "loss": 0.0271,
      "step": 38610
    },
    {
      "epoch": 1.9602517894309357,
      "grad_norm": 0.2604389786720276,
      "learning_rate": 6.931654737127097e-06,
      "loss": 0.0322,
      "step": 38615
    },
    {
      "epoch": 1.960505609421798,
      "grad_norm": 0.477633535861969,
      "learning_rate": 6.92996260385468e-06,
      "loss": 0.0521,
      "step": 38620
    },
    {
      "epoch": 1.9607594294126605,
      "grad_norm": 0.3733793795108795,
      "learning_rate": 6.928270470582263e-06,
      "loss": 0.0356,
      "step": 38625
    },
    {
      "epoch": 1.961013249403523,
      "grad_norm": 0.3341461718082428,
      "learning_rate": 6.926578337309848e-06,
      "loss": 0.0363,
      "step": 38630
    },
    {
      "epoch": 1.9612670693943857,
      "grad_norm": 0.29843002557754517,
      "learning_rate": 6.924886204037431e-06,
      "loss": 0.0367,
      "step": 38635
    },
    {
      "epoch": 1.9615208893852478,
      "grad_norm": 0.34912988543510437,
      "learning_rate": 6.923194070765014e-06,
      "loss": 0.0367,
      "step": 38640
    },
    {
      "epoch": 1.9617747093761104,
      "grad_norm": 0.4012235105037689,
      "learning_rate": 6.921501937492597e-06,
      "loss": 0.0369,
      "step": 38645
    },
    {
      "epoch": 1.962028529366973,
      "grad_norm": 0.33369386196136475,
      "learning_rate": 6.919809804220181e-06,
      "loss": 0.0381,
      "step": 38650
    },
    {
      "epoch": 1.9622823493578354,
      "grad_norm": 0.33917364478111267,
      "learning_rate": 6.9181176709477645e-06,
      "loss": 0.0394,
      "step": 38655
    },
    {
      "epoch": 1.9625361693486978,
      "grad_norm": 0.3566811680793762,
      "learning_rate": 6.916425537675347e-06,
      "loss": 0.0346,
      "step": 38660
    },
    {
      "epoch": 1.9627899893395604,
      "grad_norm": 0.35312744975090027,
      "learning_rate": 6.9147334044029316e-06,
      "loss": 0.0314,
      "step": 38665
    },
    {
      "epoch": 1.963043809330423,
      "grad_norm": 0.5093296766281128,
      "learning_rate": 6.913041271130515e-06,
      "loss": 0.037,
      "step": 38670
    },
    {
      "epoch": 1.9632976293212854,
      "grad_norm": 0.3182208836078644,
      "learning_rate": 6.911349137858099e-06,
      "loss": 0.033,
      "step": 38675
    },
    {
      "epoch": 1.9635514493121478,
      "grad_norm": 0.4049065411090851,
      "learning_rate": 6.909657004585681e-06,
      "loss": 0.0318,
      "step": 38680
    },
    {
      "epoch": 1.9638052693030104,
      "grad_norm": 0.2679682970046997,
      "learning_rate": 6.907964871313265e-06,
      "loss": 0.0338,
      "step": 38685
    },
    {
      "epoch": 1.9640590892938727,
      "grad_norm": 0.34067901968955994,
      "learning_rate": 6.906272738040849e-06,
      "loss": 0.037,
      "step": 38690
    },
    {
      "epoch": 1.9643129092847351,
      "grad_norm": 0.4129829406738281,
      "learning_rate": 6.904580604768432e-06,
      "loss": 0.0366,
      "step": 38695
    },
    {
      "epoch": 1.9645667292755977,
      "grad_norm": 0.25145605206489563,
      "learning_rate": 6.9028884714960155e-06,
      "loss": 0.0326,
      "step": 38700
    },
    {
      "epoch": 1.9648205492664603,
      "grad_norm": 0.24122470617294312,
      "learning_rate": 6.901196338223599e-06,
      "loss": 0.0414,
      "step": 38705
    },
    {
      "epoch": 1.9650743692573227,
      "grad_norm": 0.29186809062957764,
      "learning_rate": 6.899504204951183e-06,
      "loss": 0.0323,
      "step": 38710
    },
    {
      "epoch": 1.965328189248185,
      "grad_norm": 0.3512170612812042,
      "learning_rate": 6.897812071678765e-06,
      "loss": 0.0324,
      "step": 38715
    },
    {
      "epoch": 1.9655820092390477,
      "grad_norm": 0.392165869474411,
      "learning_rate": 6.89611993840635e-06,
      "loss": 0.0411,
      "step": 38720
    },
    {
      "epoch": 1.9658358292299103,
      "grad_norm": 0.3638029992580414,
      "learning_rate": 6.894427805133933e-06,
      "loss": 0.046,
      "step": 38725
    },
    {
      "epoch": 1.9660896492207727,
      "grad_norm": 0.2671540379524231,
      "learning_rate": 6.892735671861517e-06,
      "loss": 0.0388,
      "step": 38730
    },
    {
      "epoch": 1.966343469211635,
      "grad_norm": 0.2336130440235138,
      "learning_rate": 6.8910435385890995e-06,
      "loss": 0.0372,
      "step": 38735
    },
    {
      "epoch": 1.9665972892024977,
      "grad_norm": 0.3330386281013489,
      "learning_rate": 6.889351405316683e-06,
      "loss": 0.0325,
      "step": 38740
    },
    {
      "epoch": 1.96685110919336,
      "grad_norm": 1.5531961917877197,
      "learning_rate": 6.8876592720442666e-06,
      "loss": 0.0388,
      "step": 38745
    },
    {
      "epoch": 1.9671049291842224,
      "grad_norm": 0.4123472273349762,
      "learning_rate": 6.885967138771851e-06,
      "loss": 0.0439,
      "step": 38750
    },
    {
      "epoch": 1.967358749175085,
      "grad_norm": 0.3184575140476227,
      "learning_rate": 6.884275005499434e-06,
      "loss": 0.032,
      "step": 38755
    },
    {
      "epoch": 1.9676125691659476,
      "grad_norm": 0.2574605345726013,
      "learning_rate": 6.882582872227017e-06,
      "loss": 0.0315,
      "step": 38760
    },
    {
      "epoch": 1.96786638915681,
      "grad_norm": 0.3253490924835205,
      "learning_rate": 6.880890738954601e-06,
      "loss": 0.0389,
      "step": 38765
    },
    {
      "epoch": 1.9681202091476724,
      "grad_norm": 0.294781893491745,
      "learning_rate": 6.8791986056821834e-06,
      "loss": 0.032,
      "step": 38770
    },
    {
      "epoch": 1.968374029138535,
      "grad_norm": 0.29747629165649414,
      "learning_rate": 6.877506472409767e-06,
      "loss": 0.0307,
      "step": 38775
    },
    {
      "epoch": 1.9686278491293976,
      "grad_norm": 0.2784003019332886,
      "learning_rate": 6.875814339137351e-06,
      "loss": 0.0377,
      "step": 38780
    },
    {
      "epoch": 1.9688816691202597,
      "grad_norm": 0.47071167826652527,
      "learning_rate": 6.874122205864935e-06,
      "loss": 0.0408,
      "step": 38785
    },
    {
      "epoch": 1.9691354891111224,
      "grad_norm": 0.21550071239471436,
      "learning_rate": 6.872430072592518e-06,
      "loss": 0.0372,
      "step": 38790
    },
    {
      "epoch": 1.969389309101985,
      "grad_norm": 0.4224310517311096,
      "learning_rate": 6.870737939320101e-06,
      "loss": 0.0378,
      "step": 38795
    },
    {
      "epoch": 1.9696431290928473,
      "grad_norm": 0.3429742753505707,
      "learning_rate": 6.869045806047685e-06,
      "loss": 0.0348,
      "step": 38800
    },
    {
      "epoch": 1.9698969490837097,
      "grad_norm": 0.3539807200431824,
      "learning_rate": 6.867353672775269e-06,
      "loss": 0.0393,
      "step": 38805
    },
    {
      "epoch": 1.9701507690745723,
      "grad_norm": 0.281963050365448,
      "learning_rate": 6.865661539502852e-06,
      "loss": 0.0288,
      "step": 38810
    },
    {
      "epoch": 1.970404589065435,
      "grad_norm": 0.3246704936027527,
      "learning_rate": 6.863969406230435e-06,
      "loss": 0.0326,
      "step": 38815
    },
    {
      "epoch": 1.9706584090562973,
      "grad_norm": 0.3228529393672943,
      "learning_rate": 6.862277272958019e-06,
      "loss": 0.0442,
      "step": 38820
    },
    {
      "epoch": 1.9709122290471597,
      "grad_norm": 0.31963005661964417,
      "learning_rate": 6.8605851396856024e-06,
      "loss": 0.0307,
      "step": 38825
    },
    {
      "epoch": 1.9711660490380223,
      "grad_norm": 0.47388240694999695,
      "learning_rate": 6.858893006413185e-06,
      "loss": 0.0372,
      "step": 38830
    },
    {
      "epoch": 1.9714198690288847,
      "grad_norm": 0.7682002782821655,
      "learning_rate": 6.857200873140769e-06,
      "loss": 0.0369,
      "step": 38835
    },
    {
      "epoch": 1.971673689019747,
      "grad_norm": 0.23182091116905212,
      "learning_rate": 6.855508739868353e-06,
      "loss": 0.0292,
      "step": 38840
    },
    {
      "epoch": 1.9719275090106096,
      "grad_norm": 0.3109188377857208,
      "learning_rate": 6.853816606595936e-06,
      "loss": 0.0361,
      "step": 38845
    },
    {
      "epoch": 1.9721813290014722,
      "grad_norm": 0.661422848701477,
      "learning_rate": 6.852124473323519e-06,
      "loss": 0.0397,
      "step": 38850
    },
    {
      "epoch": 1.9724351489923346,
      "grad_norm": 0.2877862751483917,
      "learning_rate": 6.850432340051103e-06,
      "loss": 0.0346,
      "step": 38855
    },
    {
      "epoch": 1.972688968983197,
      "grad_norm": 0.5599194169044495,
      "learning_rate": 6.848740206778686e-06,
      "loss": 0.0361,
      "step": 38860
    },
    {
      "epoch": 1.9729427889740596,
      "grad_norm": 0.500221848487854,
      "learning_rate": 6.847048073506269e-06,
      "loss": 0.0355,
      "step": 38865
    },
    {
      "epoch": 1.9731966089649222,
      "grad_norm": 0.2395467460155487,
      "learning_rate": 6.8453559402338535e-06,
      "loss": 0.0283,
      "step": 38870
    },
    {
      "epoch": 1.9734504289557846,
      "grad_norm": 0.39513200521469116,
      "learning_rate": 6.843663806961437e-06,
      "loss": 0.033,
      "step": 38875
    },
    {
      "epoch": 1.973704248946647,
      "grad_norm": 0.2706584632396698,
      "learning_rate": 6.841971673689021e-06,
      "loss": 0.038,
      "step": 38880
    },
    {
      "epoch": 1.9739580689375096,
      "grad_norm": 0.38689085841178894,
      "learning_rate": 6.840279540416603e-06,
      "loss": 0.0322,
      "step": 38885
    },
    {
      "epoch": 1.974211888928372,
      "grad_norm": 0.25408896803855896,
      "learning_rate": 6.838587407144187e-06,
      "loss": 0.0393,
      "step": 38890
    },
    {
      "epoch": 1.9744657089192343,
      "grad_norm": 0.39390990138053894,
      "learning_rate": 6.836895273871771e-06,
      "loss": 0.0382,
      "step": 38895
    },
    {
      "epoch": 1.974719528910097,
      "grad_norm": 0.2625126540660858,
      "learning_rate": 6.835203140599355e-06,
      "loss": 0.0336,
      "step": 38900
    },
    {
      "epoch": 1.9749733489009595,
      "grad_norm": 0.3571978509426117,
      "learning_rate": 6.8335110073269375e-06,
      "loss": 0.0365,
      "step": 38905
    },
    {
      "epoch": 1.975227168891822,
      "grad_norm": 0.41177958250045776,
      "learning_rate": 6.831818874054521e-06,
      "loss": 0.0413,
      "step": 38910
    },
    {
      "epoch": 1.9754809888826843,
      "grad_norm": 0.24215035140514374,
      "learning_rate": 6.8301267407821045e-06,
      "loss": 0.0337,
      "step": 38915
    },
    {
      "epoch": 1.975734808873547,
      "grad_norm": 0.2806743085384369,
      "learning_rate": 6.828434607509689e-06,
      "loss": 0.0356,
      "step": 38920
    },
    {
      "epoch": 1.9759886288644095,
      "grad_norm": 0.26548781991004944,
      "learning_rate": 6.826742474237271e-06,
      "loss": 0.0306,
      "step": 38925
    },
    {
      "epoch": 1.976242448855272,
      "grad_norm": 0.3748215138912201,
      "learning_rate": 6.825050340964855e-06,
      "loss": 0.0314,
      "step": 38930
    },
    {
      "epoch": 1.9764962688461343,
      "grad_norm": 0.5329114198684692,
      "learning_rate": 6.823358207692439e-06,
      "loss": 0.0365,
      "step": 38935
    },
    {
      "epoch": 1.9767500888369969,
      "grad_norm": 0.33160844445228577,
      "learning_rate": 6.821666074420021e-06,
      "loss": 0.0403,
      "step": 38940
    },
    {
      "epoch": 1.9770039088278593,
      "grad_norm": 0.3684629201889038,
      "learning_rate": 6.819973941147605e-06,
      "loss": 0.0381,
      "step": 38945
    },
    {
      "epoch": 1.9772577288187216,
      "grad_norm": 0.6188549995422363,
      "learning_rate": 6.8182818078751885e-06,
      "loss": 0.042,
      "step": 38950
    },
    {
      "epoch": 1.9775115488095842,
      "grad_norm": 0.4174979627132416,
      "learning_rate": 6.816589674602773e-06,
      "loss": 0.0329,
      "step": 38955
    },
    {
      "epoch": 1.9777653688004468,
      "grad_norm": 0.4109318256378174,
      "learning_rate": 6.814897541330356e-06,
      "loss": 0.0368,
      "step": 38960
    },
    {
      "epoch": 1.9780191887913092,
      "grad_norm": 0.3277137577533722,
      "learning_rate": 6.813205408057939e-06,
      "loss": 0.0342,
      "step": 38965
    },
    {
      "epoch": 1.9782730087821716,
      "grad_norm": 0.42303919792175293,
      "learning_rate": 6.811513274785523e-06,
      "loss": 0.0453,
      "step": 38970
    },
    {
      "epoch": 1.9785268287730342,
      "grad_norm": 0.4994072914123535,
      "learning_rate": 6.809821141513106e-06,
      "loss": 0.0411,
      "step": 38975
    },
    {
      "epoch": 1.9787806487638968,
      "grad_norm": 0.6838673949241638,
      "learning_rate": 6.808129008240689e-06,
      "loss": 0.0398,
      "step": 38980
    },
    {
      "epoch": 1.979034468754759,
      "grad_norm": 0.25443944334983826,
      "learning_rate": 6.806436874968273e-06,
      "loss": 0.0348,
      "step": 38985
    },
    {
      "epoch": 1.9792882887456216,
      "grad_norm": 0.2758225202560425,
      "learning_rate": 6.804744741695857e-06,
      "loss": 0.0395,
      "step": 38990
    },
    {
      "epoch": 1.9795421087364842,
      "grad_norm": 0.2721156179904938,
      "learning_rate": 6.80305260842344e-06,
      "loss": 0.041,
      "step": 38995
    },
    {
      "epoch": 1.9797959287273466,
      "grad_norm": 0.27341485023498535,
      "learning_rate": 6.801360475151023e-06,
      "loss": 0.037,
      "step": 39000
    },
    {
      "epoch": 1.980049748718209,
      "grad_norm": 0.3476855456829071,
      "learning_rate": 6.799668341878607e-06,
      "loss": 0.0386,
      "step": 39005
    },
    {
      "epoch": 1.9803035687090715,
      "grad_norm": 0.2492963969707489,
      "learning_rate": 6.797976208606191e-06,
      "loss": 0.0339,
      "step": 39010
    },
    {
      "epoch": 1.9805573886999341,
      "grad_norm": 0.2879369854927063,
      "learning_rate": 6.796284075333774e-06,
      "loss": 0.0275,
      "step": 39015
    },
    {
      "epoch": 1.9808112086907965,
      "grad_norm": 0.32539618015289307,
      "learning_rate": 6.794591942061357e-06,
      "loss": 0.0372,
      "step": 39020
    },
    {
      "epoch": 1.981065028681659,
      "grad_norm": 0.35178670287132263,
      "learning_rate": 6.792899808788941e-06,
      "loss": 0.0351,
      "step": 39025
    },
    {
      "epoch": 1.9813188486725215,
      "grad_norm": 0.3978646993637085,
      "learning_rate": 6.791207675516524e-06,
      "loss": 0.0426,
      "step": 39030
    },
    {
      "epoch": 1.9815726686633839,
      "grad_norm": 0.3202974796295166,
      "learning_rate": 6.789515542244107e-06,
      "loss": 0.0365,
      "step": 39035
    },
    {
      "epoch": 1.9818264886542463,
      "grad_norm": 0.28949302434921265,
      "learning_rate": 6.787823408971691e-06,
      "loss": 0.0315,
      "step": 39040
    },
    {
      "epoch": 1.9820803086451089,
      "grad_norm": 0.5082337856292725,
      "learning_rate": 6.786131275699275e-06,
      "loss": 0.0332,
      "step": 39045
    },
    {
      "epoch": 1.9823341286359715,
      "grad_norm": 0.38315925002098083,
      "learning_rate": 6.7844391424268586e-06,
      "loss": 0.0389,
      "step": 39050
    },
    {
      "epoch": 1.9825879486268339,
      "grad_norm": 0.7824303507804871,
      "learning_rate": 6.782747009154441e-06,
      "loss": 0.0341,
      "step": 39055
    },
    {
      "epoch": 1.9828417686176962,
      "grad_norm": 0.3990626335144043,
      "learning_rate": 6.781054875882025e-06,
      "loss": 0.0457,
      "step": 39060
    },
    {
      "epoch": 1.9830955886085588,
      "grad_norm": 0.3116946816444397,
      "learning_rate": 6.779362742609608e-06,
      "loss": 0.0296,
      "step": 39065
    },
    {
      "epoch": 1.9833494085994214,
      "grad_norm": 0.30906444787979126,
      "learning_rate": 6.777670609337193e-06,
      "loss": 0.0322,
      "step": 39070
    },
    {
      "epoch": 1.9836032285902838,
      "grad_norm": 0.3945848345756531,
      "learning_rate": 6.7759784760647754e-06,
      "loss": 0.0448,
      "step": 39075
    },
    {
      "epoch": 1.9838570485811462,
      "grad_norm": 0.26128843426704407,
      "learning_rate": 6.774286342792359e-06,
      "loss": 0.0375,
      "step": 39080
    },
    {
      "epoch": 1.9841108685720088,
      "grad_norm": 0.32992011308670044,
      "learning_rate": 6.7725942095199425e-06,
      "loss": 0.0345,
      "step": 39085
    },
    {
      "epoch": 1.9843646885628712,
      "grad_norm": 0.2666621208190918,
      "learning_rate": 6.770902076247525e-06,
      "loss": 0.0339,
      "step": 39090
    },
    {
      "epoch": 1.9846185085537336,
      "grad_norm": 0.3576783835887909,
      "learning_rate": 6.769209942975109e-06,
      "loss": 0.0338,
      "step": 39095
    },
    {
      "epoch": 1.9848723285445962,
      "grad_norm": 0.3300950527191162,
      "learning_rate": 6.767517809702693e-06,
      "loss": 0.0317,
      "step": 39100
    },
    {
      "epoch": 1.9851261485354588,
      "grad_norm": 0.43691083788871765,
      "learning_rate": 6.765825676430277e-06,
      "loss": 0.0345,
      "step": 39105
    },
    {
      "epoch": 1.9853799685263211,
      "grad_norm": 0.2713020443916321,
      "learning_rate": 6.764133543157859e-06,
      "loss": 0.0293,
      "step": 39110
    },
    {
      "epoch": 1.9856337885171835,
      "grad_norm": 0.5607559680938721,
      "learning_rate": 6.762441409885443e-06,
      "loss": 0.0371,
      "step": 39115
    },
    {
      "epoch": 1.9858876085080461,
      "grad_norm": 1.484630823135376,
      "learning_rate": 6.7607492766130265e-06,
      "loss": 0.0378,
      "step": 39120
    },
    {
      "epoch": 1.9861414284989087,
      "grad_norm": 0.35799115896224976,
      "learning_rate": 6.759057143340611e-06,
      "loss": 0.0351,
      "step": 39125
    },
    {
      "epoch": 1.986395248489771,
      "grad_norm": 0.30156785249710083,
      "learning_rate": 6.757365010068193e-06,
      "loss": 0.0355,
      "step": 39130
    },
    {
      "epoch": 1.9866490684806335,
      "grad_norm": 0.44789522886276245,
      "learning_rate": 6.755672876795777e-06,
      "loss": 0.0395,
      "step": 39135
    },
    {
      "epoch": 1.986902888471496,
      "grad_norm": 0.5082187652587891,
      "learning_rate": 6.753980743523361e-06,
      "loss": 0.0389,
      "step": 39140
    },
    {
      "epoch": 1.9871567084623585,
      "grad_norm": 0.4441748559474945,
      "learning_rate": 6.752288610250944e-06,
      "loss": 0.031,
      "step": 39145
    },
    {
      "epoch": 1.9874105284532209,
      "grad_norm": 0.2157219499349594,
      "learning_rate": 6.750596476978527e-06,
      "loss": 0.0335,
      "step": 39150
    },
    {
      "epoch": 1.9876643484440835,
      "grad_norm": 0.4951675534248352,
      "learning_rate": 6.7489043437061104e-06,
      "loss": 0.0376,
      "step": 39155
    },
    {
      "epoch": 1.987918168434946,
      "grad_norm": 0.3324943482875824,
      "learning_rate": 6.747212210433695e-06,
      "loss": 0.0302,
      "step": 39160
    },
    {
      "epoch": 1.9881719884258084,
      "grad_norm": 0.32476142048835754,
      "learning_rate": 6.745520077161278e-06,
      "loss": 0.0422,
      "step": 39165
    },
    {
      "epoch": 1.9884258084166708,
      "grad_norm": 0.37469473481178284,
      "learning_rate": 6.743827943888861e-06,
      "loss": 0.0432,
      "step": 39170
    },
    {
      "epoch": 1.9886796284075334,
      "grad_norm": 0.672195553779602,
      "learning_rate": 6.742135810616445e-06,
      "loss": 0.0331,
      "step": 39175
    },
    {
      "epoch": 1.9889334483983958,
      "grad_norm": 0.3643190562725067,
      "learning_rate": 6.740443677344028e-06,
      "loss": 0.0344,
      "step": 39180
    },
    {
      "epoch": 1.9891872683892582,
      "grad_norm": 0.2701443135738373,
      "learning_rate": 6.738751544071611e-06,
      "loss": 0.0401,
      "step": 39185
    },
    {
      "epoch": 1.9894410883801208,
      "grad_norm": 0.4211021661758423,
      "learning_rate": 6.737059410799195e-06,
      "loss": 0.0467,
      "step": 39190
    },
    {
      "epoch": 1.9896949083709834,
      "grad_norm": 0.3575497567653656,
      "learning_rate": 6.735367277526779e-06,
      "loss": 0.0416,
      "step": 39195
    },
    {
      "epoch": 1.9899487283618458,
      "grad_norm": 0.43901878595352173,
      "learning_rate": 6.733675144254362e-06,
      "loss": 0.0411,
      "step": 39200
    },
    {
      "epoch": 1.9902025483527082,
      "grad_norm": 0.29841357469558716,
      "learning_rate": 6.731983010981945e-06,
      "loss": 0.04,
      "step": 39205
    },
    {
      "epoch": 1.9904563683435708,
      "grad_norm": 0.4036899507045746,
      "learning_rate": 6.730290877709529e-06,
      "loss": 0.0352,
      "step": 39210
    },
    {
      "epoch": 1.9907101883344334,
      "grad_norm": 0.3392904996871948,
      "learning_rate": 6.728598744437113e-06,
      "loss": 0.0313,
      "step": 39215
    },
    {
      "epoch": 1.9909640083252957,
      "grad_norm": 0.3535223603248596,
      "learning_rate": 6.7269066111646965e-06,
      "loss": 0.038,
      "step": 39220
    },
    {
      "epoch": 1.9912178283161581,
      "grad_norm": 0.5300154089927673,
      "learning_rate": 6.725214477892279e-06,
      "loss": 0.0427,
      "step": 39225
    },
    {
      "epoch": 1.9914716483070207,
      "grad_norm": 0.2561807632446289,
      "learning_rate": 6.723522344619863e-06,
      "loss": 0.0279,
      "step": 39230
    },
    {
      "epoch": 1.991725468297883,
      "grad_norm": 0.3881128132343292,
      "learning_rate": 6.721830211347446e-06,
      "loss": 0.0444,
      "step": 39235
    },
    {
      "epoch": 1.9919792882887455,
      "grad_norm": 0.340969979763031,
      "learning_rate": 6.72013807807503e-06,
      "loss": 0.0374,
      "step": 39240
    },
    {
      "epoch": 1.992233108279608,
      "grad_norm": 0.3254038989543915,
      "learning_rate": 6.7184459448026126e-06,
      "loss": 0.0313,
      "step": 39245
    },
    {
      "epoch": 1.9924869282704707,
      "grad_norm": 0.5408515930175781,
      "learning_rate": 6.716753811530197e-06,
      "loss": 0.038,
      "step": 39250
    },
    {
      "epoch": 1.992740748261333,
      "grad_norm": 0.2633402347564697,
      "learning_rate": 6.7150616782577805e-06,
      "loss": 0.0387,
      "step": 39255
    },
    {
      "epoch": 1.9929945682521955,
      "grad_norm": 0.39038726687431335,
      "learning_rate": 6.713369544985363e-06,
      "loss": 0.0333,
      "step": 39260
    },
    {
      "epoch": 1.993248388243058,
      "grad_norm": 0.46379637718200684,
      "learning_rate": 6.711677411712947e-06,
      "loss": 0.0309,
      "step": 39265
    },
    {
      "epoch": 1.9935022082339207,
      "grad_norm": 0.44145670533180237,
      "learning_rate": 6.70998527844053e-06,
      "loss": 0.0413,
      "step": 39270
    },
    {
      "epoch": 1.993756028224783,
      "grad_norm": 0.4047604501247406,
      "learning_rate": 6.708293145168115e-06,
      "loss": 0.0428,
      "step": 39275
    },
    {
      "epoch": 1.9940098482156454,
      "grad_norm": 0.4288921356201172,
      "learning_rate": 6.706601011895697e-06,
      "loss": 0.0391,
      "step": 39280
    },
    {
      "epoch": 1.994263668206508,
      "grad_norm": 0.3530460298061371,
      "learning_rate": 6.704908878623281e-06,
      "loss": 0.0404,
      "step": 39285
    },
    {
      "epoch": 1.9945174881973704,
      "grad_norm": 1.06656014919281,
      "learning_rate": 6.7032167453508645e-06,
      "loss": 0.0349,
      "step": 39290
    },
    {
      "epoch": 1.9947713081882328,
      "grad_norm": 0.32023099064826965,
      "learning_rate": 6.701524612078448e-06,
      "loss": 0.0314,
      "step": 39295
    },
    {
      "epoch": 1.9950251281790954,
      "grad_norm": 0.32165810465812683,
      "learning_rate": 6.699832478806031e-06,
      "loss": 0.0406,
      "step": 39300
    },
    {
      "epoch": 1.995278948169958,
      "grad_norm": 0.30110183358192444,
      "learning_rate": 6.698140345533615e-06,
      "loss": 0.0301,
      "step": 39305
    },
    {
      "epoch": 1.9955327681608204,
      "grad_norm": 0.2650599777698517,
      "learning_rate": 6.696448212261199e-06,
      "loss": 0.0392,
      "step": 39310
    },
    {
      "epoch": 1.9957865881516827,
      "grad_norm": 0.6036587953567505,
      "learning_rate": 6.694756078988782e-06,
      "loss": 0.0375,
      "step": 39315
    },
    {
      "epoch": 1.9960404081425454,
      "grad_norm": 1.74124276638031,
      "learning_rate": 6.693063945716365e-06,
      "loss": 0.041,
      "step": 39320
    },
    {
      "epoch": 1.9962942281334077,
      "grad_norm": 0.357609361410141,
      "learning_rate": 6.691371812443948e-06,
      "loss": 0.0394,
      "step": 39325
    },
    {
      "epoch": 1.9965480481242701,
      "grad_norm": 0.271170049905777,
      "learning_rate": 6.689679679171532e-06,
      "loss": 0.0385,
      "step": 39330
    },
    {
      "epoch": 1.9968018681151327,
      "grad_norm": 0.3494093418121338,
      "learning_rate": 6.687987545899115e-06,
      "loss": 0.045,
      "step": 39335
    },
    {
      "epoch": 1.9970556881059953,
      "grad_norm": 0.2796786427497864,
      "learning_rate": 6.686295412626699e-06,
      "loss": 0.0405,
      "step": 39340
    },
    {
      "epoch": 1.9973095080968577,
      "grad_norm": 0.26732075214385986,
      "learning_rate": 6.684603279354283e-06,
      "loss": 0.0417,
      "step": 39345
    },
    {
      "epoch": 1.99756332808772,
      "grad_norm": 0.22764670848846436,
      "learning_rate": 6.682911146081866e-06,
      "loss": 0.0277,
      "step": 39350
    },
    {
      "epoch": 1.9978171480785827,
      "grad_norm": 0.28476661443710327,
      "learning_rate": 6.681219012809449e-06,
      "loss": 0.0318,
      "step": 39355
    },
    {
      "epoch": 1.9980709680694453,
      "grad_norm": 0.21495118737220764,
      "learning_rate": 6.679526879537032e-06,
      "loss": 0.0361,
      "step": 39360
    },
    {
      "epoch": 1.9983247880603077,
      "grad_norm": 0.2934248149394989,
      "learning_rate": 6.677834746264617e-06,
      "loss": 0.0395,
      "step": 39365
    },
    {
      "epoch": 1.99857860805117,
      "grad_norm": 0.4350336790084839,
      "learning_rate": 6.6761426129922e-06,
      "loss": 0.0374,
      "step": 39370
    },
    {
      "epoch": 1.9988324280420326,
      "grad_norm": 0.31187477707862854,
      "learning_rate": 6.674450479719783e-06,
      "loss": 0.033,
      "step": 39375
    },
    {
      "epoch": 1.999086248032895,
      "grad_norm": 0.3732047379016876,
      "learning_rate": 6.6727583464473666e-06,
      "loss": 0.036,
      "step": 39380
    },
    {
      "epoch": 1.9993400680237574,
      "grad_norm": 0.24284490942955017,
      "learning_rate": 6.67106621317495e-06,
      "loss": 0.0316,
      "step": 39385
    },
    {
      "epoch": 1.99959388801462,
      "grad_norm": 0.3445446789264679,
      "learning_rate": 6.6693740799025345e-06,
      "loss": 0.0391,
      "step": 39390
    },
    {
      "epoch": 1.9998477080054826,
      "grad_norm": 0.35594990849494934,
      "learning_rate": 6.667681946630117e-06,
      "loss": 0.0401,
      "step": 39395
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.18550293147563934,
      "eval_runtime": 1774.6443,
      "eval_samples_per_second": 70.437,
      "eval_steps_per_second": 2.202,
      "step": 39398
    },
    {
      "epoch": 2.0001015279963448,
      "grad_norm": 0.48830828070640564,
      "learning_rate": 6.665989813357701e-06,
      "loss": 0.0421,
      "step": 39400
    },
    {
      "epoch": 2.0003553479872074,
      "grad_norm": 0.43993479013442993,
      "learning_rate": 6.664297680085284e-06,
      "loss": 0.0445,
      "step": 39405
    },
    {
      "epoch": 2.00060916797807,
      "grad_norm": 0.26866963505744934,
      "learning_rate": 6.662605546812867e-06,
      "loss": 0.0356,
      "step": 39410
    },
    {
      "epoch": 2.0008629879689326,
      "grad_norm": 0.356734961271286,
      "learning_rate": 6.6609134135404505e-06,
      "loss": 0.0332,
      "step": 39415
    },
    {
      "epoch": 2.0011168079597947,
      "grad_norm": 0.34094324707984924,
      "learning_rate": 6.659221280268034e-06,
      "loss": 0.0433,
      "step": 39420
    },
    {
      "epoch": 2.0013706279506573,
      "grad_norm": 0.5367624759674072,
      "learning_rate": 6.6575291469956185e-06,
      "loss": 0.0418,
      "step": 39425
    },
    {
      "epoch": 2.00162444794152,
      "grad_norm": 0.2931801676750183,
      "learning_rate": 6.655837013723201e-06,
      "loss": 0.041,
      "step": 39430
    },
    {
      "epoch": 2.0018782679323825,
      "grad_norm": 0.39484718441963196,
      "learning_rate": 6.654144880450785e-06,
      "loss": 0.0359,
      "step": 39435
    },
    {
      "epoch": 2.0021320879232447,
      "grad_norm": 0.5829231142997742,
      "learning_rate": 6.652452747178368e-06,
      "loss": 0.0352,
      "step": 39440
    },
    {
      "epoch": 2.0023859079141073,
      "grad_norm": 0.5522474050521851,
      "learning_rate": 6.650760613905952e-06,
      "loss": 0.0503,
      "step": 39445
    },
    {
      "epoch": 2.00263972790497,
      "grad_norm": 0.5528905987739563,
      "learning_rate": 6.6490684806335345e-06,
      "loss": 0.0322,
      "step": 39450
    },
    {
      "epoch": 2.002893547895832,
      "grad_norm": 0.6939111948013306,
      "learning_rate": 6.647376347361119e-06,
      "loss": 0.0387,
      "step": 39455
    },
    {
      "epoch": 2.0031473678866947,
      "grad_norm": 0.3555048108100891,
      "learning_rate": 6.6456842140887024e-06,
      "loss": 0.0384,
      "step": 39460
    },
    {
      "epoch": 2.0034011878775573,
      "grad_norm": 0.2942870855331421,
      "learning_rate": 6.643992080816286e-06,
      "loss": 0.0411,
      "step": 39465
    },
    {
      "epoch": 2.00365500786842,
      "grad_norm": 0.5334384441375732,
      "learning_rate": 6.642299947543869e-06,
      "loss": 0.0389,
      "step": 39470
    },
    {
      "epoch": 2.003908827859282,
      "grad_norm": 0.4089035987854004,
      "learning_rate": 6.640607814271452e-06,
      "loss": 0.0327,
      "step": 39475
    },
    {
      "epoch": 2.0041626478501446,
      "grad_norm": 0.6318344473838806,
      "learning_rate": 6.638915680999037e-06,
      "loss": 0.0398,
      "step": 39480
    },
    {
      "epoch": 2.0044164678410072,
      "grad_norm": 0.34425875544548035,
      "learning_rate": 6.63722354772662e-06,
      "loss": 0.0421,
      "step": 39485
    },
    {
      "epoch": 2.00467028783187,
      "grad_norm": 0.2746249735355377,
      "learning_rate": 6.635531414454203e-06,
      "loss": 0.029,
      "step": 39490
    },
    {
      "epoch": 2.004924107822732,
      "grad_norm": 0.41694292426109314,
      "learning_rate": 6.633839281181786e-06,
      "loss": 0.0364,
      "step": 39495
    },
    {
      "epoch": 2.0051779278135946,
      "grad_norm": 0.43403369188308716,
      "learning_rate": 6.63214714790937e-06,
      "loss": 0.0419,
      "step": 39500
    },
    {
      "epoch": 2.005431747804457,
      "grad_norm": 0.3920260965824127,
      "learning_rate": 6.630455014636953e-06,
      "loss": 0.0372,
      "step": 39505
    },
    {
      "epoch": 2.0056855677953194,
      "grad_norm": 0.35140088200569153,
      "learning_rate": 6.628762881364537e-06,
      "loss": 0.0368,
      "step": 39510
    },
    {
      "epoch": 2.005939387786182,
      "grad_norm": 0.5638235211372375,
      "learning_rate": 6.6270707480921206e-06,
      "loss": 0.031,
      "step": 39515
    },
    {
      "epoch": 2.0061932077770446,
      "grad_norm": 0.3450993001461029,
      "learning_rate": 6.625378614819704e-06,
      "loss": 0.0389,
      "step": 39520
    },
    {
      "epoch": 2.006447027767907,
      "grad_norm": 0.23240645229816437,
      "learning_rate": 6.623686481547287e-06,
      "loss": 0.0317,
      "step": 39525
    },
    {
      "epoch": 2.0067008477587693,
      "grad_norm": 0.2569294571876526,
      "learning_rate": 6.62199434827487e-06,
      "loss": 0.0352,
      "step": 39530
    },
    {
      "epoch": 2.006954667749632,
      "grad_norm": 0.7697589993476868,
      "learning_rate": 6.620302215002454e-06,
      "loss": 0.032,
      "step": 39535
    },
    {
      "epoch": 2.0072084877404945,
      "grad_norm": 0.39913979172706604,
      "learning_rate": 6.618610081730038e-06,
      "loss": 0.0296,
      "step": 39540
    },
    {
      "epoch": 2.007462307731357,
      "grad_norm": 0.4698142111301422,
      "learning_rate": 6.616917948457621e-06,
      "loss": 0.0382,
      "step": 39545
    },
    {
      "epoch": 2.0077161277222193,
      "grad_norm": 0.30996814370155334,
      "learning_rate": 6.6152258151852045e-06,
      "loss": 0.0349,
      "step": 39550
    },
    {
      "epoch": 2.007969947713082,
      "grad_norm": 0.3366950452327728,
      "learning_rate": 6.613533681912788e-06,
      "loss": 0.0393,
      "step": 39555
    },
    {
      "epoch": 2.0082237677039445,
      "grad_norm": 0.24016176164150238,
      "learning_rate": 6.611841548640372e-06,
      "loss": 0.0415,
      "step": 39560
    },
    {
      "epoch": 2.0084775876948067,
      "grad_norm": 0.669681191444397,
      "learning_rate": 6.610149415367954e-06,
      "loss": 0.0363,
      "step": 39565
    },
    {
      "epoch": 2.0087314076856693,
      "grad_norm": 0.22296953201293945,
      "learning_rate": 6.608457282095539e-06,
      "loss": 0.0356,
      "step": 39570
    },
    {
      "epoch": 2.008985227676532,
      "grad_norm": 0.5446509718894958,
      "learning_rate": 6.606765148823122e-06,
      "loss": 0.0316,
      "step": 39575
    },
    {
      "epoch": 2.0092390476673945,
      "grad_norm": 0.340939462184906,
      "learning_rate": 6.605073015550705e-06,
      "loss": 0.0334,
      "step": 39580
    },
    {
      "epoch": 2.0094928676582566,
      "grad_norm": 0.42687132954597473,
      "learning_rate": 6.6033808822782885e-06,
      "loss": 0.0356,
      "step": 39585
    },
    {
      "epoch": 2.0097466876491192,
      "grad_norm": 0.3017434775829315,
      "learning_rate": 6.601688749005872e-06,
      "loss": 0.039,
      "step": 39590
    },
    {
      "epoch": 2.010000507639982,
      "grad_norm": 0.24050408601760864,
      "learning_rate": 6.5999966157334564e-06,
      "loss": 0.0295,
      "step": 39595
    },
    {
      "epoch": 2.010254327630844,
      "grad_norm": 0.3019162714481354,
      "learning_rate": 6.598304482461039e-06,
      "loss": 0.0322,
      "step": 39600
    },
    {
      "epoch": 2.0105081476217066,
      "grad_norm": 0.38189876079559326,
      "learning_rate": 6.596612349188623e-06,
      "loss": 0.0326,
      "step": 39605
    },
    {
      "epoch": 2.010761967612569,
      "grad_norm": 0.5544765591621399,
      "learning_rate": 6.594920215916206e-06,
      "loss": 0.0421,
      "step": 39610
    },
    {
      "epoch": 2.011015787603432,
      "grad_norm": 0.5120232105255127,
      "learning_rate": 6.59322808264379e-06,
      "loss": 0.0424,
      "step": 39615
    },
    {
      "epoch": 2.011269607594294,
      "grad_norm": 0.29444101452827454,
      "learning_rate": 6.5915359493713725e-06,
      "loss": 0.0311,
      "step": 39620
    },
    {
      "epoch": 2.0115234275851566,
      "grad_norm": 0.24789972603321075,
      "learning_rate": 6.589843816098956e-06,
      "loss": 0.0338,
      "step": 39625
    },
    {
      "epoch": 2.011777247576019,
      "grad_norm": 0.2729002833366394,
      "learning_rate": 6.58815168282654e-06,
      "loss": 0.0328,
      "step": 39630
    },
    {
      "epoch": 2.0120310675668818,
      "grad_norm": 0.4501836895942688,
      "learning_rate": 6.586459549554124e-06,
      "loss": 0.0378,
      "step": 39635
    },
    {
      "epoch": 2.012284887557744,
      "grad_norm": 0.2863801419734955,
      "learning_rate": 6.584767416281707e-06,
      "loss": 0.0332,
      "step": 39640
    },
    {
      "epoch": 2.0125387075486065,
      "grad_norm": 0.2944399118423462,
      "learning_rate": 6.58307528300929e-06,
      "loss": 0.0374,
      "step": 39645
    },
    {
      "epoch": 2.012792527539469,
      "grad_norm": 0.8535482883453369,
      "learning_rate": 6.581383149736874e-06,
      "loss": 0.0419,
      "step": 39650
    },
    {
      "epoch": 2.0130463475303313,
      "grad_norm": 1.3947867155075073,
      "learning_rate": 6.5796910164644564e-06,
      "loss": 0.0362,
      "step": 39655
    },
    {
      "epoch": 2.013300167521194,
      "grad_norm": 0.2795107066631317,
      "learning_rate": 6.577998883192041e-06,
      "loss": 0.0351,
      "step": 39660
    },
    {
      "epoch": 2.0135539875120565,
      "grad_norm": 0.2812153398990631,
      "learning_rate": 6.576306749919624e-06,
      "loss": 0.0345,
      "step": 39665
    },
    {
      "epoch": 2.013807807502919,
      "grad_norm": 0.4150792062282562,
      "learning_rate": 6.574614616647208e-06,
      "loss": 0.0368,
      "step": 39670
    },
    {
      "epoch": 2.0140616274937813,
      "grad_norm": 0.34947192668914795,
      "learning_rate": 6.572922483374791e-06,
      "loss": 0.0341,
      "step": 39675
    },
    {
      "epoch": 2.014315447484644,
      "grad_norm": 0.28611230850219727,
      "learning_rate": 6.571230350102374e-06,
      "loss": 0.0315,
      "step": 39680
    },
    {
      "epoch": 2.0145692674755065,
      "grad_norm": 0.4013167917728424,
      "learning_rate": 6.5695382168299585e-06,
      "loss": 0.0336,
      "step": 39685
    },
    {
      "epoch": 2.014823087466369,
      "grad_norm": 0.2976538836956024,
      "learning_rate": 6.567846083557542e-06,
      "loss": 0.0383,
      "step": 39690
    },
    {
      "epoch": 2.0150769074572312,
      "grad_norm": 0.2748160660266876,
      "learning_rate": 6.566153950285125e-06,
      "loss": 0.0379,
      "step": 39695
    },
    {
      "epoch": 2.015330727448094,
      "grad_norm": 0.244331955909729,
      "learning_rate": 6.564461817012708e-06,
      "loss": 0.0316,
      "step": 39700
    },
    {
      "epoch": 2.0155845474389564,
      "grad_norm": 0.8902464509010315,
      "learning_rate": 6.562769683740292e-06,
      "loss": 0.0406,
      "step": 39705
    },
    {
      "epoch": 2.0158383674298186,
      "grad_norm": 0.3426436185836792,
      "learning_rate": 6.561077550467876e-06,
      "loss": 0.0302,
      "step": 39710
    },
    {
      "epoch": 2.016092187420681,
      "grad_norm": 0.3238757848739624,
      "learning_rate": 6.559385417195458e-06,
      "loss": 0.039,
      "step": 39715
    },
    {
      "epoch": 2.016346007411544,
      "grad_norm": 0.34030091762542725,
      "learning_rate": 6.5576932839230425e-06,
      "loss": 0.0382,
      "step": 39720
    },
    {
      "epoch": 2.0165998274024064,
      "grad_norm": 0.2633964717388153,
      "learning_rate": 6.556001150650626e-06,
      "loss": 0.0344,
      "step": 39725
    },
    {
      "epoch": 2.0168536473932686,
      "grad_norm": 0.23452314734458923,
      "learning_rate": 6.55430901737821e-06,
      "loss": 0.0405,
      "step": 39730
    },
    {
      "epoch": 2.017107467384131,
      "grad_norm": 0.34209662675857544,
      "learning_rate": 6.552616884105792e-06,
      "loss": 0.0288,
      "step": 39735
    },
    {
      "epoch": 2.0173612873749938,
      "grad_norm": 0.23917095363140106,
      "learning_rate": 6.550924750833376e-06,
      "loss": 0.0326,
      "step": 39740
    },
    {
      "epoch": 2.017615107365856,
      "grad_norm": 0.3735106289386749,
      "learning_rate": 6.54923261756096e-06,
      "loss": 0.0323,
      "step": 39745
    },
    {
      "epoch": 2.0178689273567185,
      "grad_norm": 0.517549455165863,
      "learning_rate": 6.547540484288543e-06,
      "loss": 0.0403,
      "step": 39750
    },
    {
      "epoch": 2.018122747347581,
      "grad_norm": 0.3461301922798157,
      "learning_rate": 6.5458483510161265e-06,
      "loss": 0.0271,
      "step": 39755
    },
    {
      "epoch": 2.0183765673384437,
      "grad_norm": 0.3491016626358032,
      "learning_rate": 6.54415621774371e-06,
      "loss": 0.0375,
      "step": 39760
    },
    {
      "epoch": 2.018630387329306,
      "grad_norm": 0.5119760036468506,
      "learning_rate": 6.5424640844712936e-06,
      "loss": 0.0357,
      "step": 39765
    },
    {
      "epoch": 2.0188842073201685,
      "grad_norm": 0.5701979398727417,
      "learning_rate": 6.540771951198876e-06,
      "loss": 0.0384,
      "step": 39770
    },
    {
      "epoch": 2.019138027311031,
      "grad_norm": 0.24367502331733704,
      "learning_rate": 6.539079817926461e-06,
      "loss": 0.0379,
      "step": 39775
    },
    {
      "epoch": 2.0193918473018937,
      "grad_norm": 0.3051294684410095,
      "learning_rate": 6.537387684654044e-06,
      "loss": 0.0353,
      "step": 39780
    },
    {
      "epoch": 2.019645667292756,
      "grad_norm": 0.37204432487487793,
      "learning_rate": 6.535695551381628e-06,
      "loss": 0.0352,
      "step": 39785
    },
    {
      "epoch": 2.0198994872836185,
      "grad_norm": 0.4559103846549988,
      "learning_rate": 6.5340034181092104e-06,
      "loss": 0.0304,
      "step": 39790
    },
    {
      "epoch": 2.020153307274481,
      "grad_norm": 0.37972748279571533,
      "learning_rate": 6.532311284836794e-06,
      "loss": 0.0267,
      "step": 39795
    },
    {
      "epoch": 2.020407127265343,
      "grad_norm": 0.7116328477859497,
      "learning_rate": 6.530619151564378e-06,
      "loss": 0.0325,
      "step": 39800
    },
    {
      "epoch": 2.020660947256206,
      "grad_norm": 0.3754270076751709,
      "learning_rate": 6.528927018291962e-06,
      "loss": 0.0358,
      "step": 39805
    },
    {
      "epoch": 2.0209147672470684,
      "grad_norm": 0.26524120569229126,
      "learning_rate": 6.527234885019545e-06,
      "loss": 0.0273,
      "step": 39810
    },
    {
      "epoch": 2.021168587237931,
      "grad_norm": 0.3698604106903076,
      "learning_rate": 6.525542751747128e-06,
      "loss": 0.0376,
      "step": 39815
    },
    {
      "epoch": 2.021422407228793,
      "grad_norm": 0.3561539947986603,
      "learning_rate": 6.523850618474712e-06,
      "loss": 0.0386,
      "step": 39820
    },
    {
      "epoch": 2.021676227219656,
      "grad_norm": 0.40872108936309814,
      "learning_rate": 6.522158485202294e-06,
      "loss": 0.0295,
      "step": 39825
    },
    {
      "epoch": 2.0219300472105184,
      "grad_norm": 0.3250705301761627,
      "learning_rate": 6.520466351929878e-06,
      "loss": 0.0379,
      "step": 39830
    },
    {
      "epoch": 2.022183867201381,
      "grad_norm": 0.2779284417629242,
      "learning_rate": 6.518774218657462e-06,
      "loss": 0.0317,
      "step": 39835
    },
    {
      "epoch": 2.022437687192243,
      "grad_norm": 0.4617094397544861,
      "learning_rate": 6.517082085385046e-06,
      "loss": 0.0326,
      "step": 39840
    },
    {
      "epoch": 2.0226915071831058,
      "grad_norm": 0.4745062291622162,
      "learning_rate": 6.515389952112629e-06,
      "loss": 0.0431,
      "step": 39845
    },
    {
      "epoch": 2.0229453271739684,
      "grad_norm": 0.6929597854614258,
      "learning_rate": 6.513697818840212e-06,
      "loss": 0.0315,
      "step": 39850
    },
    {
      "epoch": 2.0231991471648305,
      "grad_norm": 0.23170751333236694,
      "learning_rate": 6.512005685567796e-06,
      "loss": 0.0319,
      "step": 39855
    },
    {
      "epoch": 2.023452967155693,
      "grad_norm": 0.5685155391693115,
      "learning_rate": 6.51031355229538e-06,
      "loss": 0.0329,
      "step": 39860
    },
    {
      "epoch": 2.0237067871465557,
      "grad_norm": 0.2968323230743408,
      "learning_rate": 6.508621419022963e-06,
      "loss": 0.0354,
      "step": 39865
    },
    {
      "epoch": 2.0239606071374183,
      "grad_norm": 1.593350887298584,
      "learning_rate": 6.506929285750546e-06,
      "loss": 0.0368,
      "step": 39870
    },
    {
      "epoch": 2.0242144271282805,
      "grad_norm": 0.41603031754493713,
      "learning_rate": 6.50523715247813e-06,
      "loss": 0.0348,
      "step": 39875
    },
    {
      "epoch": 2.024468247119143,
      "grad_norm": 0.2845626473426819,
      "learning_rate": 6.503545019205713e-06,
      "loss": 0.0357,
      "step": 39880
    },
    {
      "epoch": 2.0247220671100057,
      "grad_norm": 0.4017883539199829,
      "learning_rate": 6.501852885933296e-06,
      "loss": 0.0423,
      "step": 39885
    },
    {
      "epoch": 2.024975887100868,
      "grad_norm": 0.23194077610969543,
      "learning_rate": 6.5001607526608805e-06,
      "loss": 0.0339,
      "step": 39890
    },
    {
      "epoch": 2.0252297070917304,
      "grad_norm": 0.629647970199585,
      "learning_rate": 6.498468619388464e-06,
      "loss": 0.0371,
      "step": 39895
    },
    {
      "epoch": 2.025483527082593,
      "grad_norm": 0.2321619689464569,
      "learning_rate": 6.496776486116047e-06,
      "loss": 0.0325,
      "step": 39900
    },
    {
      "epoch": 2.0257373470734557,
      "grad_norm": 0.2681446075439453,
      "learning_rate": 6.49508435284363e-06,
      "loss": 0.031,
      "step": 39905
    },
    {
      "epoch": 2.025991167064318,
      "grad_norm": 0.2765409052371979,
      "learning_rate": 6.493392219571214e-06,
      "loss": 0.0347,
      "step": 39910
    },
    {
      "epoch": 2.0262449870551804,
      "grad_norm": 0.2587858736515045,
      "learning_rate": 6.491700086298797e-06,
      "loss": 0.0363,
      "step": 39915
    },
    {
      "epoch": 2.026498807046043,
      "grad_norm": 0.39432743191719055,
      "learning_rate": 6.49000795302638e-06,
      "loss": 0.0331,
      "step": 39920
    },
    {
      "epoch": 2.0267526270369056,
      "grad_norm": 0.2792506217956543,
      "learning_rate": 6.4883158197539644e-06,
      "loss": 0.0332,
      "step": 39925
    },
    {
      "epoch": 2.0270064470277678,
      "grad_norm": 0.26611894369125366,
      "learning_rate": 6.486623686481548e-06,
      "loss": 0.0363,
      "step": 39930
    },
    {
      "epoch": 2.0272602670186304,
      "grad_norm": 0.34571000933647156,
      "learning_rate": 6.4849315532091315e-06,
      "loss": 0.0365,
      "step": 39935
    },
    {
      "epoch": 2.027514087009493,
      "grad_norm": 0.23014172911643982,
      "learning_rate": 6.483239419936714e-06,
      "loss": 0.0393,
      "step": 39940
    },
    {
      "epoch": 2.027767907000355,
      "grad_norm": 0.26579776406288147,
      "learning_rate": 6.481547286664298e-06,
      "loss": 0.0323,
      "step": 39945
    },
    {
      "epoch": 2.0280217269912177,
      "grad_norm": 0.44567152857780457,
      "learning_rate": 6.479855153391882e-06,
      "loss": 0.0447,
      "step": 39950
    },
    {
      "epoch": 2.0282755469820803,
      "grad_norm": 0.2629788815975189,
      "learning_rate": 6.478163020119466e-06,
      "loss": 0.0357,
      "step": 39955
    },
    {
      "epoch": 2.028529366972943,
      "grad_norm": 0.40638864040374756,
      "learning_rate": 6.476470886847048e-06,
      "loss": 0.0291,
      "step": 39960
    },
    {
      "epoch": 2.028783186963805,
      "grad_norm": 0.36897462606430054,
      "learning_rate": 6.474778753574632e-06,
      "loss": 0.0367,
      "step": 39965
    },
    {
      "epoch": 2.0290370069546677,
      "grad_norm": 0.468357115983963,
      "learning_rate": 6.4730866203022155e-06,
      "loss": 0.032,
      "step": 39970
    },
    {
      "epoch": 2.0292908269455303,
      "grad_norm": 0.37311598658561707,
      "learning_rate": 6.471394487029798e-06,
      "loss": 0.0422,
      "step": 39975
    },
    {
      "epoch": 2.029544646936393,
      "grad_norm": 0.3664664030075073,
      "learning_rate": 6.469702353757383e-06,
      "loss": 0.0388,
      "step": 39980
    },
    {
      "epoch": 2.029798466927255,
      "grad_norm": 0.28249913454055786,
      "learning_rate": 6.468010220484966e-06,
      "loss": 0.0364,
      "step": 39985
    },
    {
      "epoch": 2.0300522869181177,
      "grad_norm": 0.30894264578819275,
      "learning_rate": 6.46631808721255e-06,
      "loss": 0.035,
      "step": 39990
    },
    {
      "epoch": 2.0303061069089803,
      "grad_norm": 0.27944618463516235,
      "learning_rate": 6.464625953940132e-06,
      "loss": 0.0374,
      "step": 39995
    },
    {
      "epoch": 2.0305599268998424,
      "grad_norm": 0.3840200901031494,
      "learning_rate": 6.462933820667716e-06,
      "loss": 0.0393,
      "step": 40000
    },
    {
      "epoch": 2.030813746890705,
      "grad_norm": 0.3649371862411499,
      "learning_rate": 6.4612416873952995e-06,
      "loss": 0.0348,
      "step": 40005
    },
    {
      "epoch": 2.0310675668815676,
      "grad_norm": 0.2791478633880615,
      "learning_rate": 6.459549554122884e-06,
      "loss": 0.0452,
      "step": 40010
    },
    {
      "epoch": 2.0313213868724302,
      "grad_norm": 0.28353622555732727,
      "learning_rate": 6.4578574208504666e-06,
      "loss": 0.0345,
      "step": 40015
    },
    {
      "epoch": 2.0315752068632924,
      "grad_norm": 0.36117586493492126,
      "learning_rate": 6.45616528757805e-06,
      "loss": 0.0396,
      "step": 40020
    },
    {
      "epoch": 2.031829026854155,
      "grad_norm": 0.3081957995891571,
      "learning_rate": 6.454473154305634e-06,
      "loss": 0.0271,
      "step": 40025
    },
    {
      "epoch": 2.0320828468450176,
      "grad_norm": 0.25008490681648254,
      "learning_rate": 6.452781021033217e-06,
      "loss": 0.0349,
      "step": 40030
    },
    {
      "epoch": 2.03233666683588,
      "grad_norm": 0.40683382749557495,
      "learning_rate": 6.4510888877608e-06,
      "loss": 0.0342,
      "step": 40035
    },
    {
      "epoch": 2.0325904868267424,
      "grad_norm": 0.31075912714004517,
      "learning_rate": 6.449396754488384e-06,
      "loss": 0.0338,
      "step": 40040
    },
    {
      "epoch": 2.032844306817605,
      "grad_norm": 0.28491321206092834,
      "learning_rate": 6.447704621215968e-06,
      "loss": 0.0463,
      "step": 40045
    },
    {
      "epoch": 2.0330981268084676,
      "grad_norm": 0.24667423963546753,
      "learning_rate": 6.446012487943551e-06,
      "loss": 0.0372,
      "step": 40050
    },
    {
      "epoch": 2.0333519467993297,
      "grad_norm": 0.2972967028617859,
      "learning_rate": 6.444320354671134e-06,
      "loss": 0.029,
      "step": 40055
    },
    {
      "epoch": 2.0336057667901923,
      "grad_norm": 0.3517604470252991,
      "learning_rate": 6.442628221398718e-06,
      "loss": 0.0391,
      "step": 40060
    },
    {
      "epoch": 2.033859586781055,
      "grad_norm": 0.3772994875907898,
      "learning_rate": 6.440936088126302e-06,
      "loss": 0.0242,
      "step": 40065
    },
    {
      "epoch": 2.0341134067719175,
      "grad_norm": 0.39820396900177,
      "learning_rate": 6.439243954853885e-06,
      "loss": 0.0368,
      "step": 40070
    },
    {
      "epoch": 2.0343672267627797,
      "grad_norm": 0.21249252557754517,
      "learning_rate": 6.437551821581468e-06,
      "loss": 0.0421,
      "step": 40075
    },
    {
      "epoch": 2.0346210467536423,
      "grad_norm": 0.3023703694343567,
      "learning_rate": 6.435859688309052e-06,
      "loss": 0.0316,
      "step": 40080
    },
    {
      "epoch": 2.034874866744505,
      "grad_norm": 0.3016105592250824,
      "learning_rate": 6.434167555036635e-06,
      "loss": 0.0365,
      "step": 40085
    },
    {
      "epoch": 2.035128686735367,
      "grad_norm": 0.38153544068336487,
      "learning_rate": 6.432475421764218e-06,
      "loss": 0.0319,
      "step": 40090
    },
    {
      "epoch": 2.0353825067262297,
      "grad_norm": 0.33254504203796387,
      "learning_rate": 6.430783288491802e-06,
      "loss": 0.0287,
      "step": 40095
    },
    {
      "epoch": 2.0356363267170923,
      "grad_norm": 0.41204509139060974,
      "learning_rate": 6.429091155219386e-06,
      "loss": 0.0371,
      "step": 40100
    },
    {
      "epoch": 2.035890146707955,
      "grad_norm": 0.5164371728897095,
      "learning_rate": 6.4273990219469695e-06,
      "loss": 0.0291,
      "step": 40105
    },
    {
      "epoch": 2.036143966698817,
      "grad_norm": 0.30971401929855347,
      "learning_rate": 6.425706888674552e-06,
      "loss": 0.0331,
      "step": 40110
    },
    {
      "epoch": 2.0363977866896796,
      "grad_norm": 0.4166437089443207,
      "learning_rate": 6.424014755402136e-06,
      "loss": 0.0347,
      "step": 40115
    },
    {
      "epoch": 2.0366516066805422,
      "grad_norm": 0.4730953872203827,
      "learning_rate": 6.422322622129719e-06,
      "loss": 0.0418,
      "step": 40120
    },
    {
      "epoch": 2.036905426671405,
      "grad_norm": 0.2546430826187134,
      "learning_rate": 6.420630488857304e-06,
      "loss": 0.0345,
      "step": 40125
    },
    {
      "epoch": 2.037159246662267,
      "grad_norm": 0.35239869356155396,
      "learning_rate": 6.418938355584886e-06,
      "loss": 0.0342,
      "step": 40130
    },
    {
      "epoch": 2.0374130666531296,
      "grad_norm": 0.34764301776885986,
      "learning_rate": 6.41724622231247e-06,
      "loss": 0.0284,
      "step": 40135
    },
    {
      "epoch": 2.037666886643992,
      "grad_norm": 0.30288177728652954,
      "learning_rate": 6.4155540890400535e-06,
      "loss": 0.0407,
      "step": 40140
    },
    {
      "epoch": 2.0379207066348544,
      "grad_norm": 0.2945058047771454,
      "learning_rate": 6.413861955767636e-06,
      "loss": 0.0423,
      "step": 40145
    },
    {
      "epoch": 2.038174526625717,
      "grad_norm": 0.37763145565986633,
      "learning_rate": 6.41216982249522e-06,
      "loss": 0.0324,
      "step": 40150
    },
    {
      "epoch": 2.0384283466165796,
      "grad_norm": 0.5259355902671814,
      "learning_rate": 6.410477689222804e-06,
      "loss": 0.0356,
      "step": 40155
    },
    {
      "epoch": 2.038682166607442,
      "grad_norm": 0.4084901511669159,
      "learning_rate": 6.408785555950388e-06,
      "loss": 0.0414,
      "step": 40160
    },
    {
      "epoch": 2.0389359865983043,
      "grad_norm": 0.3193150758743286,
      "learning_rate": 6.40709342267797e-06,
      "loss": 0.03,
      "step": 40165
    },
    {
      "epoch": 2.039189806589167,
      "grad_norm": 0.36446109414100647,
      "learning_rate": 6.405401289405554e-06,
      "loss": 0.0317,
      "step": 40170
    },
    {
      "epoch": 2.0394436265800295,
      "grad_norm": 0.40664389729499817,
      "learning_rate": 6.4037091561331374e-06,
      "loss": 0.0339,
      "step": 40175
    },
    {
      "epoch": 2.039697446570892,
      "grad_norm": 0.4111139178276062,
      "learning_rate": 6.402017022860722e-06,
      "loss": 0.0389,
      "step": 40180
    },
    {
      "epoch": 2.0399512665617543,
      "grad_norm": 0.3648948073387146,
      "learning_rate": 6.4003248895883045e-06,
      "loss": 0.0265,
      "step": 40185
    },
    {
      "epoch": 2.040205086552617,
      "grad_norm": 0.27332165837287903,
      "learning_rate": 6.398632756315888e-06,
      "loss": 0.0345,
      "step": 40190
    },
    {
      "epoch": 2.0404589065434795,
      "grad_norm": 0.2328202873468399,
      "learning_rate": 6.396940623043472e-06,
      "loss": 0.0295,
      "step": 40195
    },
    {
      "epoch": 2.0407127265343417,
      "grad_norm": 0.3368794023990631,
      "learning_rate": 6.395248489771055e-06,
      "loss": 0.0431,
      "step": 40200
    },
    {
      "epoch": 2.0409665465252043,
      "grad_norm": 0.2832939624786377,
      "learning_rate": 6.393556356498638e-06,
      "loss": 0.0323,
      "step": 40205
    },
    {
      "epoch": 2.041220366516067,
      "grad_norm": 0.6089285612106323,
      "learning_rate": 6.391864223226221e-06,
      "loss": 0.042,
      "step": 40210
    },
    {
      "epoch": 2.0414741865069295,
      "grad_norm": 0.341775506734848,
      "learning_rate": 6.390172089953806e-06,
      "loss": 0.0338,
      "step": 40215
    },
    {
      "epoch": 2.0417280064977916,
      "grad_norm": 0.24220286309719086,
      "learning_rate": 6.3884799566813885e-06,
      "loss": 0.0295,
      "step": 40220
    },
    {
      "epoch": 2.0419818264886542,
      "grad_norm": 0.31153371930122375,
      "learning_rate": 6.386787823408972e-06,
      "loss": 0.035,
      "step": 40225
    },
    {
      "epoch": 2.042235646479517,
      "grad_norm": 0.3575878441333771,
      "learning_rate": 6.385095690136556e-06,
      "loss": 0.0377,
      "step": 40230
    },
    {
      "epoch": 2.042489466470379,
      "grad_norm": 0.44863882660865784,
      "learning_rate": 6.383403556864139e-06,
      "loss": 0.0419,
      "step": 40235
    },
    {
      "epoch": 2.0427432864612416,
      "grad_norm": 0.24326546490192413,
      "learning_rate": 6.381711423591722e-06,
      "loss": 0.0317,
      "step": 40240
    },
    {
      "epoch": 2.042997106452104,
      "grad_norm": 0.5627332329750061,
      "learning_rate": 6.380019290319306e-06,
      "loss": 0.0346,
      "step": 40245
    },
    {
      "epoch": 2.043250926442967,
      "grad_norm": 0.22772236168384552,
      "learning_rate": 6.37832715704689e-06,
      "loss": 0.0389,
      "step": 40250
    },
    {
      "epoch": 2.043504746433829,
      "grad_norm": 0.383705735206604,
      "learning_rate": 6.376635023774473e-06,
      "loss": 0.0403,
      "step": 40255
    },
    {
      "epoch": 2.0437585664246916,
      "grad_norm": 0.40597113966941833,
      "learning_rate": 6.374942890502056e-06,
      "loss": 0.0349,
      "step": 40260
    },
    {
      "epoch": 2.044012386415554,
      "grad_norm": 0.3816896378993988,
      "learning_rate": 6.3732507572296395e-06,
      "loss": 0.0363,
      "step": 40265
    },
    {
      "epoch": 2.0442662064064168,
      "grad_norm": 0.29251110553741455,
      "learning_rate": 6.371558623957224e-06,
      "loss": 0.0321,
      "step": 40270
    },
    {
      "epoch": 2.044520026397279,
      "grad_norm": 0.33574673533439636,
      "learning_rate": 6.3698664906848075e-06,
      "loss": 0.0334,
      "step": 40275
    },
    {
      "epoch": 2.0447738463881415,
      "grad_norm": 0.24987970292568207,
      "learning_rate": 6.36817435741239e-06,
      "loss": 0.0312,
      "step": 40280
    },
    {
      "epoch": 2.045027666379004,
      "grad_norm": 0.2474692314863205,
      "learning_rate": 6.366482224139974e-06,
      "loss": 0.0359,
      "step": 40285
    },
    {
      "epoch": 2.0452814863698663,
      "grad_norm": 0.24791835248470306,
      "learning_rate": 6.364790090867557e-06,
      "loss": 0.0355,
      "step": 40290
    },
    {
      "epoch": 2.045535306360729,
      "grad_norm": 0.41668781638145447,
      "learning_rate": 6.363097957595142e-06,
      "loss": 0.0379,
      "step": 40295
    },
    {
      "epoch": 2.0457891263515915,
      "grad_norm": 0.4789046347141266,
      "learning_rate": 6.3614058243227235e-06,
      "loss": 0.0323,
      "step": 40300
    },
    {
      "epoch": 2.046042946342454,
      "grad_norm": 0.4169135093688965,
      "learning_rate": 6.359713691050308e-06,
      "loss": 0.0282,
      "step": 40305
    },
    {
      "epoch": 2.0462967663333163,
      "grad_norm": 0.3398691713809967,
      "learning_rate": 6.3580215577778914e-06,
      "loss": 0.0372,
      "step": 40310
    },
    {
      "epoch": 2.046550586324179,
      "grad_norm": 0.3975592255592346,
      "learning_rate": 6.356329424505474e-06,
      "loss": 0.0342,
      "step": 40315
    },
    {
      "epoch": 2.0468044063150415,
      "grad_norm": 0.28930965065956116,
      "learning_rate": 6.354637291233058e-06,
      "loss": 0.0398,
      "step": 40320
    },
    {
      "epoch": 2.047058226305904,
      "grad_norm": 0.311261922121048,
      "learning_rate": 6.352945157960641e-06,
      "loss": 0.0354,
      "step": 40325
    },
    {
      "epoch": 2.047312046296766,
      "grad_norm": 0.441555917263031,
      "learning_rate": 6.351253024688226e-06,
      "loss": 0.0357,
      "step": 40330
    },
    {
      "epoch": 2.047565866287629,
      "grad_norm": 0.8066316843032837,
      "learning_rate": 6.349560891415808e-06,
      "loss": 0.038,
      "step": 40335
    },
    {
      "epoch": 2.0478196862784914,
      "grad_norm": 0.39569395780563354,
      "learning_rate": 6.347868758143392e-06,
      "loss": 0.0428,
      "step": 40340
    },
    {
      "epoch": 2.0480735062693536,
      "grad_norm": 0.28354886174201965,
      "learning_rate": 6.346176624870975e-06,
      "loss": 0.0329,
      "step": 40345
    },
    {
      "epoch": 2.048327326260216,
      "grad_norm": 0.29512515664100647,
      "learning_rate": 6.344484491598559e-06,
      "loss": 0.0334,
      "step": 40350
    },
    {
      "epoch": 2.048581146251079,
      "grad_norm": 0.35090482234954834,
      "learning_rate": 6.342792358326142e-06,
      "loss": 0.0335,
      "step": 40355
    },
    {
      "epoch": 2.0488349662419414,
      "grad_norm": 0.48609450459480286,
      "learning_rate": 6.341100225053726e-06,
      "loss": 0.0357,
      "step": 40360
    },
    {
      "epoch": 2.0490887862328035,
      "grad_norm": 0.2837776839733124,
      "learning_rate": 6.33940809178131e-06,
      "loss": 0.0336,
      "step": 40365
    },
    {
      "epoch": 2.049342606223666,
      "grad_norm": 0.8645208477973938,
      "learning_rate": 6.337715958508893e-06,
      "loss": 0.0291,
      "step": 40370
    },
    {
      "epoch": 2.0495964262145288,
      "grad_norm": 0.3983393907546997,
      "learning_rate": 6.336023825236476e-06,
      "loss": 0.0422,
      "step": 40375
    },
    {
      "epoch": 2.0498502462053914,
      "grad_norm": 0.3151146173477173,
      "learning_rate": 6.334331691964059e-06,
      "loss": 0.0325,
      "step": 40380
    },
    {
      "epoch": 2.0501040661962535,
      "grad_norm": 0.27322548627853394,
      "learning_rate": 6.332639558691644e-06,
      "loss": 0.0348,
      "step": 40385
    },
    {
      "epoch": 2.050357886187116,
      "grad_norm": 0.3726036846637726,
      "learning_rate": 6.330947425419226e-06,
      "loss": 0.0324,
      "step": 40390
    },
    {
      "epoch": 2.0506117061779787,
      "grad_norm": 0.3639608323574066,
      "learning_rate": 6.32925529214681e-06,
      "loss": 0.0334,
      "step": 40395
    },
    {
      "epoch": 2.050865526168841,
      "grad_norm": 0.2874867618083954,
      "learning_rate": 6.3275631588743936e-06,
      "loss": 0.0311,
      "step": 40400
    },
    {
      "epoch": 2.0511193461597035,
      "grad_norm": 0.4578050374984741,
      "learning_rate": 6.325871025601977e-06,
      "loss": 0.0338,
      "step": 40405
    },
    {
      "epoch": 2.051373166150566,
      "grad_norm": 0.5080977082252502,
      "learning_rate": 6.32417889232956e-06,
      "loss": 0.0354,
      "step": 40410
    },
    {
      "epoch": 2.0516269861414287,
      "grad_norm": 0.4109554886817932,
      "learning_rate": 6.322486759057143e-06,
      "loss": 0.0367,
      "step": 40415
    },
    {
      "epoch": 2.051880806132291,
      "grad_norm": 0.5904496312141418,
      "learning_rate": 6.320794625784728e-06,
      "loss": 0.0429,
      "step": 40420
    },
    {
      "epoch": 2.0521346261231534,
      "grad_norm": 0.43072524666786194,
      "learning_rate": 6.319102492512311e-06,
      "loss": 0.0319,
      "step": 40425
    },
    {
      "epoch": 2.052388446114016,
      "grad_norm": 0.29243141412734985,
      "learning_rate": 6.317410359239894e-06,
      "loss": 0.034,
      "step": 40430
    },
    {
      "epoch": 2.052642266104878,
      "grad_norm": 0.35898685455322266,
      "learning_rate": 6.3157182259674775e-06,
      "loss": 0.0325,
      "step": 40435
    },
    {
      "epoch": 2.052896086095741,
      "grad_norm": 1.1760319471359253,
      "learning_rate": 6.314026092695061e-06,
      "loss": 0.0293,
      "step": 40440
    },
    {
      "epoch": 2.0531499060866034,
      "grad_norm": 0.2817917466163635,
      "learning_rate": 6.3123339594226455e-06,
      "loss": 0.0349,
      "step": 40445
    },
    {
      "epoch": 2.053403726077466,
      "grad_norm": 0.35304921865463257,
      "learning_rate": 6.310641826150228e-06,
      "loss": 0.0337,
      "step": 40450
    },
    {
      "epoch": 2.053657546068328,
      "grad_norm": 0.4008921682834625,
      "learning_rate": 6.308949692877812e-06,
      "loss": 0.0426,
      "step": 40455
    },
    {
      "epoch": 2.053911366059191,
      "grad_norm": 0.3372136950492859,
      "learning_rate": 6.307257559605395e-06,
      "loss": 0.0345,
      "step": 40460
    },
    {
      "epoch": 2.0541651860500534,
      "grad_norm": 0.5483426451683044,
      "learning_rate": 6.305565426332978e-06,
      "loss": 0.0302,
      "step": 40465
    },
    {
      "epoch": 2.054419006040916,
      "grad_norm": 0.18743745982646942,
      "learning_rate": 6.3038732930605615e-06,
      "loss": 0.0402,
      "step": 40470
    },
    {
      "epoch": 2.054672826031778,
      "grad_norm": 0.3917071521282196,
      "learning_rate": 6.302181159788146e-06,
      "loss": 0.0295,
      "step": 40475
    },
    {
      "epoch": 2.0549266460226407,
      "grad_norm": 0.3281704783439636,
      "learning_rate": 6.300489026515729e-06,
      "loss": 0.036,
      "step": 40480
    },
    {
      "epoch": 2.0551804660135033,
      "grad_norm": 0.3832402527332306,
      "learning_rate": 6.298796893243312e-06,
      "loss": 0.0338,
      "step": 40485
    },
    {
      "epoch": 2.0554342860043655,
      "grad_norm": 0.4939913749694824,
      "learning_rate": 6.297104759970896e-06,
      "loss": 0.037,
      "step": 40490
    },
    {
      "epoch": 2.055688105995228,
      "grad_norm": 0.34781306982040405,
      "learning_rate": 6.295412626698479e-06,
      "loss": 0.0351,
      "step": 40495
    },
    {
      "epoch": 2.0559419259860907,
      "grad_norm": 0.516234278678894,
      "learning_rate": 6.293720493426063e-06,
      "loss": 0.0343,
      "step": 40500
    },
    {
      "epoch": 2.0561957459769533,
      "grad_norm": 0.3306788206100464,
      "learning_rate": 6.2920283601536455e-06,
      "loss": 0.0442,
      "step": 40505
    },
    {
      "epoch": 2.0564495659678155,
      "grad_norm": 0.2988586127758026,
      "learning_rate": 6.29033622688123e-06,
      "loss": 0.0357,
      "step": 40510
    },
    {
      "epoch": 2.056703385958678,
      "grad_norm": 0.25617992877960205,
      "learning_rate": 6.288644093608813e-06,
      "loss": 0.0292,
      "step": 40515
    },
    {
      "epoch": 2.0569572059495407,
      "grad_norm": 0.2515580356121063,
      "learning_rate": 6.286951960336397e-06,
      "loss": 0.0303,
      "step": 40520
    },
    {
      "epoch": 2.0572110259404033,
      "grad_norm": 0.20995844900608063,
      "learning_rate": 6.28525982706398e-06,
      "loss": 0.0347,
      "step": 40525
    },
    {
      "epoch": 2.0574648459312654,
      "grad_norm": 0.6654472947120667,
      "learning_rate": 6.283567693791563e-06,
      "loss": 0.0363,
      "step": 40530
    },
    {
      "epoch": 2.057718665922128,
      "grad_norm": 0.47852277755737305,
      "learning_rate": 6.2818755605191476e-06,
      "loss": 0.0385,
      "step": 40535
    },
    {
      "epoch": 2.0579724859129906,
      "grad_norm": 0.4235854744911194,
      "learning_rate": 6.28018342724673e-06,
      "loss": 0.0353,
      "step": 40540
    },
    {
      "epoch": 2.058226305903853,
      "grad_norm": 0.4624616801738739,
      "learning_rate": 6.278491293974314e-06,
      "loss": 0.0372,
      "step": 40545
    },
    {
      "epoch": 2.0584801258947154,
      "grad_norm": 0.3644164204597473,
      "learning_rate": 6.276799160701897e-06,
      "loss": 0.0349,
      "step": 40550
    },
    {
      "epoch": 2.058733945885578,
      "grad_norm": 0.39000046253204346,
      "learning_rate": 6.275107027429481e-06,
      "loss": 0.0387,
      "step": 40555
    },
    {
      "epoch": 2.0589877658764406,
      "grad_norm": 0.3902978301048279,
      "learning_rate": 6.273414894157064e-06,
      "loss": 0.0314,
      "step": 40560
    },
    {
      "epoch": 2.0592415858673028,
      "grad_norm": 0.3080569803714752,
      "learning_rate": 6.271722760884648e-06,
      "loss": 0.0396,
      "step": 40565
    },
    {
      "epoch": 2.0594954058581654,
      "grad_norm": 0.4491880536079407,
      "learning_rate": 6.2700306276122315e-06,
      "loss": 0.0316,
      "step": 40570
    },
    {
      "epoch": 2.059749225849028,
      "grad_norm": 0.4176236689090729,
      "learning_rate": 6.268338494339815e-06,
      "loss": 0.0359,
      "step": 40575
    },
    {
      "epoch": 2.06000304583989,
      "grad_norm": 0.334221214056015,
      "learning_rate": 6.266646361067398e-06,
      "loss": 0.0385,
      "step": 40580
    },
    {
      "epoch": 2.0602568658307527,
      "grad_norm": 0.37594276666641235,
      "learning_rate": 6.264954227794981e-06,
      "loss": 0.0371,
      "step": 40585
    },
    {
      "epoch": 2.0605106858216153,
      "grad_norm": 0.50478595495224,
      "learning_rate": 6.263262094522565e-06,
      "loss": 0.0351,
      "step": 40590
    },
    {
      "epoch": 2.060764505812478,
      "grad_norm": 0.31555843353271484,
      "learning_rate": 6.261569961250149e-06,
      "loss": 0.0399,
      "step": 40595
    },
    {
      "epoch": 2.06101832580334,
      "grad_norm": 0.40124326944351196,
      "learning_rate": 6.259877827977732e-06,
      "loss": 0.0326,
      "step": 40600
    },
    {
      "epoch": 2.0612721457942027,
      "grad_norm": 0.2583628296852112,
      "learning_rate": 6.2581856947053155e-06,
      "loss": 0.0318,
      "step": 40605
    },
    {
      "epoch": 2.0615259657850653,
      "grad_norm": 0.24880623817443848,
      "learning_rate": 6.256493561432899e-06,
      "loss": 0.0449,
      "step": 40610
    },
    {
      "epoch": 2.061779785775928,
      "grad_norm": 0.30163612961769104,
      "learning_rate": 6.254801428160483e-06,
      "loss": 0.0431,
      "step": 40615
    },
    {
      "epoch": 2.06203360576679,
      "grad_norm": 0.2596006989479065,
      "learning_rate": 6.253109294888065e-06,
      "loss": 0.0284,
      "step": 40620
    },
    {
      "epoch": 2.0622874257576527,
      "grad_norm": 0.2562941014766693,
      "learning_rate": 6.25141716161565e-06,
      "loss": 0.0301,
      "step": 40625
    },
    {
      "epoch": 2.0625412457485153,
      "grad_norm": 0.3008427619934082,
      "learning_rate": 6.249725028343233e-06,
      "loss": 0.0318,
      "step": 40630
    },
    {
      "epoch": 2.0627950657393774,
      "grad_norm": 0.335764080286026,
      "learning_rate": 6.248032895070816e-06,
      "loss": 0.0344,
      "step": 40635
    },
    {
      "epoch": 2.06304888573024,
      "grad_norm": 0.5868695378303528,
      "learning_rate": 6.2463407617983995e-06,
      "loss": 0.0366,
      "step": 40640
    },
    {
      "epoch": 2.0633027057211026,
      "grad_norm": 0.38782617449760437,
      "learning_rate": 6.244648628525983e-06,
      "loss": 0.03,
      "step": 40645
    },
    {
      "epoch": 2.0635565257119652,
      "grad_norm": 0.5476816296577454,
      "learning_rate": 6.242956495253567e-06,
      "loss": 0.0364,
      "step": 40650
    },
    {
      "epoch": 2.0638103457028274,
      "grad_norm": 0.27162158489227295,
      "learning_rate": 6.24126436198115e-06,
      "loss": 0.0307,
      "step": 40655
    },
    {
      "epoch": 2.06406416569369,
      "grad_norm": 0.41091370582580566,
      "learning_rate": 6.239572228708734e-06,
      "loss": 0.0273,
      "step": 40660
    },
    {
      "epoch": 2.0643179856845526,
      "grad_norm": 0.32241737842559814,
      "learning_rate": 6.237880095436317e-06,
      "loss": 0.0356,
      "step": 40665
    },
    {
      "epoch": 2.064571805675415,
      "grad_norm": 0.25782039761543274,
      "learning_rate": 6.236187962163901e-06,
      "loss": 0.0388,
      "step": 40670
    },
    {
      "epoch": 2.0648256256662774,
      "grad_norm": 0.3799375295639038,
      "learning_rate": 6.2344958288914834e-06,
      "loss": 0.0437,
      "step": 40675
    },
    {
      "epoch": 2.06507944565714,
      "grad_norm": 0.5712692141532898,
      "learning_rate": 6.232803695619068e-06,
      "loss": 0.0388,
      "step": 40680
    },
    {
      "epoch": 2.0653332656480026,
      "grad_norm": 0.47142764925956726,
      "learning_rate": 6.231111562346651e-06,
      "loss": 0.0337,
      "step": 40685
    },
    {
      "epoch": 2.0655870856388647,
      "grad_norm": 0.3855939507484436,
      "learning_rate": 6.229419429074235e-06,
      "loss": 0.0405,
      "step": 40690
    },
    {
      "epoch": 2.0658409056297273,
      "grad_norm": 0.36656343936920166,
      "learning_rate": 6.227727295801818e-06,
      "loss": 0.0339,
      "step": 40695
    },
    {
      "epoch": 2.06609472562059,
      "grad_norm": 0.529723048210144,
      "learning_rate": 6.226035162529401e-06,
      "loss": 0.0302,
      "step": 40700
    },
    {
      "epoch": 2.0663485456114525,
      "grad_norm": 0.41368111968040466,
      "learning_rate": 6.224343029256985e-06,
      "loss": 0.0269,
      "step": 40705
    },
    {
      "epoch": 2.0666023656023147,
      "grad_norm": 0.26356250047683716,
      "learning_rate": 6.222650895984567e-06,
      "loss": 0.0311,
      "step": 40710
    },
    {
      "epoch": 2.0668561855931773,
      "grad_norm": 0.42043641209602356,
      "learning_rate": 6.220958762712152e-06,
      "loss": 0.0305,
      "step": 40715
    },
    {
      "epoch": 2.06711000558404,
      "grad_norm": 0.3392391502857208,
      "learning_rate": 6.219266629439735e-06,
      "loss": 0.0395,
      "step": 40720
    },
    {
      "epoch": 2.067363825574902,
      "grad_norm": 0.43280452489852905,
      "learning_rate": 6.217574496167319e-06,
      "loss": 0.0341,
      "step": 40725
    },
    {
      "epoch": 2.0676176455657647,
      "grad_norm": 0.3381582200527191,
      "learning_rate": 6.2158823628949016e-06,
      "loss": 0.0269,
      "step": 40730
    },
    {
      "epoch": 2.0678714655566273,
      "grad_norm": 0.34322628378868103,
      "learning_rate": 6.214190229622485e-06,
      "loss": 0.036,
      "step": 40735
    },
    {
      "epoch": 2.06812528554749,
      "grad_norm": 0.40060219168663025,
      "learning_rate": 6.2124980963500695e-06,
      "loss": 0.033,
      "step": 40740
    },
    {
      "epoch": 2.068379105538352,
      "grad_norm": 0.21552379429340363,
      "learning_rate": 6.210805963077653e-06,
      "loss": 0.0284,
      "step": 40745
    },
    {
      "epoch": 2.0686329255292146,
      "grad_norm": 0.30288293957710266,
      "learning_rate": 6.209113829805236e-06,
      "loss": 0.0363,
      "step": 40750
    },
    {
      "epoch": 2.0688867455200772,
      "grad_norm": 0.29889294505119324,
      "learning_rate": 6.207421696532819e-06,
      "loss": 0.0335,
      "step": 40755
    },
    {
      "epoch": 2.06914056551094,
      "grad_norm": 0.24941956996917725,
      "learning_rate": 6.205729563260403e-06,
      "loss": 0.034,
      "step": 40760
    },
    {
      "epoch": 2.069394385501802,
      "grad_norm": 0.3261270523071289,
      "learning_rate": 6.204037429987987e-06,
      "loss": 0.0371,
      "step": 40765
    },
    {
      "epoch": 2.0696482054926646,
      "grad_norm": 0.3423829674720764,
      "learning_rate": 6.20234529671557e-06,
      "loss": 0.0502,
      "step": 40770
    },
    {
      "epoch": 2.069902025483527,
      "grad_norm": 0.34635868668556213,
      "learning_rate": 6.2006531634431535e-06,
      "loss": 0.0316,
      "step": 40775
    },
    {
      "epoch": 2.0701558454743894,
      "grad_norm": 0.2521602213382721,
      "learning_rate": 6.198961030170737e-06,
      "loss": 0.0366,
      "step": 40780
    },
    {
      "epoch": 2.070409665465252,
      "grad_norm": 0.22092673182487488,
      "learning_rate": 6.19726889689832e-06,
      "loss": 0.0309,
      "step": 40785
    },
    {
      "epoch": 2.0706634854561146,
      "grad_norm": 0.33411675691604614,
      "learning_rate": 6.195576763625903e-06,
      "loss": 0.039,
      "step": 40790
    },
    {
      "epoch": 2.070917305446977,
      "grad_norm": 0.22911129891872406,
      "learning_rate": 6.193884630353487e-06,
      "loss": 0.028,
      "step": 40795
    },
    {
      "epoch": 2.0711711254378393,
      "grad_norm": 4.144477844238281,
      "learning_rate": 6.192192497081071e-06,
      "loss": 0.0308,
      "step": 40800
    },
    {
      "epoch": 2.071424945428702,
      "grad_norm": 0.24567130208015442,
      "learning_rate": 6.190500363808654e-06,
      "loss": 0.0316,
      "step": 40805
    },
    {
      "epoch": 2.0716787654195645,
      "grad_norm": 0.3412927985191345,
      "learning_rate": 6.1888082305362374e-06,
      "loss": 0.0327,
      "step": 40810
    },
    {
      "epoch": 2.071932585410427,
      "grad_norm": 0.2085569053888321,
      "learning_rate": 6.187116097263821e-06,
      "loss": 0.0341,
      "step": 40815
    },
    {
      "epoch": 2.0721864054012893,
      "grad_norm": 0.2844689190387726,
      "learning_rate": 6.1854239639914045e-06,
      "loss": 0.0332,
      "step": 40820
    },
    {
      "epoch": 2.072440225392152,
      "grad_norm": 0.431959331035614,
      "learning_rate": 6.183731830718987e-06,
      "loss": 0.033,
      "step": 40825
    },
    {
      "epoch": 2.0726940453830145,
      "grad_norm": 0.2555539011955261,
      "learning_rate": 6.182039697446572e-06,
      "loss": 0.0349,
      "step": 40830
    },
    {
      "epoch": 2.0729478653738767,
      "grad_norm": 0.35910919308662415,
      "learning_rate": 6.180347564174155e-06,
      "loss": 0.0355,
      "step": 40835
    },
    {
      "epoch": 2.0732016853647393,
      "grad_norm": 0.34894514083862305,
      "learning_rate": 6.178655430901739e-06,
      "loss": 0.0316,
      "step": 40840
    },
    {
      "epoch": 2.073455505355602,
      "grad_norm": 0.25370797514915466,
      "learning_rate": 6.176963297629321e-06,
      "loss": 0.0401,
      "step": 40845
    },
    {
      "epoch": 2.0737093253464645,
      "grad_norm": 0.3521358072757721,
      "learning_rate": 6.175271164356905e-06,
      "loss": 0.0334,
      "step": 40850
    },
    {
      "epoch": 2.0739631453373266,
      "grad_norm": 0.3908250331878662,
      "learning_rate": 6.173579031084489e-06,
      "loss": 0.0355,
      "step": 40855
    },
    {
      "epoch": 2.074216965328189,
      "grad_norm": 0.31335461139678955,
      "learning_rate": 6.171886897812073e-06,
      "loss": 0.035,
      "step": 40860
    },
    {
      "epoch": 2.074470785319052,
      "grad_norm": 0.404501736164093,
      "learning_rate": 6.170194764539656e-06,
      "loss": 0.0357,
      "step": 40865
    },
    {
      "epoch": 2.074724605309914,
      "grad_norm": 0.39813360571861267,
      "learning_rate": 6.168502631267239e-06,
      "loss": 0.0325,
      "step": 40870
    },
    {
      "epoch": 2.0749784253007766,
      "grad_norm": 0.3120538592338562,
      "learning_rate": 6.166810497994823e-06,
      "loss": 0.0332,
      "step": 40875
    },
    {
      "epoch": 2.075232245291639,
      "grad_norm": 0.24379944801330566,
      "learning_rate": 6.165118364722405e-06,
      "loss": 0.0328,
      "step": 40880
    },
    {
      "epoch": 2.075486065282502,
      "grad_norm": 0.3763482868671417,
      "learning_rate": 6.163426231449989e-06,
      "loss": 0.0434,
      "step": 40885
    },
    {
      "epoch": 2.075739885273364,
      "grad_norm": 0.27463313937187195,
      "learning_rate": 6.161734098177573e-06,
      "loss": 0.0314,
      "step": 40890
    },
    {
      "epoch": 2.0759937052642266,
      "grad_norm": 0.3216022849082947,
      "learning_rate": 6.160041964905157e-06,
      "loss": 0.0356,
      "step": 40895
    },
    {
      "epoch": 2.076247525255089,
      "grad_norm": 0.4241223633289337,
      "learning_rate": 6.1583498316327395e-06,
      "loss": 0.0361,
      "step": 40900
    },
    {
      "epoch": 2.0765013452459518,
      "grad_norm": 0.4173457622528076,
      "learning_rate": 6.156657698360323e-06,
      "loss": 0.0287,
      "step": 40905
    },
    {
      "epoch": 2.076755165236814,
      "grad_norm": 0.4344845712184906,
      "learning_rate": 6.154965565087907e-06,
      "loss": 0.0383,
      "step": 40910
    },
    {
      "epoch": 2.0770089852276765,
      "grad_norm": 0.38608500361442566,
      "learning_rate": 6.153273431815491e-06,
      "loss": 0.0305,
      "step": 40915
    },
    {
      "epoch": 2.077262805218539,
      "grad_norm": 0.37664130330085754,
      "learning_rate": 6.151581298543074e-06,
      "loss": 0.0358,
      "step": 40920
    },
    {
      "epoch": 2.0775166252094013,
      "grad_norm": 0.36590301990509033,
      "learning_rate": 6.149889165270657e-06,
      "loss": 0.0372,
      "step": 40925
    },
    {
      "epoch": 2.077770445200264,
      "grad_norm": 0.2959653437137604,
      "learning_rate": 6.148197031998241e-06,
      "loss": 0.0309,
      "step": 40930
    },
    {
      "epoch": 2.0780242651911265,
      "grad_norm": 0.20683987438678741,
      "learning_rate": 6.146504898725824e-06,
      "loss": 0.0359,
      "step": 40935
    },
    {
      "epoch": 2.078278085181989,
      "grad_norm": 0.4567297101020813,
      "learning_rate": 6.144812765453407e-06,
      "loss": 0.0349,
      "step": 40940
    },
    {
      "epoch": 2.0785319051728512,
      "grad_norm": 0.4936054050922394,
      "learning_rate": 6.1431206321809914e-06,
      "loss": 0.0291,
      "step": 40945
    },
    {
      "epoch": 2.078785725163714,
      "grad_norm": 0.45569735765457153,
      "learning_rate": 6.141428498908575e-06,
      "loss": 0.0383,
      "step": 40950
    },
    {
      "epoch": 2.0790395451545765,
      "grad_norm": 0.26285237073898315,
      "learning_rate": 6.139736365636158e-06,
      "loss": 0.0269,
      "step": 40955
    },
    {
      "epoch": 2.079293365145439,
      "grad_norm": 0.4176620841026306,
      "learning_rate": 6.138044232363741e-06,
      "loss": 0.0335,
      "step": 40960
    },
    {
      "epoch": 2.079547185136301,
      "grad_norm": 0.3247475028038025,
      "learning_rate": 6.136352099091325e-06,
      "loss": 0.0369,
      "step": 40965
    },
    {
      "epoch": 2.079801005127164,
      "grad_norm": 0.37378448247909546,
      "learning_rate": 6.134659965818909e-06,
      "loss": 0.0439,
      "step": 40970
    },
    {
      "epoch": 2.0800548251180264,
      "grad_norm": 0.35308516025543213,
      "learning_rate": 6.132967832546491e-06,
      "loss": 0.0385,
      "step": 40975
    },
    {
      "epoch": 2.0803086451088886,
      "grad_norm": 0.3680042624473572,
      "learning_rate": 6.131275699274075e-06,
      "loss": 0.0321,
      "step": 40980
    },
    {
      "epoch": 2.080562465099751,
      "grad_norm": 0.27900105714797974,
      "learning_rate": 6.129583566001659e-06,
      "loss": 0.0325,
      "step": 40985
    },
    {
      "epoch": 2.080816285090614,
      "grad_norm": 0.3695995509624481,
      "learning_rate": 6.1278914327292425e-06,
      "loss": 0.0301,
      "step": 40990
    },
    {
      "epoch": 2.0810701050814764,
      "grad_norm": 0.5086187124252319,
      "learning_rate": 6.126199299456825e-06,
      "loss": 0.0364,
      "step": 40995
    },
    {
      "epoch": 2.0813239250723385,
      "grad_norm": 0.38260045647621155,
      "learning_rate": 6.124507166184409e-06,
      "loss": 0.0364,
      "step": 41000
    },
    {
      "epoch": 2.081577745063201,
      "grad_norm": 0.3308844566345215,
      "learning_rate": 6.122815032911993e-06,
      "loss": 0.028,
      "step": 41005
    },
    {
      "epoch": 2.0818315650540637,
      "grad_norm": 1.5857940912246704,
      "learning_rate": 6.121122899639577e-06,
      "loss": 0.0435,
      "step": 41010
    },
    {
      "epoch": 2.0820853850449264,
      "grad_norm": 0.3290964365005493,
      "learning_rate": 6.119430766367159e-06,
      "loss": 0.03,
      "step": 41015
    },
    {
      "epoch": 2.0823392050357885,
      "grad_norm": 0.9360296130180359,
      "learning_rate": 6.117738633094743e-06,
      "loss": 0.0372,
      "step": 41020
    },
    {
      "epoch": 2.082593025026651,
      "grad_norm": 0.435088574886322,
      "learning_rate": 6.1160464998223265e-06,
      "loss": 0.0323,
      "step": 41025
    },
    {
      "epoch": 2.0828468450175137,
      "grad_norm": 0.307736873626709,
      "learning_rate": 6.114354366549909e-06,
      "loss": 0.0336,
      "step": 41030
    },
    {
      "epoch": 2.083100665008376,
      "grad_norm": 0.34247100353240967,
      "learning_rate": 6.1126622332774936e-06,
      "loss": 0.0398,
      "step": 41035
    },
    {
      "epoch": 2.0833544849992385,
      "grad_norm": 0.2559192478656769,
      "learning_rate": 6.110970100005077e-06,
      "loss": 0.0318,
      "step": 41040
    },
    {
      "epoch": 2.083608304990101,
      "grad_norm": 0.42116183042526245,
      "learning_rate": 6.109277966732661e-06,
      "loss": 0.0294,
      "step": 41045
    },
    {
      "epoch": 2.0838621249809637,
      "grad_norm": 0.36998796463012695,
      "learning_rate": 6.107585833460243e-06,
      "loss": 0.0314,
      "step": 41050
    },
    {
      "epoch": 2.084115944971826,
      "grad_norm": 0.3762056231498718,
      "learning_rate": 6.105893700187827e-06,
      "loss": 0.0422,
      "step": 41055
    },
    {
      "epoch": 2.0843697649626884,
      "grad_norm": 0.24426433444023132,
      "learning_rate": 6.104201566915411e-06,
      "loss": 0.0381,
      "step": 41060
    },
    {
      "epoch": 2.084623584953551,
      "grad_norm": 0.21887318789958954,
      "learning_rate": 6.102509433642995e-06,
      "loss": 0.0313,
      "step": 41065
    },
    {
      "epoch": 2.0848774049444136,
      "grad_norm": 0.3165340721607208,
      "learning_rate": 6.1008173003705775e-06,
      "loss": 0.0316,
      "step": 41070
    },
    {
      "epoch": 2.085131224935276,
      "grad_norm": 0.4518713355064392,
      "learning_rate": 6.099125167098161e-06,
      "loss": 0.0359,
      "step": 41075
    },
    {
      "epoch": 2.0853850449261384,
      "grad_norm": 0.3209543824195862,
      "learning_rate": 6.097433033825745e-06,
      "loss": 0.0383,
      "step": 41080
    },
    {
      "epoch": 2.085638864917001,
      "grad_norm": 0.2539142668247223,
      "learning_rate": 6.095740900553328e-06,
      "loss": 0.0288,
      "step": 41085
    },
    {
      "epoch": 2.085892684907863,
      "grad_norm": 0.327890008687973,
      "learning_rate": 6.094048767280911e-06,
      "loss": 0.0335,
      "step": 41090
    },
    {
      "epoch": 2.0861465048987258,
      "grad_norm": 0.40916168689727783,
      "learning_rate": 6.092356634008495e-06,
      "loss": 0.032,
      "step": 41095
    },
    {
      "epoch": 2.0864003248895884,
      "grad_norm": 0.5362857580184937,
      "learning_rate": 6.090664500736079e-06,
      "loss": 0.0312,
      "step": 41100
    },
    {
      "epoch": 2.086654144880451,
      "grad_norm": 0.20822367072105408,
      "learning_rate": 6.0889723674636615e-06,
      "loss": 0.0347,
      "step": 41105
    },
    {
      "epoch": 2.086907964871313,
      "grad_norm": 0.25241541862487793,
      "learning_rate": 6.087280234191245e-06,
      "loss": 0.0347,
      "step": 41110
    },
    {
      "epoch": 2.0871617848621757,
      "grad_norm": 0.3435940742492676,
      "learning_rate": 6.0855881009188286e-06,
      "loss": 0.0375,
      "step": 41115
    },
    {
      "epoch": 2.0874156048530383,
      "grad_norm": 0.37276408076286316,
      "learning_rate": 6.083895967646413e-06,
      "loss": 0.0413,
      "step": 41120
    },
    {
      "epoch": 2.0876694248439005,
      "grad_norm": 0.30192676186561584,
      "learning_rate": 6.082203834373996e-06,
      "loss": 0.0288,
      "step": 41125
    },
    {
      "epoch": 2.087923244834763,
      "grad_norm": 0.304162859916687,
      "learning_rate": 6.080511701101579e-06,
      "loss": 0.0346,
      "step": 41130
    },
    {
      "epoch": 2.0881770648256257,
      "grad_norm": 0.3105781078338623,
      "learning_rate": 6.078819567829163e-06,
      "loss": 0.0358,
      "step": 41135
    },
    {
      "epoch": 2.0884308848164883,
      "grad_norm": 0.29993224143981934,
      "learning_rate": 6.077127434556746e-06,
      "loss": 0.0362,
      "step": 41140
    },
    {
      "epoch": 2.0886847048073505,
      "grad_norm": 0.3587764501571655,
      "learning_rate": 6.075435301284329e-06,
      "loss": 0.0332,
      "step": 41145
    },
    {
      "epoch": 2.088938524798213,
      "grad_norm": 0.40033718943595886,
      "learning_rate": 6.073743168011913e-06,
      "loss": 0.0415,
      "step": 41150
    },
    {
      "epoch": 2.0891923447890757,
      "grad_norm": 0.26234421133995056,
      "learning_rate": 6.072051034739497e-06,
      "loss": 0.0298,
      "step": 41155
    },
    {
      "epoch": 2.0894461647799383,
      "grad_norm": 0.6442959904670715,
      "learning_rate": 6.0703589014670805e-06,
      "loss": 0.0369,
      "step": 41160
    },
    {
      "epoch": 2.0896999847708004,
      "grad_norm": 3.047675848007202,
      "learning_rate": 6.068666768194663e-06,
      "loss": 0.0298,
      "step": 41165
    },
    {
      "epoch": 2.089953804761663,
      "grad_norm": 0.3974636197090149,
      "learning_rate": 6.066974634922247e-06,
      "loss": 0.0372,
      "step": 41170
    },
    {
      "epoch": 2.0902076247525256,
      "grad_norm": 0.3301776945590973,
      "learning_rate": 6.06528250164983e-06,
      "loss": 0.0358,
      "step": 41175
    },
    {
      "epoch": 2.090461444743388,
      "grad_norm": 0.4375455975532532,
      "learning_rate": 6.063590368377415e-06,
      "loss": 0.0282,
      "step": 41180
    },
    {
      "epoch": 2.0907152647342504,
      "grad_norm": 0.28423193097114563,
      "learning_rate": 6.061898235104997e-06,
      "loss": 0.0269,
      "step": 41185
    },
    {
      "epoch": 2.090969084725113,
      "grad_norm": 0.3169099986553192,
      "learning_rate": 6.060206101832581e-06,
      "loss": 0.0336,
      "step": 41190
    },
    {
      "epoch": 2.0912229047159756,
      "grad_norm": 0.3071502447128296,
      "learning_rate": 6.0585139685601644e-06,
      "loss": 0.041,
      "step": 41195
    },
    {
      "epoch": 2.0914767247068378,
      "grad_norm": 0.29963594675064087,
      "learning_rate": 6.056821835287747e-06,
      "loss": 0.0348,
      "step": 41200
    },
    {
      "epoch": 2.0917305446977004,
      "grad_norm": 0.3469637632369995,
      "learning_rate": 6.055129702015331e-06,
      "loss": 0.0343,
      "step": 41205
    },
    {
      "epoch": 2.091984364688563,
      "grad_norm": 0.6181375980377197,
      "learning_rate": 6.053437568742915e-06,
      "loss": 0.041,
      "step": 41210
    },
    {
      "epoch": 2.0922381846794256,
      "grad_norm": 0.7354362607002258,
      "learning_rate": 6.051745435470499e-06,
      "loss": 0.0401,
      "step": 41215
    },
    {
      "epoch": 2.0924920046702877,
      "grad_norm": 0.3352373242378235,
      "learning_rate": 6.050053302198081e-06,
      "loss": 0.0297,
      "step": 41220
    },
    {
      "epoch": 2.0927458246611503,
      "grad_norm": 0.4805290102958679,
      "learning_rate": 6.048361168925665e-06,
      "loss": 0.0429,
      "step": 41225
    },
    {
      "epoch": 2.092999644652013,
      "grad_norm": 0.47889429330825806,
      "learning_rate": 6.046669035653248e-06,
      "loss": 0.0407,
      "step": 41230
    },
    {
      "epoch": 2.093253464642875,
      "grad_norm": 0.2577539086341858,
      "learning_rate": 6.044976902380833e-06,
      "loss": 0.0312,
      "step": 41235
    },
    {
      "epoch": 2.0935072846337377,
      "grad_norm": 0.29059699177742004,
      "learning_rate": 6.0432847691084155e-06,
      "loss": 0.0312,
      "step": 41240
    },
    {
      "epoch": 2.0937611046246003,
      "grad_norm": 0.23989014327526093,
      "learning_rate": 6.041592635835999e-06,
      "loss": 0.0355,
      "step": 41245
    },
    {
      "epoch": 2.094014924615463,
      "grad_norm": 0.41318488121032715,
      "learning_rate": 6.039900502563583e-06,
      "loss": 0.0325,
      "step": 41250
    },
    {
      "epoch": 2.094268744606325,
      "grad_norm": 0.37093064188957214,
      "learning_rate": 6.038208369291166e-06,
      "loss": 0.0387,
      "step": 41255
    },
    {
      "epoch": 2.0945225645971877,
      "grad_norm": 0.28160449862480164,
      "learning_rate": 6.036516236018749e-06,
      "loss": 0.0339,
      "step": 41260
    },
    {
      "epoch": 2.0947763845880503,
      "grad_norm": 0.25263282656669617,
      "learning_rate": 6.034824102746333e-06,
      "loss": 0.029,
      "step": 41265
    },
    {
      "epoch": 2.0950302045789124,
      "grad_norm": 0.3500441610813141,
      "learning_rate": 6.033131969473917e-06,
      "loss": 0.0303,
      "step": 41270
    },
    {
      "epoch": 2.095284024569775,
      "grad_norm": 0.25448429584503174,
      "learning_rate": 6.0314398362014995e-06,
      "loss": 0.0351,
      "step": 41275
    },
    {
      "epoch": 2.0955378445606376,
      "grad_norm": 0.3285086154937744,
      "learning_rate": 6.029747702929083e-06,
      "loss": 0.0357,
      "step": 41280
    },
    {
      "epoch": 2.0957916645515002,
      "grad_norm": 0.20502544939517975,
      "learning_rate": 6.0280555696566665e-06,
      "loss": 0.0267,
      "step": 41285
    },
    {
      "epoch": 2.0960454845423624,
      "grad_norm": 0.30411866307258606,
      "learning_rate": 6.02636343638425e-06,
      "loss": 0.0342,
      "step": 41290
    },
    {
      "epoch": 2.096299304533225,
      "grad_norm": 0.3867494761943817,
      "learning_rate": 6.024671303111833e-06,
      "loss": 0.0381,
      "step": 41295
    },
    {
      "epoch": 2.0965531245240876,
      "grad_norm": 0.3713706433773041,
      "learning_rate": 6.022979169839417e-06,
      "loss": 0.032,
      "step": 41300
    },
    {
      "epoch": 2.09680694451495,
      "grad_norm": 0.3718228042125702,
      "learning_rate": 6.021287036567001e-06,
      "loss": 0.0355,
      "step": 41305
    },
    {
      "epoch": 2.0970607645058124,
      "grad_norm": 0.34682971239089966,
      "learning_rate": 6.019594903294584e-06,
      "loss": 0.0349,
      "step": 41310
    },
    {
      "epoch": 2.097314584496675,
      "grad_norm": 0.4789373278617859,
      "learning_rate": 6.017902770022167e-06,
      "loss": 0.0343,
      "step": 41315
    },
    {
      "epoch": 2.0975684044875376,
      "grad_norm": 0.3536965548992157,
      "learning_rate": 6.0162106367497505e-06,
      "loss": 0.0354,
      "step": 41320
    },
    {
      "epoch": 2.0978222244783997,
      "grad_norm": 2.8167803287506104,
      "learning_rate": 6.014518503477335e-06,
      "loss": 0.0352,
      "step": 41325
    },
    {
      "epoch": 2.0980760444692623,
      "grad_norm": 0.3835851550102234,
      "learning_rate": 6.0128263702049184e-06,
      "loss": 0.0314,
      "step": 41330
    },
    {
      "epoch": 2.098329864460125,
      "grad_norm": 0.5738430023193359,
      "learning_rate": 6.011134236932501e-06,
      "loss": 0.0293,
      "step": 41335
    },
    {
      "epoch": 2.0985836844509875,
      "grad_norm": 0.37173283100128174,
      "learning_rate": 6.009442103660085e-06,
      "loss": 0.0275,
      "step": 41340
    },
    {
      "epoch": 2.0988375044418497,
      "grad_norm": 0.4102475643157959,
      "learning_rate": 6.007749970387668e-06,
      "loss": 0.0387,
      "step": 41345
    },
    {
      "epoch": 2.0990913244327123,
      "grad_norm": 0.24621905386447906,
      "learning_rate": 6.006057837115251e-06,
      "loss": 0.0317,
      "step": 41350
    },
    {
      "epoch": 2.099345144423575,
      "grad_norm": 0.35181790590286255,
      "learning_rate": 6.004365703842835e-06,
      "loss": 0.0308,
      "step": 41355
    },
    {
      "epoch": 2.0995989644144375,
      "grad_norm": 0.5117323398590088,
      "learning_rate": 6.002673570570419e-06,
      "loss": 0.0285,
      "step": 41360
    },
    {
      "epoch": 2.0998527844052997,
      "grad_norm": 0.27516618371009827,
      "learning_rate": 6.000981437298002e-06,
      "loss": 0.0259,
      "step": 41365
    },
    {
      "epoch": 2.1001066043961623,
      "grad_norm": 0.369096577167511,
      "learning_rate": 5.999289304025585e-06,
      "loss": 0.0331,
      "step": 41370
    },
    {
      "epoch": 2.100360424387025,
      "grad_norm": 0.3615378737449646,
      "learning_rate": 5.997597170753169e-06,
      "loss": 0.0345,
      "step": 41375
    },
    {
      "epoch": 2.100614244377887,
      "grad_norm": 0.24888181686401367,
      "learning_rate": 5.995905037480752e-06,
      "loss": 0.0267,
      "step": 41380
    },
    {
      "epoch": 2.1008680643687496,
      "grad_norm": 0.5934954881668091,
      "learning_rate": 5.994212904208337e-06,
      "loss": 0.0307,
      "step": 41385
    },
    {
      "epoch": 2.1011218843596122,
      "grad_norm": 0.3345963656902313,
      "learning_rate": 5.992520770935919e-06,
      "loss": 0.0291,
      "step": 41390
    },
    {
      "epoch": 2.101375704350475,
      "grad_norm": 0.25391513109207153,
      "learning_rate": 5.990828637663503e-06,
      "loss": 0.0349,
      "step": 41395
    },
    {
      "epoch": 2.101629524341337,
      "grad_norm": 1.7726422548294067,
      "learning_rate": 5.989136504391086e-06,
      "loss": 0.0315,
      "step": 41400
    },
    {
      "epoch": 2.1018833443321996,
      "grad_norm": 0.37963372468948364,
      "learning_rate": 5.98744437111867e-06,
      "loss": 0.0361,
      "step": 41405
    },
    {
      "epoch": 2.102137164323062,
      "grad_norm": 0.2904397249221802,
      "learning_rate": 5.985752237846253e-06,
      "loss": 0.0329,
      "step": 41410
    },
    {
      "epoch": 2.1023909843139243,
      "grad_norm": 0.2872005105018616,
      "learning_rate": 5.984060104573837e-06,
      "loss": 0.0336,
      "step": 41415
    },
    {
      "epoch": 2.102644804304787,
      "grad_norm": 0.28669625520706177,
      "learning_rate": 5.9823679713014206e-06,
      "loss": 0.0311,
      "step": 41420
    },
    {
      "epoch": 2.1028986242956496,
      "grad_norm": 0.29623928666114807,
      "learning_rate": 5.980675838029004e-06,
      "loss": 0.0372,
      "step": 41425
    },
    {
      "epoch": 2.103152444286512,
      "grad_norm": 0.27423080801963806,
      "learning_rate": 5.978983704756587e-06,
      "loss": 0.0305,
      "step": 41430
    },
    {
      "epoch": 2.1034062642773743,
      "grad_norm": 0.32243502140045166,
      "learning_rate": 5.97729157148417e-06,
      "loss": 0.0364,
      "step": 41435
    },
    {
      "epoch": 2.103660084268237,
      "grad_norm": 0.2616148889064789,
      "learning_rate": 5.975599438211755e-06,
      "loss": 0.0269,
      "step": 41440
    },
    {
      "epoch": 2.1039139042590995,
      "grad_norm": 0.31910741329193115,
      "learning_rate": 5.9739073049393374e-06,
      "loss": 0.0337,
      "step": 41445
    },
    {
      "epoch": 2.104167724249962,
      "grad_norm": 0.21999460458755493,
      "learning_rate": 5.972215171666921e-06,
      "loss": 0.0311,
      "step": 41450
    },
    {
      "epoch": 2.1044215442408243,
      "grad_norm": 0.2903193533420563,
      "learning_rate": 5.9705230383945045e-06,
      "loss": 0.0343,
      "step": 41455
    },
    {
      "epoch": 2.104675364231687,
      "grad_norm": 0.3697342574596405,
      "learning_rate": 5.968830905122088e-06,
      "loss": 0.0359,
      "step": 41460
    },
    {
      "epoch": 2.1049291842225495,
      "grad_norm": 0.7846786975860596,
      "learning_rate": 5.967138771849671e-06,
      "loss": 0.0352,
      "step": 41465
    },
    {
      "epoch": 2.1051830042134116,
      "grad_norm": 0.3305487036705017,
      "learning_rate": 5.965446638577254e-06,
      "loss": 0.0393,
      "step": 41470
    },
    {
      "epoch": 2.1054368242042742,
      "grad_norm": 0.7854042053222656,
      "learning_rate": 5.963754505304839e-06,
      "loss": 0.0355,
      "step": 41475
    },
    {
      "epoch": 2.105690644195137,
      "grad_norm": 0.26362791657447815,
      "learning_rate": 5.962062372032422e-06,
      "loss": 0.0368,
      "step": 41480
    },
    {
      "epoch": 2.1059444641859995,
      "grad_norm": 0.348886102437973,
      "learning_rate": 5.960370238760005e-06,
      "loss": 0.0334,
      "step": 41485
    },
    {
      "epoch": 2.1061982841768616,
      "grad_norm": 0.538774311542511,
      "learning_rate": 5.9586781054875885e-06,
      "loss": 0.0375,
      "step": 41490
    },
    {
      "epoch": 2.106452104167724,
      "grad_norm": 0.31403154134750366,
      "learning_rate": 5.956985972215172e-06,
      "loss": 0.0318,
      "step": 41495
    },
    {
      "epoch": 2.106705924158587,
      "grad_norm": 0.3984089493751526,
      "learning_rate": 5.955293838942756e-06,
      "loss": 0.0364,
      "step": 41500
    },
    {
      "epoch": 2.1069597441494494,
      "grad_norm": 0.2857896089553833,
      "learning_rate": 5.953601705670339e-06,
      "loss": 0.0301,
      "step": 41505
    },
    {
      "epoch": 2.1072135641403116,
      "grad_norm": 0.4356675148010254,
      "learning_rate": 5.951909572397923e-06,
      "loss": 0.038,
      "step": 41510
    },
    {
      "epoch": 2.107467384131174,
      "grad_norm": 0.24989667534828186,
      "learning_rate": 5.950217439125506e-06,
      "loss": 0.036,
      "step": 41515
    },
    {
      "epoch": 2.107721204122037,
      "grad_norm": 0.38163310289382935,
      "learning_rate": 5.948525305853089e-06,
      "loss": 0.0369,
      "step": 41520
    },
    {
      "epoch": 2.107975024112899,
      "grad_norm": 0.2924136221408844,
      "learning_rate": 5.9468331725806724e-06,
      "loss": 0.0252,
      "step": 41525
    },
    {
      "epoch": 2.1082288441037615,
      "grad_norm": 0.6070223450660706,
      "learning_rate": 5.945141039308257e-06,
      "loss": 0.0415,
      "step": 41530
    },
    {
      "epoch": 2.108482664094624,
      "grad_norm": 0.3942844569683075,
      "learning_rate": 5.94344890603584e-06,
      "loss": 0.0342,
      "step": 41535
    },
    {
      "epoch": 2.1087364840854868,
      "grad_norm": 0.2800804376602173,
      "learning_rate": 5.941756772763423e-06,
      "loss": 0.0367,
      "step": 41540
    },
    {
      "epoch": 2.108990304076349,
      "grad_norm": 0.49201416969299316,
      "learning_rate": 5.940064639491007e-06,
      "loss": 0.0329,
      "step": 41545
    },
    {
      "epoch": 2.1092441240672115,
      "grad_norm": 0.35287120938301086,
      "learning_rate": 5.93837250621859e-06,
      "loss": 0.0311,
      "step": 41550
    },
    {
      "epoch": 2.109497944058074,
      "grad_norm": 0.5795586705207825,
      "learning_rate": 5.9366803729461746e-06,
      "loss": 0.0402,
      "step": 41555
    },
    {
      "epoch": 2.1097517640489363,
      "grad_norm": 0.3506717085838318,
      "learning_rate": 5.934988239673756e-06,
      "loss": 0.036,
      "step": 41560
    },
    {
      "epoch": 2.110005584039799,
      "grad_norm": 0.46954965591430664,
      "learning_rate": 5.933296106401341e-06,
      "loss": 0.0356,
      "step": 41565
    },
    {
      "epoch": 2.1102594040306615,
      "grad_norm": 0.3808066248893738,
      "learning_rate": 5.931603973128924e-06,
      "loss": 0.0353,
      "step": 41570
    },
    {
      "epoch": 2.110513224021524,
      "grad_norm": 0.3136650025844574,
      "learning_rate": 5.929911839856508e-06,
      "loss": 0.035,
      "step": 41575
    },
    {
      "epoch": 2.1107670440123862,
      "grad_norm": 0.3546306788921356,
      "learning_rate": 5.928219706584091e-06,
      "loss": 0.0314,
      "step": 41580
    },
    {
      "epoch": 2.111020864003249,
      "grad_norm": 0.2899860441684723,
      "learning_rate": 5.926527573311674e-06,
      "loss": 0.032,
      "step": 41585
    },
    {
      "epoch": 2.1112746839941114,
      "grad_norm": 0.3703233003616333,
      "learning_rate": 5.9248354400392585e-06,
      "loss": 0.0407,
      "step": 41590
    },
    {
      "epoch": 2.111528503984974,
      "grad_norm": 0.3205859661102295,
      "learning_rate": 5.923143306766841e-06,
      "loss": 0.0343,
      "step": 41595
    },
    {
      "epoch": 2.111782323975836,
      "grad_norm": 0.32839179039001465,
      "learning_rate": 5.921451173494425e-06,
      "loss": 0.0289,
      "step": 41600
    },
    {
      "epoch": 2.112036143966699,
      "grad_norm": 0.3662111461162567,
      "learning_rate": 5.919759040222008e-06,
      "loss": 0.0358,
      "step": 41605
    },
    {
      "epoch": 2.1122899639575614,
      "grad_norm": 0.2723025977611542,
      "learning_rate": 5.918066906949592e-06,
      "loss": 0.029,
      "step": 41610
    },
    {
      "epoch": 2.1125437839484236,
      "grad_norm": 0.4181530773639679,
      "learning_rate": 5.9163747736771746e-06,
      "loss": 0.0378,
      "step": 41615
    },
    {
      "epoch": 2.112797603939286,
      "grad_norm": 0.33553892374038696,
      "learning_rate": 5.914682640404759e-06,
      "loss": 0.0318,
      "step": 41620
    },
    {
      "epoch": 2.1130514239301488,
      "grad_norm": 0.3390558958053589,
      "learning_rate": 5.9129905071323425e-06,
      "loss": 0.0371,
      "step": 41625
    },
    {
      "epoch": 2.1133052439210114,
      "grad_norm": 0.4047679007053375,
      "learning_rate": 5.911298373859926e-06,
      "loss": 0.0403,
      "step": 41630
    },
    {
      "epoch": 2.1135590639118735,
      "grad_norm": 0.32319176197052,
      "learning_rate": 5.909606240587509e-06,
      "loss": 0.0321,
      "step": 41635
    },
    {
      "epoch": 2.113812883902736,
      "grad_norm": 0.3039274513721466,
      "learning_rate": 5.907914107315092e-06,
      "loss": 0.0327,
      "step": 41640
    },
    {
      "epoch": 2.1140667038935987,
      "grad_norm": 0.31280750036239624,
      "learning_rate": 5.906221974042677e-06,
      "loss": 0.0274,
      "step": 41645
    },
    {
      "epoch": 2.1143205238844613,
      "grad_norm": 0.4254741966724396,
      "learning_rate": 5.90452984077026e-06,
      "loss": 0.0325,
      "step": 41650
    },
    {
      "epoch": 2.1145743438753235,
      "grad_norm": 0.5669242739677429,
      "learning_rate": 5.902837707497843e-06,
      "loss": 0.0361,
      "step": 41655
    },
    {
      "epoch": 2.114828163866186,
      "grad_norm": 0.2737228572368622,
      "learning_rate": 5.9011455742254265e-06,
      "loss": 0.035,
      "step": 41660
    },
    {
      "epoch": 2.1150819838570487,
      "grad_norm": 0.35538411140441895,
      "learning_rate": 5.89945344095301e-06,
      "loss": 0.0387,
      "step": 41665
    },
    {
      "epoch": 2.115335803847911,
      "grad_norm": 0.23223480582237244,
      "learning_rate": 5.897761307680593e-06,
      "loss": 0.0377,
      "step": 41670
    },
    {
      "epoch": 2.1155896238387735,
      "grad_norm": 0.34843796491622925,
      "learning_rate": 5.896069174408176e-06,
      "loss": 0.0363,
      "step": 41675
    },
    {
      "epoch": 2.115843443829636,
      "grad_norm": 0.3045257329940796,
      "learning_rate": 5.894377041135761e-06,
      "loss": 0.0285,
      "step": 41680
    },
    {
      "epoch": 2.1160972638204987,
      "grad_norm": 0.6797808408737183,
      "learning_rate": 5.892684907863344e-06,
      "loss": 0.0337,
      "step": 41685
    },
    {
      "epoch": 2.116351083811361,
      "grad_norm": 0.8594637513160706,
      "learning_rate": 5.890992774590927e-06,
      "loss": 0.0361,
      "step": 41690
    },
    {
      "epoch": 2.1166049038022234,
      "grad_norm": 0.6106091141700745,
      "learning_rate": 5.88930064131851e-06,
      "loss": 0.0309,
      "step": 41695
    },
    {
      "epoch": 2.116858723793086,
      "grad_norm": 0.3803764879703522,
      "learning_rate": 5.887608508046094e-06,
      "loss": 0.0416,
      "step": 41700
    },
    {
      "epoch": 2.117112543783948,
      "grad_norm": 0.29211854934692383,
      "learning_rate": 5.885916374773678e-06,
      "loss": 0.0315,
      "step": 41705
    },
    {
      "epoch": 2.117366363774811,
      "grad_norm": 0.3258429765701294,
      "learning_rate": 5.884224241501261e-06,
      "loss": 0.0343,
      "step": 41710
    },
    {
      "epoch": 2.1176201837656734,
      "grad_norm": 0.35452401638031006,
      "learning_rate": 5.882532108228845e-06,
      "loss": 0.0339,
      "step": 41715
    },
    {
      "epoch": 2.117874003756536,
      "grad_norm": 0.3310200572013855,
      "learning_rate": 5.880839974956428e-06,
      "loss": 0.0315,
      "step": 41720
    },
    {
      "epoch": 2.118127823747398,
      "grad_norm": 0.43584612011909485,
      "learning_rate": 5.879147841684012e-06,
      "loss": 0.0368,
      "step": 41725
    },
    {
      "epoch": 2.1183816437382608,
      "grad_norm": 0.29797083139419556,
      "learning_rate": 5.877455708411594e-06,
      "loss": 0.031,
      "step": 41730
    },
    {
      "epoch": 2.1186354637291234,
      "grad_norm": 0.3942992389202118,
      "learning_rate": 5.875763575139179e-06,
      "loss": 0.0329,
      "step": 41735
    },
    {
      "epoch": 2.118889283719986,
      "grad_norm": 0.272582083940506,
      "learning_rate": 5.874071441866762e-06,
      "loss": 0.0319,
      "step": 41740
    },
    {
      "epoch": 2.119143103710848,
      "grad_norm": 0.3063197135925293,
      "learning_rate": 5.872379308594346e-06,
      "loss": 0.032,
      "step": 41745
    },
    {
      "epoch": 2.1193969237017107,
      "grad_norm": 0.24674548208713531,
      "learning_rate": 5.8706871753219286e-06,
      "loss": 0.0317,
      "step": 41750
    },
    {
      "epoch": 2.1196507436925733,
      "grad_norm": 0.38824862241744995,
      "learning_rate": 5.868995042049512e-06,
      "loss": 0.0332,
      "step": 41755
    },
    {
      "epoch": 2.1199045636834355,
      "grad_norm": 0.46050703525543213,
      "learning_rate": 5.867302908777096e-06,
      "loss": 0.0383,
      "step": 41760
    },
    {
      "epoch": 2.120158383674298,
      "grad_norm": 0.34193238615989685,
      "learning_rate": 5.865610775504678e-06,
      "loss": 0.0346,
      "step": 41765
    },
    {
      "epoch": 2.1204122036651607,
      "grad_norm": 0.5228917002677917,
      "learning_rate": 5.863918642232263e-06,
      "loss": 0.0437,
      "step": 41770
    },
    {
      "epoch": 2.1206660236560233,
      "grad_norm": 0.33700859546661377,
      "learning_rate": 5.862226508959846e-06,
      "loss": 0.0338,
      "step": 41775
    },
    {
      "epoch": 2.1209198436468855,
      "grad_norm": 1.1780332326889038,
      "learning_rate": 5.86053437568743e-06,
      "loss": 0.0357,
      "step": 41780
    },
    {
      "epoch": 2.121173663637748,
      "grad_norm": 0.34263256192207336,
      "learning_rate": 5.8588422424150125e-06,
      "loss": 0.0346,
      "step": 41785
    },
    {
      "epoch": 2.1214274836286107,
      "grad_norm": 0.35251888632774353,
      "learning_rate": 5.857150109142596e-06,
      "loss": 0.039,
      "step": 41790
    },
    {
      "epoch": 2.1216813036194733,
      "grad_norm": 0.5435646772384644,
      "learning_rate": 5.8554579758701805e-06,
      "loss": 0.0302,
      "step": 41795
    },
    {
      "epoch": 2.1219351236103354,
      "grad_norm": 0.2682788074016571,
      "learning_rate": 5.853765842597764e-06,
      "loss": 0.0342,
      "step": 41800
    },
    {
      "epoch": 2.122188943601198,
      "grad_norm": 0.33146414160728455,
      "learning_rate": 5.852073709325347e-06,
      "loss": 0.0371,
      "step": 41805
    },
    {
      "epoch": 2.1224427635920606,
      "grad_norm": 0.41672345995903015,
      "learning_rate": 5.85038157605293e-06,
      "loss": 0.0298,
      "step": 41810
    },
    {
      "epoch": 2.122696583582923,
      "grad_norm": 0.4548164904117584,
      "learning_rate": 5.848689442780514e-06,
      "loss": 0.0322,
      "step": 41815
    },
    {
      "epoch": 2.1229504035737854,
      "grad_norm": 0.2541426122188568,
      "learning_rate": 5.846997309508098e-06,
      "loss": 0.0278,
      "step": 41820
    },
    {
      "epoch": 2.123204223564648,
      "grad_norm": 0.313416451215744,
      "learning_rate": 5.845305176235681e-06,
      "loss": 0.0288,
      "step": 41825
    },
    {
      "epoch": 2.1234580435555106,
      "grad_norm": 1.0031988620758057,
      "learning_rate": 5.8436130429632644e-06,
      "loss": 0.0385,
      "step": 41830
    },
    {
      "epoch": 2.1237118635463728,
      "grad_norm": 0.3969992399215698,
      "learning_rate": 5.841920909690848e-06,
      "loss": 0.0382,
      "step": 41835
    },
    {
      "epoch": 2.1239656835372354,
      "grad_norm": 0.2746165096759796,
      "learning_rate": 5.840228776418431e-06,
      "loss": 0.0272,
      "step": 41840
    },
    {
      "epoch": 2.124219503528098,
      "grad_norm": 0.257400780916214,
      "learning_rate": 5.838536643146014e-06,
      "loss": 0.0276,
      "step": 41845
    },
    {
      "epoch": 2.1244733235189606,
      "grad_norm": 0.2490048110485077,
      "learning_rate": 5.836844509873599e-06,
      "loss": 0.0282,
      "step": 41850
    },
    {
      "epoch": 2.1247271435098227,
      "grad_norm": 0.45560091733932495,
      "learning_rate": 5.835152376601182e-06,
      "loss": 0.0403,
      "step": 41855
    },
    {
      "epoch": 2.1249809635006853,
      "grad_norm": 0.6095450520515442,
      "learning_rate": 5.833460243328765e-06,
      "loss": 0.0305,
      "step": 41860
    },
    {
      "epoch": 2.125234783491548,
      "grad_norm": 0.3171636164188385,
      "learning_rate": 5.831768110056348e-06,
      "loss": 0.0452,
      "step": 41865
    },
    {
      "epoch": 2.12548860348241,
      "grad_norm": 0.38941559195518494,
      "learning_rate": 5.830075976783932e-06,
      "loss": 0.037,
      "step": 41870
    },
    {
      "epoch": 2.1257424234732727,
      "grad_norm": 0.2944948077201843,
      "learning_rate": 5.8283838435115155e-06,
      "loss": 0.0305,
      "step": 41875
    },
    {
      "epoch": 2.1259962434641353,
      "grad_norm": 0.3468674123287201,
      "learning_rate": 5.826691710239098e-06,
      "loss": 0.0391,
      "step": 41880
    },
    {
      "epoch": 2.126250063454998,
      "grad_norm": 0.305381178855896,
      "learning_rate": 5.8249995769666826e-06,
      "loss": 0.0326,
      "step": 41885
    },
    {
      "epoch": 2.12650388344586,
      "grad_norm": 0.27479448914527893,
      "learning_rate": 5.823307443694266e-06,
      "loss": 0.0348,
      "step": 41890
    },
    {
      "epoch": 2.1267577034367227,
      "grad_norm": 0.2829623222351074,
      "learning_rate": 5.82161531042185e-06,
      "loss": 0.0357,
      "step": 41895
    },
    {
      "epoch": 2.1270115234275853,
      "grad_norm": 0.4399655759334564,
      "learning_rate": 5.819923177149432e-06,
      "loss": 0.0402,
      "step": 41900
    },
    {
      "epoch": 2.127265343418448,
      "grad_norm": 0.31989338994026184,
      "learning_rate": 5.818231043877016e-06,
      "loss": 0.0377,
      "step": 41905
    },
    {
      "epoch": 2.12751916340931,
      "grad_norm": 0.37366315722465515,
      "learning_rate": 5.8165389106046e-06,
      "loss": 0.0294,
      "step": 41910
    },
    {
      "epoch": 2.1277729834001726,
      "grad_norm": 0.2678751051425934,
      "learning_rate": 5.814846777332183e-06,
      "loss": 0.0261,
      "step": 41915
    },
    {
      "epoch": 2.1280268033910352,
      "grad_norm": 0.2653253674507141,
      "learning_rate": 5.8131546440597665e-06,
      "loss": 0.0299,
      "step": 41920
    },
    {
      "epoch": 2.1282806233818974,
      "grad_norm": 0.3034153878688812,
      "learning_rate": 5.81146251078735e-06,
      "loss": 0.0295,
      "step": 41925
    },
    {
      "epoch": 2.12853444337276,
      "grad_norm": 0.31852924823760986,
      "learning_rate": 5.809770377514934e-06,
      "loss": 0.0288,
      "step": 41930
    },
    {
      "epoch": 2.1287882633636226,
      "grad_norm": 0.4647199511528015,
      "learning_rate": 5.808078244242516e-06,
      "loss": 0.0321,
      "step": 41935
    },
    {
      "epoch": 2.129042083354485,
      "grad_norm": 0.22736875712871552,
      "learning_rate": 5.806386110970101e-06,
      "loss": 0.0317,
      "step": 41940
    },
    {
      "epoch": 2.1292959033453474,
      "grad_norm": 0.43213286995887756,
      "learning_rate": 5.804693977697684e-06,
      "loss": 0.0463,
      "step": 41945
    },
    {
      "epoch": 2.12954972333621,
      "grad_norm": 0.26801225543022156,
      "learning_rate": 5.803001844425268e-06,
      "loss": 0.0342,
      "step": 41950
    },
    {
      "epoch": 2.1298035433270726,
      "grad_norm": 0.3421599268913269,
      "learning_rate": 5.8013097111528505e-06,
      "loss": 0.0366,
      "step": 41955
    },
    {
      "epoch": 2.1300573633179347,
      "grad_norm": 0.5038048028945923,
      "learning_rate": 5.799617577880434e-06,
      "loss": 0.045,
      "step": 41960
    },
    {
      "epoch": 2.1303111833087973,
      "grad_norm": 0.3382954001426697,
      "learning_rate": 5.797925444608018e-06,
      "loss": 0.0389,
      "step": 41965
    },
    {
      "epoch": 2.13056500329966,
      "grad_norm": 0.3887106478214264,
      "learning_rate": 5.796233311335602e-06,
      "loss": 0.0386,
      "step": 41970
    },
    {
      "epoch": 2.1308188232905225,
      "grad_norm": 0.3497892916202545,
      "learning_rate": 5.794541178063185e-06,
      "loss": 0.0339,
      "step": 41975
    },
    {
      "epoch": 2.1310726432813847,
      "grad_norm": 0.49205145239830017,
      "learning_rate": 5.792849044790768e-06,
      "loss": 0.0336,
      "step": 41980
    },
    {
      "epoch": 2.1313264632722473,
      "grad_norm": 0.2803315818309784,
      "learning_rate": 5.791156911518352e-06,
      "loss": 0.0351,
      "step": 41985
    },
    {
      "epoch": 2.13158028326311,
      "grad_norm": 0.4156321883201599,
      "learning_rate": 5.789464778245935e-06,
      "loss": 0.0338,
      "step": 41990
    },
    {
      "epoch": 2.1318341032539725,
      "grad_norm": 0.29502180218696594,
      "learning_rate": 5.787772644973518e-06,
      "loss": 0.0357,
      "step": 41995
    },
    {
      "epoch": 2.1320879232448346,
      "grad_norm": 0.3212575614452362,
      "learning_rate": 5.786080511701102e-06,
      "loss": 0.0279,
      "step": 42000
    },
    {
      "epoch": 2.1323417432356973,
      "grad_norm": 0.40423792600631714,
      "learning_rate": 5.784388378428686e-06,
      "loss": 0.0396,
      "step": 42005
    },
    {
      "epoch": 2.13259556322656,
      "grad_norm": 0.31288251280784607,
      "learning_rate": 5.782696245156269e-06,
      "loss": 0.0329,
      "step": 42010
    },
    {
      "epoch": 2.132849383217422,
      "grad_norm": 0.26179739832878113,
      "learning_rate": 5.781004111883852e-06,
      "loss": 0.0369,
      "step": 42015
    },
    {
      "epoch": 2.1331032032082846,
      "grad_norm": 0.3306056559085846,
      "learning_rate": 5.779311978611436e-06,
      "loss": 0.0338,
      "step": 42020
    },
    {
      "epoch": 2.133357023199147,
      "grad_norm": 0.314375102519989,
      "learning_rate": 5.77761984533902e-06,
      "loss": 0.0268,
      "step": 42025
    },
    {
      "epoch": 2.13361084319001,
      "grad_norm": 0.2780863046646118,
      "learning_rate": 5.775927712066603e-06,
      "loss": 0.0318,
      "step": 42030
    },
    {
      "epoch": 2.133864663180872,
      "grad_norm": 0.8828794956207275,
      "learning_rate": 5.774235578794186e-06,
      "loss": 0.0348,
      "step": 42035
    },
    {
      "epoch": 2.1341184831717346,
      "grad_norm": 0.3718717098236084,
      "learning_rate": 5.77254344552177e-06,
      "loss": 0.0327,
      "step": 42040
    },
    {
      "epoch": 2.134372303162597,
      "grad_norm": 0.3083139657974243,
      "learning_rate": 5.7708513122493535e-06,
      "loss": 0.0315,
      "step": 42045
    },
    {
      "epoch": 2.13462612315346,
      "grad_norm": 0.39442887902259827,
      "learning_rate": 5.769159178976936e-06,
      "loss": 0.0329,
      "step": 42050
    },
    {
      "epoch": 2.134879943144322,
      "grad_norm": 0.31543081998825073,
      "learning_rate": 5.76746704570452e-06,
      "loss": 0.0283,
      "step": 42055
    },
    {
      "epoch": 2.1351337631351845,
      "grad_norm": 0.3375324606895447,
      "learning_rate": 5.765774912432104e-06,
      "loss": 0.0304,
      "step": 42060
    },
    {
      "epoch": 2.135387583126047,
      "grad_norm": 0.3673302233219147,
      "learning_rate": 5.764082779159688e-06,
      "loss": 0.0307,
      "step": 42065
    },
    {
      "epoch": 2.1356414031169093,
      "grad_norm": 0.3264317214488983,
      "learning_rate": 5.76239064588727e-06,
      "loss": 0.0348,
      "step": 42070
    },
    {
      "epoch": 2.135895223107772,
      "grad_norm": 0.3236130177974701,
      "learning_rate": 5.760698512614854e-06,
      "loss": 0.036,
      "step": 42075
    },
    {
      "epoch": 2.1361490430986345,
      "grad_norm": 0.46524056792259216,
      "learning_rate": 5.759006379342437e-06,
      "loss": 0.0344,
      "step": 42080
    },
    {
      "epoch": 2.136402863089497,
      "grad_norm": 0.3885677456855774,
      "learning_rate": 5.75731424607002e-06,
      "loss": 0.0339,
      "step": 42085
    },
    {
      "epoch": 2.1366566830803593,
      "grad_norm": 0.3115951418876648,
      "learning_rate": 5.7556221127976045e-06,
      "loss": 0.0305,
      "step": 42090
    },
    {
      "epoch": 2.136910503071222,
      "grad_norm": 0.28554677963256836,
      "learning_rate": 5.753929979525188e-06,
      "loss": 0.0333,
      "step": 42095
    },
    {
      "epoch": 2.1371643230620845,
      "grad_norm": 0.2696676254272461,
      "learning_rate": 5.752237846252772e-06,
      "loss": 0.0334,
      "step": 42100
    },
    {
      "epoch": 2.1374181430529466,
      "grad_norm": 0.2900834083557129,
      "learning_rate": 5.750545712980354e-06,
      "loss": 0.038,
      "step": 42105
    },
    {
      "epoch": 2.1376719630438092,
      "grad_norm": 0.7020983099937439,
      "learning_rate": 5.748853579707938e-06,
      "loss": 0.0291,
      "step": 42110
    },
    {
      "epoch": 2.137925783034672,
      "grad_norm": 0.2436659336090088,
      "learning_rate": 5.747161446435522e-06,
      "loss": 0.0376,
      "step": 42115
    },
    {
      "epoch": 2.1381796030255344,
      "grad_norm": 0.2586725652217865,
      "learning_rate": 5.745469313163106e-06,
      "loss": 0.0286,
      "step": 42120
    },
    {
      "epoch": 2.1384334230163966,
      "grad_norm": 0.414315789937973,
      "learning_rate": 5.7437771798906885e-06,
      "loss": 0.03,
      "step": 42125
    },
    {
      "epoch": 2.138687243007259,
      "grad_norm": 0.3736642897129059,
      "learning_rate": 5.742085046618272e-06,
      "loss": 0.0359,
      "step": 42130
    },
    {
      "epoch": 2.138941062998122,
      "grad_norm": 0.44996002316474915,
      "learning_rate": 5.7403929133458556e-06,
      "loss": 0.0366,
      "step": 42135
    },
    {
      "epoch": 2.1391948829889844,
      "grad_norm": 0.39824178814888,
      "learning_rate": 5.73870078007344e-06,
      "loss": 0.0373,
      "step": 42140
    },
    {
      "epoch": 2.1394487029798466,
      "grad_norm": 0.29749274253845215,
      "learning_rate": 5.737008646801022e-06,
      "loss": 0.0359,
      "step": 42145
    },
    {
      "epoch": 2.139702522970709,
      "grad_norm": 0.2442183643579483,
      "learning_rate": 5.735316513528606e-06,
      "loss": 0.0293,
      "step": 42150
    },
    {
      "epoch": 2.139956342961572,
      "grad_norm": 0.32782718539237976,
      "learning_rate": 5.73362438025619e-06,
      "loss": 0.0348,
      "step": 42155
    },
    {
      "epoch": 2.140210162952434,
      "grad_norm": 0.362077534198761,
      "learning_rate": 5.7319322469837724e-06,
      "loss": 0.0315,
      "step": 42160
    },
    {
      "epoch": 2.1404639829432965,
      "grad_norm": 0.4710308313369751,
      "learning_rate": 5.730240113711356e-06,
      "loss": 0.0364,
      "step": 42165
    },
    {
      "epoch": 2.140717802934159,
      "grad_norm": 0.37128207087516785,
      "learning_rate": 5.7285479804389395e-06,
      "loss": 0.0361,
      "step": 42170
    },
    {
      "epoch": 2.1409716229250217,
      "grad_norm": 0.2547430694103241,
      "learning_rate": 5.726855847166524e-06,
      "loss": 0.0332,
      "step": 42175
    },
    {
      "epoch": 2.141225442915884,
      "grad_norm": 0.2352115660905838,
      "learning_rate": 5.725163713894107e-06,
      "loss": 0.0246,
      "step": 42180
    },
    {
      "epoch": 2.1414792629067465,
      "grad_norm": 0.276828408241272,
      "learning_rate": 5.72347158062169e-06,
      "loss": 0.0316,
      "step": 42185
    },
    {
      "epoch": 2.141733082897609,
      "grad_norm": 0.29261428117752075,
      "learning_rate": 5.721779447349274e-06,
      "loss": 0.0292,
      "step": 42190
    },
    {
      "epoch": 2.1419869028884717,
      "grad_norm": 0.43690693378448486,
      "learning_rate": 5.720087314076857e-06,
      "loss": 0.0321,
      "step": 42195
    },
    {
      "epoch": 2.142240722879334,
      "grad_norm": 0.35671156644821167,
      "learning_rate": 5.71839518080444e-06,
      "loss": 0.0402,
      "step": 42200
    },
    {
      "epoch": 2.1424945428701965,
      "grad_norm": 0.32940271496772766,
      "learning_rate": 5.716703047532024e-06,
      "loss": 0.0322,
      "step": 42205
    },
    {
      "epoch": 2.142748362861059,
      "grad_norm": 0.3026338815689087,
      "learning_rate": 5.715010914259608e-06,
      "loss": 0.03,
      "step": 42210
    },
    {
      "epoch": 2.1430021828519212,
      "grad_norm": 0.29456955194473267,
      "learning_rate": 5.7133187809871914e-06,
      "loss": 0.0311,
      "step": 42215
    },
    {
      "epoch": 2.143256002842784,
      "grad_norm": 0.4180431067943573,
      "learning_rate": 5.711626647714774e-06,
      "loss": 0.0335,
      "step": 42220
    },
    {
      "epoch": 2.1435098228336464,
      "grad_norm": 0.25527551770210266,
      "learning_rate": 5.709934514442358e-06,
      "loss": 0.0317,
      "step": 42225
    },
    {
      "epoch": 2.143763642824509,
      "grad_norm": 0.38631191849708557,
      "learning_rate": 5.708242381169942e-06,
      "loss": 0.0316,
      "step": 42230
    },
    {
      "epoch": 2.144017462815371,
      "grad_norm": 0.4208427667617798,
      "learning_rate": 5.706550247897526e-06,
      "loss": 0.0349,
      "step": 42235
    },
    {
      "epoch": 2.144271282806234,
      "grad_norm": 0.3130198121070862,
      "learning_rate": 5.704858114625108e-06,
      "loss": 0.0315,
      "step": 42240
    },
    {
      "epoch": 2.1445251027970964,
      "grad_norm": 0.3700083792209625,
      "learning_rate": 5.703165981352692e-06,
      "loss": 0.0371,
      "step": 42245
    },
    {
      "epoch": 2.1447789227879586,
      "grad_norm": 0.3512057363986969,
      "learning_rate": 5.701473848080275e-06,
      "loss": 0.0342,
      "step": 42250
    },
    {
      "epoch": 2.145032742778821,
      "grad_norm": 0.34538981318473816,
      "learning_rate": 5.699781714807858e-06,
      "loss": 0.0325,
      "step": 42255
    },
    {
      "epoch": 2.1452865627696838,
      "grad_norm": 0.3043919801712036,
      "learning_rate": 5.698089581535442e-06,
      "loss": 0.0276,
      "step": 42260
    },
    {
      "epoch": 2.1455403827605464,
      "grad_norm": 0.30897659063339233,
      "learning_rate": 5.696397448263026e-06,
      "loss": 0.035,
      "step": 42265
    },
    {
      "epoch": 2.1457942027514085,
      "grad_norm": 0.3704676032066345,
      "learning_rate": 5.6947053149906096e-06,
      "loss": 0.0356,
      "step": 42270
    },
    {
      "epoch": 2.146048022742271,
      "grad_norm": 0.3414614498615265,
      "learning_rate": 5.693013181718192e-06,
      "loss": 0.0348,
      "step": 42275
    },
    {
      "epoch": 2.1463018427331337,
      "grad_norm": 0.3303713798522949,
      "learning_rate": 5.691321048445776e-06,
      "loss": 0.0324,
      "step": 42280
    },
    {
      "epoch": 2.1465556627239963,
      "grad_norm": 0.24423152208328247,
      "learning_rate": 5.689628915173359e-06,
      "loss": 0.0315,
      "step": 42285
    },
    {
      "epoch": 2.1468094827148585,
      "grad_norm": 0.43744367361068726,
      "learning_rate": 5.687936781900944e-06,
      "loss": 0.0395,
      "step": 42290
    },
    {
      "epoch": 2.147063302705721,
      "grad_norm": 0.3257816433906555,
      "learning_rate": 5.6862446486285264e-06,
      "loss": 0.0375,
      "step": 42295
    },
    {
      "epoch": 2.1473171226965837,
      "grad_norm": 0.4677280783653259,
      "learning_rate": 5.68455251535611e-06,
      "loss": 0.0381,
      "step": 42300
    },
    {
      "epoch": 2.147570942687446,
      "grad_norm": 0.5847957134246826,
      "learning_rate": 5.6828603820836935e-06,
      "loss": 0.0322,
      "step": 42305
    },
    {
      "epoch": 2.1478247626783085,
      "grad_norm": 0.34915176033973694,
      "learning_rate": 5.681168248811277e-06,
      "loss": 0.0299,
      "step": 42310
    },
    {
      "epoch": 2.148078582669171,
      "grad_norm": 0.29204973578453064,
      "learning_rate": 5.67947611553886e-06,
      "loss": 0.0323,
      "step": 42315
    },
    {
      "epoch": 2.1483324026600337,
      "grad_norm": 1.1503249406814575,
      "learning_rate": 5.677783982266444e-06,
      "loss": 0.0396,
      "step": 42320
    },
    {
      "epoch": 2.148586222650896,
      "grad_norm": 0.32928937673568726,
      "learning_rate": 5.676091848994028e-06,
      "loss": 0.0354,
      "step": 42325
    },
    {
      "epoch": 2.1488400426417584,
      "grad_norm": 0.28010591864585876,
      "learning_rate": 5.67439971572161e-06,
      "loss": 0.0268,
      "step": 42330
    },
    {
      "epoch": 2.149093862632621,
      "grad_norm": 0.22533346712589264,
      "learning_rate": 5.672707582449194e-06,
      "loss": 0.0278,
      "step": 42335
    },
    {
      "epoch": 2.1493476826234836,
      "grad_norm": 0.27991658449172974,
      "learning_rate": 5.6710154491767775e-06,
      "loss": 0.0332,
      "step": 42340
    },
    {
      "epoch": 2.149601502614346,
      "grad_norm": 0.2624974548816681,
      "learning_rate": 5.669323315904362e-06,
      "loss": 0.0388,
      "step": 42345
    },
    {
      "epoch": 2.1498553226052084,
      "grad_norm": 0.45382624864578247,
      "learning_rate": 5.667631182631944e-06,
      "loss": 0.027,
      "step": 42350
    },
    {
      "epoch": 2.150109142596071,
      "grad_norm": 0.27515923976898193,
      "learning_rate": 5.665939049359528e-06,
      "loss": 0.0313,
      "step": 42355
    },
    {
      "epoch": 2.150362962586933,
      "grad_norm": 0.3498297929763794,
      "learning_rate": 5.664246916087112e-06,
      "loss": 0.0286,
      "step": 42360
    },
    {
      "epoch": 2.1506167825777958,
      "grad_norm": 0.32744863629341125,
      "learning_rate": 5.662554782814695e-06,
      "loss": 0.0281,
      "step": 42365
    },
    {
      "epoch": 2.1508706025686584,
      "grad_norm": 0.4372187852859497,
      "learning_rate": 5.660862649542278e-06,
      "loss": 0.0405,
      "step": 42370
    },
    {
      "epoch": 2.151124422559521,
      "grad_norm": 0.23590414226055145,
      "learning_rate": 5.6591705162698615e-06,
      "loss": 0.0291,
      "step": 42375
    },
    {
      "epoch": 2.151378242550383,
      "grad_norm": 0.33515384793281555,
      "learning_rate": 5.657478382997446e-06,
      "loss": 0.0315,
      "step": 42380
    },
    {
      "epoch": 2.1516320625412457,
      "grad_norm": 0.2991335093975067,
      "learning_rate": 5.655786249725029e-06,
      "loss": 0.0414,
      "step": 42385
    },
    {
      "epoch": 2.1518858825321083,
      "grad_norm": 0.40804606676101685,
      "learning_rate": 5.654094116452612e-06,
      "loss": 0.0335,
      "step": 42390
    },
    {
      "epoch": 2.1521397025229705,
      "grad_norm": 0.42320457100868225,
      "learning_rate": 5.652401983180196e-06,
      "loss": 0.0363,
      "step": 42395
    },
    {
      "epoch": 2.152393522513833,
      "grad_norm": 0.5444016456604004,
      "learning_rate": 5.650709849907779e-06,
      "loss": 0.029,
      "step": 42400
    },
    {
      "epoch": 2.1526473425046957,
      "grad_norm": 0.3651138246059418,
      "learning_rate": 5.649017716635362e-06,
      "loss": 0.0399,
      "step": 42405
    },
    {
      "epoch": 2.1529011624955583,
      "grad_norm": 0.30745598673820496,
      "learning_rate": 5.647325583362946e-06,
      "loss": 0.0348,
      "step": 42410
    },
    {
      "epoch": 2.1531549824864205,
      "grad_norm": 0.2215747833251953,
      "learning_rate": 5.64563345009053e-06,
      "loss": 0.0342,
      "step": 42415
    },
    {
      "epoch": 2.153408802477283,
      "grad_norm": 0.47790786623954773,
      "learning_rate": 5.643941316818113e-06,
      "loss": 0.0371,
      "step": 42420
    },
    {
      "epoch": 2.1536626224681457,
      "grad_norm": 0.4156583547592163,
      "learning_rate": 5.642249183545696e-06,
      "loss": 0.0326,
      "step": 42425
    },
    {
      "epoch": 2.1539164424590083,
      "grad_norm": 0.38761693239212036,
      "learning_rate": 5.64055705027328e-06,
      "loss": 0.0358,
      "step": 42430
    },
    {
      "epoch": 2.1541702624498704,
      "grad_norm": 0.3796803057193756,
      "learning_rate": 5.638864917000864e-06,
      "loss": 0.0363,
      "step": 42435
    },
    {
      "epoch": 2.154424082440733,
      "grad_norm": 0.3898187279701233,
      "learning_rate": 5.6371727837284475e-06,
      "loss": 0.0339,
      "step": 42440
    },
    {
      "epoch": 2.1546779024315956,
      "grad_norm": 0.31213703751564026,
      "learning_rate": 5.63548065045603e-06,
      "loss": 0.039,
      "step": 42445
    },
    {
      "epoch": 2.154931722422458,
      "grad_norm": 0.2264726459980011,
      "learning_rate": 5.633788517183614e-06,
      "loss": 0.0302,
      "step": 42450
    },
    {
      "epoch": 2.1551855424133204,
      "grad_norm": 0.24250595271587372,
      "learning_rate": 5.632096383911197e-06,
      "loss": 0.0376,
      "step": 42455
    },
    {
      "epoch": 2.155439362404183,
      "grad_norm": 0.32829833030700684,
      "learning_rate": 5.630404250638781e-06,
      "loss": 0.0352,
      "step": 42460
    },
    {
      "epoch": 2.1556931823950456,
      "grad_norm": 0.3167964220046997,
      "learning_rate": 5.6287121173663636e-06,
      "loss": 0.0368,
      "step": 42465
    },
    {
      "epoch": 2.1559470023859078,
      "grad_norm": 0.3327024579048157,
      "learning_rate": 5.627019984093948e-06,
      "loss": 0.0368,
      "step": 42470
    },
    {
      "epoch": 2.1562008223767704,
      "grad_norm": 0.337005078792572,
      "learning_rate": 5.6253278508215315e-06,
      "loss": 0.0324,
      "step": 42475
    },
    {
      "epoch": 2.156454642367633,
      "grad_norm": 0.34376224875450134,
      "learning_rate": 5.623635717549114e-06,
      "loss": 0.0346,
      "step": 42480
    },
    {
      "epoch": 2.1567084623584956,
      "grad_norm": 0.41502124071121216,
      "learning_rate": 5.621943584276698e-06,
      "loss": 0.0333,
      "step": 42485
    },
    {
      "epoch": 2.1569622823493577,
      "grad_norm": 0.34838420152664185,
      "learning_rate": 5.620251451004281e-06,
      "loss": 0.0334,
      "step": 42490
    },
    {
      "epoch": 2.1572161023402203,
      "grad_norm": 0.36978450417518616,
      "learning_rate": 5.618559317731866e-06,
      "loss": 0.0338,
      "step": 42495
    },
    {
      "epoch": 2.157469922331083,
      "grad_norm": 0.2665274441242218,
      "learning_rate": 5.616867184459448e-06,
      "loss": 0.0275,
      "step": 42500
    },
    {
      "epoch": 2.157723742321945,
      "grad_norm": 0.389095664024353,
      "learning_rate": 5.615175051187032e-06,
      "loss": 0.0333,
      "step": 42505
    },
    {
      "epoch": 2.1579775623128077,
      "grad_norm": 0.2917749881744385,
      "learning_rate": 5.6134829179146155e-06,
      "loss": 0.0332,
      "step": 42510
    },
    {
      "epoch": 2.1582313823036703,
      "grad_norm": 0.3830176889896393,
      "learning_rate": 5.611790784642199e-06,
      "loss": 0.034,
      "step": 42515
    },
    {
      "epoch": 2.158485202294533,
      "grad_norm": 0.3689527213573456,
      "learning_rate": 5.610098651369782e-06,
      "loss": 0.0304,
      "step": 42520
    },
    {
      "epoch": 2.158739022285395,
      "grad_norm": 0.517031729221344,
      "learning_rate": 5.608406518097366e-06,
      "loss": 0.0376,
      "step": 42525
    },
    {
      "epoch": 2.1589928422762577,
      "grad_norm": 0.3284299671649933,
      "learning_rate": 5.60671438482495e-06,
      "loss": 0.039,
      "step": 42530
    },
    {
      "epoch": 2.1592466622671203,
      "grad_norm": 0.3636420667171478,
      "learning_rate": 5.605022251552533e-06,
      "loss": 0.0334,
      "step": 42535
    },
    {
      "epoch": 2.1595004822579824,
      "grad_norm": 0.5183312892913818,
      "learning_rate": 5.603330118280116e-06,
      "loss": 0.0377,
      "step": 42540
    },
    {
      "epoch": 2.159754302248845,
      "grad_norm": 0.30915188789367676,
      "learning_rate": 5.6016379850076994e-06,
      "loss": 0.0271,
      "step": 42545
    },
    {
      "epoch": 2.1600081222397076,
      "grad_norm": 0.3173023760318756,
      "learning_rate": 5.599945851735283e-06,
      "loss": 0.0332,
      "step": 42550
    },
    {
      "epoch": 2.16026194223057,
      "grad_norm": 0.31991812586784363,
      "learning_rate": 5.598253718462867e-06,
      "loss": 0.0442,
      "step": 42555
    },
    {
      "epoch": 2.1605157622214324,
      "grad_norm": 0.7277874946594238,
      "learning_rate": 5.59656158519045e-06,
      "loss": 0.0327,
      "step": 42560
    },
    {
      "epoch": 2.160769582212295,
      "grad_norm": 0.23353658616542816,
      "learning_rate": 5.594869451918034e-06,
      "loss": 0.035,
      "step": 42565
    },
    {
      "epoch": 2.1610234022031576,
      "grad_norm": 0.5625307559967041,
      "learning_rate": 5.593177318645617e-06,
      "loss": 0.0313,
      "step": 42570
    },
    {
      "epoch": 2.16127722219402,
      "grad_norm": 0.41298919916152954,
      "learning_rate": 5.5914851853732e-06,
      "loss": 0.0377,
      "step": 42575
    },
    {
      "epoch": 2.1615310421848823,
      "grad_norm": 0.26099902391433716,
      "learning_rate": 5.589793052100783e-06,
      "loss": 0.0321,
      "step": 42580
    },
    {
      "epoch": 2.161784862175745,
      "grad_norm": 0.6761044263839722,
      "learning_rate": 5.588100918828368e-06,
      "loss": 0.0325,
      "step": 42585
    },
    {
      "epoch": 2.1620386821666076,
      "grad_norm": 0.3050651550292969,
      "learning_rate": 5.586408785555951e-06,
      "loss": 0.0327,
      "step": 42590
    },
    {
      "epoch": 2.16229250215747,
      "grad_norm": 0.4124506115913391,
      "learning_rate": 5.584716652283534e-06,
      "loss": 0.0396,
      "step": 42595
    },
    {
      "epoch": 2.1625463221483323,
      "grad_norm": 0.5697929263114929,
      "learning_rate": 5.583024519011118e-06,
      "loss": 0.0379,
      "step": 42600
    },
    {
      "epoch": 2.162800142139195,
      "grad_norm": 0.471778929233551,
      "learning_rate": 5.581332385738701e-06,
      "loss": 0.0291,
      "step": 42605
    },
    {
      "epoch": 2.1630539621300575,
      "grad_norm": 0.27507635951042175,
      "learning_rate": 5.5796402524662855e-06,
      "loss": 0.0331,
      "step": 42610
    },
    {
      "epoch": 2.1633077821209197,
      "grad_norm": 0.4065435230731964,
      "learning_rate": 5.577948119193868e-06,
      "loss": 0.0319,
      "step": 42615
    },
    {
      "epoch": 2.1635616021117823,
      "grad_norm": 0.32734885811805725,
      "learning_rate": 5.576255985921452e-06,
      "loss": 0.0301,
      "step": 42620
    },
    {
      "epoch": 2.163815422102645,
      "grad_norm": 0.27780577540397644,
      "learning_rate": 5.574563852649035e-06,
      "loss": 0.0255,
      "step": 42625
    },
    {
      "epoch": 2.1640692420935075,
      "grad_norm": 0.37050920724868774,
      "learning_rate": 5.572871719376619e-06,
      "loss": 0.0355,
      "step": 42630
    },
    {
      "epoch": 2.1643230620843696,
      "grad_norm": 0.3847156763076782,
      "learning_rate": 5.5711795861042015e-06,
      "loss": 0.0324,
      "step": 42635
    },
    {
      "epoch": 2.1645768820752322,
      "grad_norm": 0.4115941822528839,
      "learning_rate": 5.569487452831785e-06,
      "loss": 0.0343,
      "step": 42640
    },
    {
      "epoch": 2.164830702066095,
      "grad_norm": 0.3248206377029419,
      "learning_rate": 5.5677953195593695e-06,
      "loss": 0.0293,
      "step": 42645
    },
    {
      "epoch": 2.165084522056957,
      "grad_norm": 0.5444241762161255,
      "learning_rate": 5.566103186286952e-06,
      "loss": 0.0392,
      "step": 42650
    },
    {
      "epoch": 2.1653383420478196,
      "grad_norm": 0.4807383716106415,
      "learning_rate": 5.564411053014536e-06,
      "loss": 0.0359,
      "step": 42655
    },
    {
      "epoch": 2.165592162038682,
      "grad_norm": 0.2330743968486786,
      "learning_rate": 5.562718919742119e-06,
      "loss": 0.0249,
      "step": 42660
    },
    {
      "epoch": 2.165845982029545,
      "grad_norm": 0.29443061351776123,
      "learning_rate": 5.561026786469703e-06,
      "loss": 0.0321,
      "step": 42665
    },
    {
      "epoch": 2.166099802020407,
      "grad_norm": 0.42073678970336914,
      "learning_rate": 5.5593346531972855e-06,
      "loss": 0.0345,
      "step": 42670
    },
    {
      "epoch": 2.1663536220112696,
      "grad_norm": 0.2619229853153229,
      "learning_rate": 5.55764251992487e-06,
      "loss": 0.0242,
      "step": 42675
    },
    {
      "epoch": 2.166607442002132,
      "grad_norm": 0.37656205892562866,
      "learning_rate": 5.5559503866524534e-06,
      "loss": 0.0338,
      "step": 42680
    },
    {
      "epoch": 2.1668612619929943,
      "grad_norm": 0.3815825283527374,
      "learning_rate": 5.554258253380037e-06,
      "loss": 0.0316,
      "step": 42685
    },
    {
      "epoch": 2.167115081983857,
      "grad_norm": 0.24443581700325012,
      "learning_rate": 5.55256612010762e-06,
      "loss": 0.0399,
      "step": 42690
    },
    {
      "epoch": 2.1673689019747195,
      "grad_norm": 0.6075119376182556,
      "learning_rate": 5.550873986835203e-06,
      "loss": 0.0304,
      "step": 42695
    },
    {
      "epoch": 2.167622721965582,
      "grad_norm": 0.4012862741947174,
      "learning_rate": 5.549181853562788e-06,
      "loss": 0.0352,
      "step": 42700
    },
    {
      "epoch": 2.1678765419564443,
      "grad_norm": 0.5233157277107239,
      "learning_rate": 5.547489720290371e-06,
      "loss": 0.0303,
      "step": 42705
    },
    {
      "epoch": 2.168130361947307,
      "grad_norm": 0.23460358381271362,
      "learning_rate": 5.545797587017954e-06,
      "loss": 0.0379,
      "step": 42710
    },
    {
      "epoch": 2.1683841819381695,
      "grad_norm": 0.3687634766101837,
      "learning_rate": 5.544105453745537e-06,
      "loss": 0.035,
      "step": 42715
    },
    {
      "epoch": 2.168638001929032,
      "grad_norm": 0.49674105644226074,
      "learning_rate": 5.542413320473121e-06,
      "loss": 0.0305,
      "step": 42720
    },
    {
      "epoch": 2.1688918219198943,
      "grad_norm": 0.2974553406238556,
      "learning_rate": 5.540721187200704e-06,
      "loss": 0.0301,
      "step": 42725
    },
    {
      "epoch": 2.169145641910757,
      "grad_norm": 0.3256213068962097,
      "learning_rate": 5.539029053928287e-06,
      "loss": 0.0383,
      "step": 42730
    },
    {
      "epoch": 2.1693994619016195,
      "grad_norm": 0.2907634377479553,
      "learning_rate": 5.537336920655872e-06,
      "loss": 0.038,
      "step": 42735
    },
    {
      "epoch": 2.169653281892482,
      "grad_norm": 0.6651555895805359,
      "learning_rate": 5.535644787383455e-06,
      "loss": 0.0322,
      "step": 42740
    },
    {
      "epoch": 2.1699071018833442,
      "grad_norm": 0.3532056212425232,
      "learning_rate": 5.533952654111038e-06,
      "loss": 0.0327,
      "step": 42745
    },
    {
      "epoch": 2.170160921874207,
      "grad_norm": 0.2968956232070923,
      "learning_rate": 5.532260520838621e-06,
      "loss": 0.0339,
      "step": 42750
    },
    {
      "epoch": 2.1704147418650694,
      "grad_norm": 0.40109357237815857,
      "learning_rate": 5.530568387566205e-06,
      "loss": 0.0283,
      "step": 42755
    },
    {
      "epoch": 2.1706685618559316,
      "grad_norm": 0.44607171416282654,
      "learning_rate": 5.528876254293789e-06,
      "loss": 0.0346,
      "step": 42760
    },
    {
      "epoch": 2.170922381846794,
      "grad_norm": 0.4335692524909973,
      "learning_rate": 5.527184121021372e-06,
      "loss": 0.0409,
      "step": 42765
    },
    {
      "epoch": 2.171176201837657,
      "grad_norm": 0.40640366077423096,
      "learning_rate": 5.5254919877489556e-06,
      "loss": 0.0264,
      "step": 42770
    },
    {
      "epoch": 2.1714300218285194,
      "grad_norm": 0.2704562544822693,
      "learning_rate": 5.523799854476539e-06,
      "loss": 0.0287,
      "step": 42775
    },
    {
      "epoch": 2.1716838418193816,
      "grad_norm": 0.2660040855407715,
      "learning_rate": 5.522107721204123e-06,
      "loss": 0.0326,
      "step": 42780
    },
    {
      "epoch": 2.171937661810244,
      "grad_norm": 0.3302189111709595,
      "learning_rate": 5.520415587931705e-06,
      "loss": 0.044,
      "step": 42785
    },
    {
      "epoch": 2.1721914818011068,
      "grad_norm": 0.300412118434906,
      "learning_rate": 5.51872345465929e-06,
      "loss": 0.0349,
      "step": 42790
    },
    {
      "epoch": 2.172445301791969,
      "grad_norm": 0.26021671295166016,
      "learning_rate": 5.517031321386873e-06,
      "loss": 0.0375,
      "step": 42795
    },
    {
      "epoch": 2.1726991217828315,
      "grad_norm": 0.3481993079185486,
      "learning_rate": 5.515339188114457e-06,
      "loss": 0.0339,
      "step": 42800
    },
    {
      "epoch": 2.172952941773694,
      "grad_norm": 0.37204688787460327,
      "learning_rate": 5.5136470548420395e-06,
      "loss": 0.0304,
      "step": 42805
    },
    {
      "epoch": 2.1732067617645567,
      "grad_norm": 0.453274667263031,
      "learning_rate": 5.511954921569623e-06,
      "loss": 0.0302,
      "step": 42810
    },
    {
      "epoch": 2.173460581755419,
      "grad_norm": 0.4201031029224396,
      "learning_rate": 5.5102627882972075e-06,
      "loss": 0.0379,
      "step": 42815
    },
    {
      "epoch": 2.1737144017462815,
      "grad_norm": 0.3628084063529968,
      "learning_rate": 5.508570655024789e-06,
      "loss": 0.0356,
      "step": 42820
    },
    {
      "epoch": 2.173968221737144,
      "grad_norm": 0.4329802691936493,
      "learning_rate": 5.506878521752374e-06,
      "loss": 0.0347,
      "step": 42825
    },
    {
      "epoch": 2.1742220417280067,
      "grad_norm": 0.32375532388687134,
      "learning_rate": 5.505186388479957e-06,
      "loss": 0.039,
      "step": 42830
    },
    {
      "epoch": 2.174475861718869,
      "grad_norm": 0.7113686800003052,
      "learning_rate": 5.503494255207541e-06,
      "loss": 0.0349,
      "step": 42835
    },
    {
      "epoch": 2.1747296817097315,
      "grad_norm": 0.3115473985671997,
      "learning_rate": 5.5018021219351235e-06,
      "loss": 0.0399,
      "step": 42840
    },
    {
      "epoch": 2.174983501700594,
      "grad_norm": 0.7726762890815735,
      "learning_rate": 5.500109988662707e-06,
      "loss": 0.0312,
      "step": 42845
    },
    {
      "epoch": 2.1752373216914562,
      "grad_norm": 0.34752291440963745,
      "learning_rate": 5.498417855390291e-06,
      "loss": 0.037,
      "step": 42850
    },
    {
      "epoch": 2.175491141682319,
      "grad_norm": 0.8168660998344421,
      "learning_rate": 5.496725722117875e-06,
      "loss": 0.029,
      "step": 42855
    },
    {
      "epoch": 2.1757449616731814,
      "grad_norm": 0.27437952160835266,
      "learning_rate": 5.495033588845458e-06,
      "loss": 0.0328,
      "step": 42860
    },
    {
      "epoch": 2.175998781664044,
      "grad_norm": 0.30334383249282837,
      "learning_rate": 5.493341455573041e-06,
      "loss": 0.0389,
      "step": 42865
    },
    {
      "epoch": 2.176252601654906,
      "grad_norm": 0.2583600878715515,
      "learning_rate": 5.491649322300625e-06,
      "loss": 0.0318,
      "step": 42870
    },
    {
      "epoch": 2.176506421645769,
      "grad_norm": 0.42259642481803894,
      "learning_rate": 5.489957189028209e-06,
      "loss": 0.0279,
      "step": 42875
    },
    {
      "epoch": 2.1767602416366314,
      "grad_norm": 1.2351564168930054,
      "learning_rate": 5.488265055755792e-06,
      "loss": 0.0346,
      "step": 42880
    },
    {
      "epoch": 2.177014061627494,
      "grad_norm": 0.36466923356056213,
      "learning_rate": 5.486572922483375e-06,
      "loss": 0.0321,
      "step": 42885
    },
    {
      "epoch": 2.177267881618356,
      "grad_norm": 0.3623839020729065,
      "learning_rate": 5.484880789210959e-06,
      "loss": 0.0354,
      "step": 42890
    },
    {
      "epoch": 2.1775217016092188,
      "grad_norm": 0.3960568308830261,
      "learning_rate": 5.483188655938542e-06,
      "loss": 0.0315,
      "step": 42895
    },
    {
      "epoch": 2.1777755216000814,
      "grad_norm": 0.23207826912403107,
      "learning_rate": 5.481496522666125e-06,
      "loss": 0.0332,
      "step": 42900
    },
    {
      "epoch": 2.1780293415909435,
      "grad_norm": 0.34986627101898193,
      "learning_rate": 5.4798043893937096e-06,
      "loss": 0.0365,
      "step": 42905
    },
    {
      "epoch": 2.178283161581806,
      "grad_norm": 0.2791183888912201,
      "learning_rate": 5.478112256121293e-06,
      "loss": 0.0346,
      "step": 42910
    },
    {
      "epoch": 2.1785369815726687,
      "grad_norm": 0.4135153889656067,
      "learning_rate": 5.476420122848876e-06,
      "loss": 0.0387,
      "step": 42915
    },
    {
      "epoch": 2.1787908015635313,
      "grad_norm": 0.4401041567325592,
      "learning_rate": 5.474727989576459e-06,
      "loss": 0.0348,
      "step": 42920
    },
    {
      "epoch": 2.1790446215543935,
      "grad_norm": 0.25939372181892395,
      "learning_rate": 5.473035856304043e-06,
      "loss": 0.0341,
      "step": 42925
    },
    {
      "epoch": 2.179298441545256,
      "grad_norm": 0.8912258744239807,
      "learning_rate": 5.471343723031627e-06,
      "loss": 0.0365,
      "step": 42930
    },
    {
      "epoch": 2.1795522615361187,
      "grad_norm": 0.40229907631874084,
      "learning_rate": 5.469651589759209e-06,
      "loss": 0.0332,
      "step": 42935
    },
    {
      "epoch": 2.179806081526981,
      "grad_norm": 0.2511097192764282,
      "learning_rate": 5.4679594564867935e-06,
      "loss": 0.0314,
      "step": 42940
    },
    {
      "epoch": 2.1800599015178435,
      "grad_norm": 0.25943753123283386,
      "learning_rate": 5.466267323214377e-06,
      "loss": 0.0307,
      "step": 42945
    },
    {
      "epoch": 2.180313721508706,
      "grad_norm": 0.3757547438144684,
      "learning_rate": 5.464575189941961e-06,
      "loss": 0.0378,
      "step": 42950
    },
    {
      "epoch": 2.1805675414995687,
      "grad_norm": 0.3526191711425781,
      "learning_rate": 5.462883056669543e-06,
      "loss": 0.0337,
      "step": 42955
    },
    {
      "epoch": 2.180821361490431,
      "grad_norm": 0.2390231043100357,
      "learning_rate": 5.461190923397127e-06,
      "loss": 0.0336,
      "step": 42960
    },
    {
      "epoch": 2.1810751814812934,
      "grad_norm": 0.3768102526664734,
      "learning_rate": 5.459498790124711e-06,
      "loss": 0.04,
      "step": 42965
    },
    {
      "epoch": 2.181329001472156,
      "grad_norm": 0.345241516828537,
      "learning_rate": 5.457806656852294e-06,
      "loss": 0.0326,
      "step": 42970
    },
    {
      "epoch": 2.1815828214630186,
      "grad_norm": 0.3643832504749298,
      "learning_rate": 5.4561145235798775e-06,
      "loss": 0.0364,
      "step": 42975
    },
    {
      "epoch": 2.181836641453881,
      "grad_norm": 0.3406466245651245,
      "learning_rate": 5.454422390307461e-06,
      "loss": 0.0287,
      "step": 42980
    },
    {
      "epoch": 2.1820904614447434,
      "grad_norm": 0.46112820506095886,
      "learning_rate": 5.452730257035045e-06,
      "loss": 0.0402,
      "step": 42985
    },
    {
      "epoch": 2.182344281435606,
      "grad_norm": 0.24325571954250336,
      "learning_rate": 5.451038123762627e-06,
      "loss": 0.0299,
      "step": 42990
    },
    {
      "epoch": 2.182598101426468,
      "grad_norm": 0.3688530921936035,
      "learning_rate": 5.449345990490212e-06,
      "loss": 0.0279,
      "step": 42995
    },
    {
      "epoch": 2.1828519214173308,
      "grad_norm": 0.29081302881240845,
      "learning_rate": 5.447653857217795e-06,
      "loss": 0.0299,
      "step": 43000
    },
    {
      "epoch": 2.1831057414081934,
      "grad_norm": 0.45404350757598877,
      "learning_rate": 5.445961723945379e-06,
      "loss": 0.0308,
      "step": 43005
    },
    {
      "epoch": 2.183359561399056,
      "grad_norm": 0.41532981395721436,
      "learning_rate": 5.4442695906729615e-06,
      "loss": 0.0366,
      "step": 43010
    },
    {
      "epoch": 2.183613381389918,
      "grad_norm": 0.38594257831573486,
      "learning_rate": 5.442577457400545e-06,
      "loss": 0.0334,
      "step": 43015
    },
    {
      "epoch": 2.1838672013807807,
      "grad_norm": 0.2794020473957062,
      "learning_rate": 5.440885324128129e-06,
      "loss": 0.0337,
      "step": 43020
    },
    {
      "epoch": 2.1841210213716433,
      "grad_norm": 0.26633644104003906,
      "learning_rate": 5.439193190855713e-06,
      "loss": 0.0313,
      "step": 43025
    },
    {
      "epoch": 2.184374841362506,
      "grad_norm": 0.41068437695503235,
      "learning_rate": 5.437501057583296e-06,
      "loss": 0.0329,
      "step": 43030
    },
    {
      "epoch": 2.184628661353368,
      "grad_norm": 0.4237484931945801,
      "learning_rate": 5.435808924310879e-06,
      "loss": 0.0361,
      "step": 43035
    },
    {
      "epoch": 2.1848824813442307,
      "grad_norm": 0.2985211908817291,
      "learning_rate": 5.434116791038463e-06,
      "loss": 0.0355,
      "step": 43040
    },
    {
      "epoch": 2.1851363013350933,
      "grad_norm": 0.2558479309082031,
      "learning_rate": 5.432424657766045e-06,
      "loss": 0.0307,
      "step": 43045
    },
    {
      "epoch": 2.1853901213259554,
      "grad_norm": 0.2939911484718323,
      "learning_rate": 5.430732524493629e-06,
      "loss": 0.0323,
      "step": 43050
    },
    {
      "epoch": 2.185643941316818,
      "grad_norm": 0.4152364730834961,
      "learning_rate": 5.429040391221213e-06,
      "loss": 0.0306,
      "step": 43055
    },
    {
      "epoch": 2.1858977613076807,
      "grad_norm": 0.3391212224960327,
      "learning_rate": 5.427348257948797e-06,
      "loss": 0.0375,
      "step": 43060
    },
    {
      "epoch": 2.1861515812985433,
      "grad_norm": 0.3118323087692261,
      "learning_rate": 5.42565612467638e-06,
      "loss": 0.0341,
      "step": 43065
    },
    {
      "epoch": 2.1864054012894054,
      "grad_norm": 0.278476744890213,
      "learning_rate": 5.423963991403963e-06,
      "loss": 0.0301,
      "step": 43070
    },
    {
      "epoch": 2.186659221280268,
      "grad_norm": 0.31256920099258423,
      "learning_rate": 5.422271858131547e-06,
      "loss": 0.0342,
      "step": 43075
    },
    {
      "epoch": 2.1869130412711306,
      "grad_norm": 0.27027949690818787,
      "learning_rate": 5.420579724859131e-06,
      "loss": 0.0312,
      "step": 43080
    },
    {
      "epoch": 2.187166861261993,
      "grad_norm": 1.943433403968811,
      "learning_rate": 5.418887591586714e-06,
      "loss": 0.0378,
      "step": 43085
    },
    {
      "epoch": 2.1874206812528554,
      "grad_norm": 0.5797852873802185,
      "learning_rate": 5.417195458314297e-06,
      "loss": 0.0346,
      "step": 43090
    },
    {
      "epoch": 2.187674501243718,
      "grad_norm": 0.39422717690467834,
      "learning_rate": 5.415503325041881e-06,
      "loss": 0.0347,
      "step": 43095
    },
    {
      "epoch": 2.1879283212345806,
      "grad_norm": 0.26226067543029785,
      "learning_rate": 5.413811191769464e-06,
      "loss": 0.0319,
      "step": 43100
    },
    {
      "epoch": 2.1881821412254427,
      "grad_norm": 0.3121758699417114,
      "learning_rate": 5.412119058497047e-06,
      "loss": 0.0367,
      "step": 43105
    },
    {
      "epoch": 2.1884359612163053,
      "grad_norm": 0.36041760444641113,
      "learning_rate": 5.4104269252246315e-06,
      "loss": 0.0336,
      "step": 43110
    },
    {
      "epoch": 2.188689781207168,
      "grad_norm": 0.25645551085472107,
      "learning_rate": 5.408734791952215e-06,
      "loss": 0.028,
      "step": 43115
    },
    {
      "epoch": 2.1889436011980306,
      "grad_norm": 0.30976834893226624,
      "learning_rate": 5.407042658679799e-06,
      "loss": 0.0392,
      "step": 43120
    },
    {
      "epoch": 2.1891974211888927,
      "grad_norm": 0.3137769103050232,
      "learning_rate": 5.405350525407381e-06,
      "loss": 0.0323,
      "step": 43125
    },
    {
      "epoch": 2.1894512411797553,
      "grad_norm": 0.5139391422271729,
      "learning_rate": 5.403658392134965e-06,
      "loss": 0.0342,
      "step": 43130
    },
    {
      "epoch": 2.189705061170618,
      "grad_norm": 0.32713380455970764,
      "learning_rate": 5.401966258862548e-06,
      "loss": 0.0334,
      "step": 43135
    },
    {
      "epoch": 2.18995888116148,
      "grad_norm": 0.36082905530929565,
      "learning_rate": 5.400274125590131e-06,
      "loss": 0.0321,
      "step": 43140
    },
    {
      "epoch": 2.1902127011523427,
      "grad_norm": 0.2427683025598526,
      "learning_rate": 5.3985819923177155e-06,
      "loss": 0.0413,
      "step": 43145
    },
    {
      "epoch": 2.1904665211432053,
      "grad_norm": 0.6607348322868347,
      "learning_rate": 5.396889859045299e-06,
      "loss": 0.0409,
      "step": 43150
    },
    {
      "epoch": 2.190720341134068,
      "grad_norm": 0.4588119387626648,
      "learning_rate": 5.3951977257728826e-06,
      "loss": 0.0358,
      "step": 43155
    },
    {
      "epoch": 2.19097416112493,
      "grad_norm": 0.3139321208000183,
      "learning_rate": 5.393505592500465e-06,
      "loss": 0.0299,
      "step": 43160
    },
    {
      "epoch": 2.1912279811157926,
      "grad_norm": 0.24087227880954742,
      "learning_rate": 5.391813459228049e-06,
      "loss": 0.0316,
      "step": 43165
    },
    {
      "epoch": 2.1914818011066552,
      "grad_norm": 0.338569313287735,
      "learning_rate": 5.390121325955633e-06,
      "loss": 0.0268,
      "step": 43170
    },
    {
      "epoch": 2.191735621097518,
      "grad_norm": 0.3832125961780548,
      "learning_rate": 5.388429192683217e-06,
      "loss": 0.0354,
      "step": 43175
    },
    {
      "epoch": 2.19198944108838,
      "grad_norm": 0.2771073877811432,
      "learning_rate": 5.3867370594107994e-06,
      "loss": 0.0356,
      "step": 43180
    },
    {
      "epoch": 2.1922432610792426,
      "grad_norm": 0.27626606822013855,
      "learning_rate": 5.385044926138383e-06,
      "loss": 0.0333,
      "step": 43185
    },
    {
      "epoch": 2.192497081070105,
      "grad_norm": 0.3390451669692993,
      "learning_rate": 5.3833527928659665e-06,
      "loss": 0.0294,
      "step": 43190
    },
    {
      "epoch": 2.1927509010609674,
      "grad_norm": 0.2988022565841675,
      "learning_rate": 5.381660659593551e-06,
      "loss": 0.034,
      "step": 43195
    },
    {
      "epoch": 2.19300472105183,
      "grad_norm": 0.41277876496315,
      "learning_rate": 5.379968526321134e-06,
      "loss": 0.0328,
      "step": 43200
    },
    {
      "epoch": 2.1932585410426926,
      "grad_norm": 0.3886488080024719,
      "learning_rate": 5.378276393048717e-06,
      "loss": 0.0329,
      "step": 43205
    },
    {
      "epoch": 2.193512361033555,
      "grad_norm": 0.4166080057621002,
      "learning_rate": 5.376584259776301e-06,
      "loss": 0.0345,
      "step": 43210
    },
    {
      "epoch": 2.1937661810244173,
      "grad_norm": 0.311166912317276,
      "learning_rate": 5.374892126503883e-06,
      "loss": 0.0259,
      "step": 43215
    },
    {
      "epoch": 2.19402000101528,
      "grad_norm": 1.2612603902816772,
      "learning_rate": 5.373199993231467e-06,
      "loss": 0.0365,
      "step": 43220
    },
    {
      "epoch": 2.1942738210061425,
      "grad_norm": 0.2629720866680145,
      "learning_rate": 5.3715078599590505e-06,
      "loss": 0.0387,
      "step": 43225
    },
    {
      "epoch": 2.1945276409970047,
      "grad_norm": 0.41876986622810364,
      "learning_rate": 5.369815726686635e-06,
      "loss": 0.0417,
      "step": 43230
    },
    {
      "epoch": 2.1947814609878673,
      "grad_norm": 0.4134247303009033,
      "learning_rate": 5.3681235934142176e-06,
      "loss": 0.043,
      "step": 43235
    },
    {
      "epoch": 2.19503528097873,
      "grad_norm": 0.7171376347541809,
      "learning_rate": 5.366431460141801e-06,
      "loss": 0.0327,
      "step": 43240
    },
    {
      "epoch": 2.1952891009695925,
      "grad_norm": 0.4790825843811035,
      "learning_rate": 5.364739326869385e-06,
      "loss": 0.0343,
      "step": 43245
    },
    {
      "epoch": 2.1955429209604547,
      "grad_norm": 0.47442811727523804,
      "learning_rate": 5.363047193596968e-06,
      "loss": 0.032,
      "step": 43250
    },
    {
      "epoch": 2.1957967409513173,
      "grad_norm": 0.32504549622535706,
      "learning_rate": 5.361355060324551e-06,
      "loss": 0.0348,
      "step": 43255
    },
    {
      "epoch": 2.19605056094218,
      "grad_norm": 0.3139643371105194,
      "learning_rate": 5.359662927052135e-06,
      "loss": 0.0274,
      "step": 43260
    },
    {
      "epoch": 2.1963043809330425,
      "grad_norm": 0.3576580882072449,
      "learning_rate": 5.357970793779719e-06,
      "loss": 0.0382,
      "step": 43265
    },
    {
      "epoch": 2.1965582009239046,
      "grad_norm": 0.3130418658256531,
      "learning_rate": 5.356278660507302e-06,
      "loss": 0.0297,
      "step": 43270
    },
    {
      "epoch": 2.1968120209147672,
      "grad_norm": 0.2839604616165161,
      "learning_rate": 5.354586527234885e-06,
      "loss": 0.0261,
      "step": 43275
    },
    {
      "epoch": 2.19706584090563,
      "grad_norm": 0.5264848470687866,
      "learning_rate": 5.352894393962469e-06,
      "loss": 0.0336,
      "step": 43280
    },
    {
      "epoch": 2.197319660896492,
      "grad_norm": 0.40158921480178833,
      "learning_rate": 5.351202260690053e-06,
      "loss": 0.0384,
      "step": 43285
    },
    {
      "epoch": 2.1975734808873546,
      "grad_norm": 0.35445520281791687,
      "learning_rate": 5.349510127417636e-06,
      "loss": 0.0316,
      "step": 43290
    },
    {
      "epoch": 2.197827300878217,
      "grad_norm": 0.2767607867717743,
      "learning_rate": 5.347817994145219e-06,
      "loss": 0.0325,
      "step": 43295
    },
    {
      "epoch": 2.19808112086908,
      "grad_norm": 0.2724688649177551,
      "learning_rate": 5.346125860872803e-06,
      "loss": 0.0289,
      "step": 43300
    },
    {
      "epoch": 2.198334940859942,
      "grad_norm": 0.3401924967765808,
      "learning_rate": 5.344433727600386e-06,
      "loss": 0.0364,
      "step": 43305
    },
    {
      "epoch": 2.1985887608508046,
      "grad_norm": 0.27552342414855957,
      "learning_rate": 5.342741594327969e-06,
      "loss": 0.0298,
      "step": 43310
    },
    {
      "epoch": 2.198842580841667,
      "grad_norm": 0.31632283329963684,
      "learning_rate": 5.341049461055553e-06,
      "loss": 0.0272,
      "step": 43315
    },
    {
      "epoch": 2.1990964008325298,
      "grad_norm": 0.5958532094955444,
      "learning_rate": 5.339357327783137e-06,
      "loss": 0.0328,
      "step": 43320
    },
    {
      "epoch": 2.199350220823392,
      "grad_norm": 0.29851803183555603,
      "learning_rate": 5.3376651945107205e-06,
      "loss": 0.0306,
      "step": 43325
    },
    {
      "epoch": 2.1996040408142545,
      "grad_norm": 0.30227798223495483,
      "learning_rate": 5.335973061238303e-06,
      "loss": 0.0267,
      "step": 43330
    },
    {
      "epoch": 2.199857860805117,
      "grad_norm": 0.31471794843673706,
      "learning_rate": 5.334280927965887e-06,
      "loss": 0.0315,
      "step": 43335
    },
    {
      "epoch": 2.2001116807959793,
      "grad_norm": 0.3955003619194031,
      "learning_rate": 5.33258879469347e-06,
      "loss": 0.0348,
      "step": 43340
    },
    {
      "epoch": 2.200365500786842,
      "grad_norm": 0.2753443717956543,
      "learning_rate": 5.330896661421055e-06,
      "loss": 0.0343,
      "step": 43345
    },
    {
      "epoch": 2.2006193207777045,
      "grad_norm": 0.282761812210083,
      "learning_rate": 5.329204528148637e-06,
      "loss": 0.0376,
      "step": 43350
    },
    {
      "epoch": 2.200873140768567,
      "grad_norm": 0.37203195691108704,
      "learning_rate": 5.327512394876221e-06,
      "loss": 0.0445,
      "step": 43355
    },
    {
      "epoch": 2.2011269607594293,
      "grad_norm": 0.4049202501773834,
      "learning_rate": 5.3258202616038045e-06,
      "loss": 0.033,
      "step": 43360
    },
    {
      "epoch": 2.201380780750292,
      "grad_norm": 0.4423353374004364,
      "learning_rate": 5.324128128331388e-06,
      "loss": 0.0341,
      "step": 43365
    },
    {
      "epoch": 2.2016346007411545,
      "grad_norm": 0.3491520881652832,
      "learning_rate": 5.322435995058971e-06,
      "loss": 0.0329,
      "step": 43370
    },
    {
      "epoch": 2.2018884207320166,
      "grad_norm": 0.3284947872161865,
      "learning_rate": 5.320743861786555e-06,
      "loss": 0.032,
      "step": 43375
    },
    {
      "epoch": 2.2021422407228792,
      "grad_norm": 0.36112505197525024,
      "learning_rate": 5.319051728514139e-06,
      "loss": 0.0288,
      "step": 43380
    },
    {
      "epoch": 2.202396060713742,
      "grad_norm": 0.3680697977542877,
      "learning_rate": 5.317359595241721e-06,
      "loss": 0.0314,
      "step": 43385
    },
    {
      "epoch": 2.2026498807046044,
      "grad_norm": 0.26703742146492004,
      "learning_rate": 5.315667461969305e-06,
      "loss": 0.0287,
      "step": 43390
    },
    {
      "epoch": 2.2029037006954666,
      "grad_norm": 0.4103862941265106,
      "learning_rate": 5.3139753286968885e-06,
      "loss": 0.034,
      "step": 43395
    },
    {
      "epoch": 2.203157520686329,
      "grad_norm": 0.34333786368370056,
      "learning_rate": 5.312283195424473e-06,
      "loss": 0.0294,
      "step": 43400
    },
    {
      "epoch": 2.203411340677192,
      "grad_norm": 0.5722442269325256,
      "learning_rate": 5.310591062152055e-06,
      "loss": 0.0352,
      "step": 43405
    },
    {
      "epoch": 2.2036651606680544,
      "grad_norm": 0.44682955741882324,
      "learning_rate": 5.308898928879639e-06,
      "loss": 0.0376,
      "step": 43410
    },
    {
      "epoch": 2.2039189806589166,
      "grad_norm": 1.0077766180038452,
      "learning_rate": 5.307206795607223e-06,
      "loss": 0.032,
      "step": 43415
    },
    {
      "epoch": 2.204172800649779,
      "grad_norm": 0.40870773792266846,
      "learning_rate": 5.305514662334806e-06,
      "loss": 0.0338,
      "step": 43420
    },
    {
      "epoch": 2.2044266206406418,
      "grad_norm": 0.2662786543369293,
      "learning_rate": 5.303822529062389e-06,
      "loss": 0.0325,
      "step": 43425
    },
    {
      "epoch": 2.2046804406315044,
      "grad_norm": 0.6228799223899841,
      "learning_rate": 5.302130395789972e-06,
      "loss": 0.0335,
      "step": 43430
    },
    {
      "epoch": 2.2049342606223665,
      "grad_norm": 0.44985294342041016,
      "learning_rate": 5.300438262517557e-06,
      "loss": 0.0378,
      "step": 43435
    },
    {
      "epoch": 2.205188080613229,
      "grad_norm": 0.36874616146087646,
      "learning_rate": 5.29874612924514e-06,
      "loss": 0.033,
      "step": 43440
    },
    {
      "epoch": 2.2054419006040917,
      "grad_norm": 0.8388116359710693,
      "learning_rate": 5.297053995972723e-06,
      "loss": 0.0335,
      "step": 43445
    },
    {
      "epoch": 2.205695720594954,
      "grad_norm": 0.3380875885486603,
      "learning_rate": 5.295361862700307e-06,
      "loss": 0.0337,
      "step": 43450
    },
    {
      "epoch": 2.2059495405858165,
      "grad_norm": 0.27034997940063477,
      "learning_rate": 5.29366972942789e-06,
      "loss": 0.0401,
      "step": 43455
    },
    {
      "epoch": 2.206203360576679,
      "grad_norm": 0.31091082096099854,
      "learning_rate": 5.291977596155473e-06,
      "loss": 0.0298,
      "step": 43460
    },
    {
      "epoch": 2.2064571805675417,
      "grad_norm": 0.33562663197517395,
      "learning_rate": 5.290285462883057e-06,
      "loss": 0.0291,
      "step": 43465
    },
    {
      "epoch": 2.206711000558404,
      "grad_norm": 0.42769908905029297,
      "learning_rate": 5.288593329610641e-06,
      "loss": 0.0266,
      "step": 43470
    },
    {
      "epoch": 2.2069648205492665,
      "grad_norm": 0.28732359409332275,
      "learning_rate": 5.286901196338224e-06,
      "loss": 0.0259,
      "step": 43475
    },
    {
      "epoch": 2.207218640540129,
      "grad_norm": 0.4161680042743683,
      "learning_rate": 5.285209063065807e-06,
      "loss": 0.0345,
      "step": 43480
    },
    {
      "epoch": 2.207472460530991,
      "grad_norm": 0.3423847556114197,
      "learning_rate": 5.2835169297933906e-06,
      "loss": 0.0316,
      "step": 43485
    },
    {
      "epoch": 2.207726280521854,
      "grad_norm": 0.32624608278274536,
      "learning_rate": 5.281824796520975e-06,
      "loss": 0.0316,
      "step": 43490
    },
    {
      "epoch": 2.2079801005127164,
      "grad_norm": 0.43581536412239075,
      "learning_rate": 5.2801326632485585e-06,
      "loss": 0.0351,
      "step": 43495
    },
    {
      "epoch": 2.208233920503579,
      "grad_norm": 0.3912333846092224,
      "learning_rate": 5.278440529976141e-06,
      "loss": 0.0382,
      "step": 43500
    },
    {
      "epoch": 2.208487740494441,
      "grad_norm": 0.260891318321228,
      "learning_rate": 5.276748396703725e-06,
      "loss": 0.0283,
      "step": 43505
    },
    {
      "epoch": 2.208741560485304,
      "grad_norm": 0.42657461762428284,
      "learning_rate": 5.275056263431308e-06,
      "loss": 0.0328,
      "step": 43510
    },
    {
      "epoch": 2.2089953804761664,
      "grad_norm": 0.3546578288078308,
      "learning_rate": 5.273364130158893e-06,
      "loss": 0.0335,
      "step": 43515
    },
    {
      "epoch": 2.2092492004670286,
      "grad_norm": 0.2557455599308014,
      "learning_rate": 5.2716719968864745e-06,
      "loss": 0.0332,
      "step": 43520
    },
    {
      "epoch": 2.209503020457891,
      "grad_norm": 0.3160721957683563,
      "learning_rate": 5.269979863614059e-06,
      "loss": 0.0341,
      "step": 43525
    },
    {
      "epoch": 2.2097568404487538,
      "grad_norm": 0.27429550886154175,
      "learning_rate": 5.2682877303416425e-06,
      "loss": 0.0334,
      "step": 43530
    },
    {
      "epoch": 2.2100106604396164,
      "grad_norm": 0.2348918616771698,
      "learning_rate": 5.266595597069225e-06,
      "loss": 0.034,
      "step": 43535
    },
    {
      "epoch": 2.2102644804304785,
      "grad_norm": 0.19068334996700287,
      "learning_rate": 5.264903463796809e-06,
      "loss": 0.032,
      "step": 43540
    },
    {
      "epoch": 2.210518300421341,
      "grad_norm": 0.3050518035888672,
      "learning_rate": 5.263211330524392e-06,
      "loss": 0.0305,
      "step": 43545
    },
    {
      "epoch": 2.2107721204122037,
      "grad_norm": 0.5854632258415222,
      "learning_rate": 5.261519197251977e-06,
      "loss": 0.0424,
      "step": 43550
    },
    {
      "epoch": 2.2110259404030663,
      "grad_norm": 0.3303602933883667,
      "learning_rate": 5.259827063979559e-06,
      "loss": 0.027,
      "step": 43555
    },
    {
      "epoch": 2.2112797603939285,
      "grad_norm": 0.304310142993927,
      "learning_rate": 5.258134930707143e-06,
      "loss": 0.0386,
      "step": 43560
    },
    {
      "epoch": 2.211533580384791,
      "grad_norm": 0.38438746333122253,
      "learning_rate": 5.2564427974347264e-06,
      "loss": 0.0353,
      "step": 43565
    },
    {
      "epoch": 2.2117874003756537,
      "grad_norm": 0.20120958983898163,
      "learning_rate": 5.25475066416231e-06,
      "loss": 0.034,
      "step": 43570
    },
    {
      "epoch": 2.2120412203665163,
      "grad_norm": 0.2356599122285843,
      "learning_rate": 5.253058530889893e-06,
      "loss": 0.0395,
      "step": 43575
    },
    {
      "epoch": 2.2122950403573785,
      "grad_norm": 0.3216328024864197,
      "learning_rate": 5.251366397617477e-06,
      "loss": 0.0285,
      "step": 43580
    },
    {
      "epoch": 2.212548860348241,
      "grad_norm": 0.737366259098053,
      "learning_rate": 5.249674264345061e-06,
      "loss": 0.0331,
      "step": 43585
    },
    {
      "epoch": 2.2128026803391037,
      "grad_norm": 0.27575215697288513,
      "learning_rate": 5.247982131072644e-06,
      "loss": 0.037,
      "step": 43590
    },
    {
      "epoch": 2.213056500329966,
      "grad_norm": 0.2615464925765991,
      "learning_rate": 5.246289997800227e-06,
      "loss": 0.0316,
      "step": 43595
    },
    {
      "epoch": 2.2133103203208284,
      "grad_norm": 0.2448047250509262,
      "learning_rate": 5.24459786452781e-06,
      "loss": 0.0326,
      "step": 43600
    },
    {
      "epoch": 2.213564140311691,
      "grad_norm": 0.37527599930763245,
      "learning_rate": 5.242905731255395e-06,
      "loss": 0.0335,
      "step": 43605
    },
    {
      "epoch": 2.2138179603025536,
      "grad_norm": 0.5040178298950195,
      "learning_rate": 5.241213597982977e-06,
      "loss": 0.0371,
      "step": 43610
    },
    {
      "epoch": 2.214071780293416,
      "grad_norm": 0.2776201069355011,
      "learning_rate": 5.239521464710561e-06,
      "loss": 0.029,
      "step": 43615
    },
    {
      "epoch": 2.2143256002842784,
      "grad_norm": 0.25941890478134155,
      "learning_rate": 5.2378293314381446e-06,
      "loss": 0.0282,
      "step": 43620
    },
    {
      "epoch": 2.214579420275141,
      "grad_norm": 0.4714052677154541,
      "learning_rate": 5.236137198165728e-06,
      "loss": 0.0397,
      "step": 43625
    },
    {
      "epoch": 2.214833240266003,
      "grad_norm": 0.1847241073846817,
      "learning_rate": 5.234445064893311e-06,
      "loss": 0.0312,
      "step": 43630
    },
    {
      "epoch": 2.2150870602568657,
      "grad_norm": 0.7910279631614685,
      "learning_rate": 5.232752931620894e-06,
      "loss": 0.0323,
      "step": 43635
    },
    {
      "epoch": 2.2153408802477284,
      "grad_norm": 0.3663308322429657,
      "learning_rate": 5.231060798348479e-06,
      "loss": 0.0341,
      "step": 43640
    },
    {
      "epoch": 2.215594700238591,
      "grad_norm": 0.3408203125,
      "learning_rate": 5.229368665076062e-06,
      "loss": 0.0352,
      "step": 43645
    },
    {
      "epoch": 2.215848520229453,
      "grad_norm": 1.7612500190734863,
      "learning_rate": 5.227676531803645e-06,
      "loss": 0.0296,
      "step": 43650
    },
    {
      "epoch": 2.2161023402203157,
      "grad_norm": 0.46181654930114746,
      "learning_rate": 5.2259843985312285e-06,
      "loss": 0.0389,
      "step": 43655
    },
    {
      "epoch": 2.2163561602111783,
      "grad_norm": 0.25822097063064575,
      "learning_rate": 5.224292265258812e-06,
      "loss": 0.0345,
      "step": 43660
    },
    {
      "epoch": 2.216609980202041,
      "grad_norm": 0.5843444466590881,
      "learning_rate": 5.2226001319863965e-06,
      "loss": 0.0396,
      "step": 43665
    },
    {
      "epoch": 2.216863800192903,
      "grad_norm": 0.31277167797088623,
      "learning_rate": 5.220907998713979e-06,
      "loss": 0.0311,
      "step": 43670
    },
    {
      "epoch": 2.2171176201837657,
      "grad_norm": 0.5508875846862793,
      "learning_rate": 5.219215865441563e-06,
      "loss": 0.0385,
      "step": 43675
    },
    {
      "epoch": 2.2173714401746283,
      "grad_norm": 0.31969529390335083,
      "learning_rate": 5.217523732169146e-06,
      "loss": 0.0308,
      "step": 43680
    },
    {
      "epoch": 2.2176252601654904,
      "grad_norm": 0.357572078704834,
      "learning_rate": 5.21583159889673e-06,
      "loss": 0.0301,
      "step": 43685
    },
    {
      "epoch": 2.217879080156353,
      "grad_norm": 0.3687649667263031,
      "learning_rate": 5.2141394656243125e-06,
      "loss": 0.0362,
      "step": 43690
    },
    {
      "epoch": 2.2181329001472156,
      "grad_norm": 0.4379350543022156,
      "learning_rate": 5.212447332351897e-06,
      "loss": 0.0313,
      "step": 43695
    },
    {
      "epoch": 2.2183867201380782,
      "grad_norm": 0.270060271024704,
      "learning_rate": 5.2107551990794804e-06,
      "loss": 0.0282,
      "step": 43700
    },
    {
      "epoch": 2.2186405401289404,
      "grad_norm": 0.3274651765823364,
      "learning_rate": 5.209063065807063e-06,
      "loss": 0.0246,
      "step": 43705
    },
    {
      "epoch": 2.218894360119803,
      "grad_norm": 0.4745948314666748,
      "learning_rate": 5.207370932534647e-06,
      "loss": 0.0292,
      "step": 43710
    },
    {
      "epoch": 2.2191481801106656,
      "grad_norm": 0.3572133481502533,
      "learning_rate": 5.20567879926223e-06,
      "loss": 0.0325,
      "step": 43715
    },
    {
      "epoch": 2.219402000101528,
      "grad_norm": 0.3448382616043091,
      "learning_rate": 5.203986665989814e-06,
      "loss": 0.0356,
      "step": 43720
    },
    {
      "epoch": 2.2196558200923904,
      "grad_norm": 0.2912856638431549,
      "learning_rate": 5.2022945327173965e-06,
      "loss": 0.0295,
      "step": 43725
    },
    {
      "epoch": 2.219909640083253,
      "grad_norm": 0.41613078117370605,
      "learning_rate": 5.200602399444981e-06,
      "loss": 0.0326,
      "step": 43730
    },
    {
      "epoch": 2.2201634600741156,
      "grad_norm": 0.2688710391521454,
      "learning_rate": 5.198910266172564e-06,
      "loss": 0.0296,
      "step": 43735
    },
    {
      "epoch": 2.2204172800649777,
      "grad_norm": 0.40279969573020935,
      "learning_rate": 5.197218132900148e-06,
      "loss": 0.0368,
      "step": 43740
    },
    {
      "epoch": 2.2206711000558403,
      "grad_norm": 0.34636586904525757,
      "learning_rate": 5.195525999627731e-06,
      "loss": 0.0299,
      "step": 43745
    },
    {
      "epoch": 2.220924920046703,
      "grad_norm": 0.40201225876808167,
      "learning_rate": 5.193833866355314e-06,
      "loss": 0.0297,
      "step": 43750
    },
    {
      "epoch": 2.2211787400375655,
      "grad_norm": 0.2993839383125305,
      "learning_rate": 5.192141733082899e-06,
      "loss": 0.0272,
      "step": 43755
    },
    {
      "epoch": 2.2214325600284277,
      "grad_norm": 0.3276756703853607,
      "learning_rate": 5.190449599810482e-06,
      "loss": 0.0333,
      "step": 43760
    },
    {
      "epoch": 2.2216863800192903,
      "grad_norm": 0.22368964552879333,
      "learning_rate": 5.188757466538065e-06,
      "loss": 0.0304,
      "step": 43765
    },
    {
      "epoch": 2.221940200010153,
      "grad_norm": 0.42625653743743896,
      "learning_rate": 5.187065333265648e-06,
      "loss": 0.0367,
      "step": 43770
    },
    {
      "epoch": 2.222194020001015,
      "grad_norm": 0.3062908947467804,
      "learning_rate": 5.185373199993232e-06,
      "loss": 0.0266,
      "step": 43775
    },
    {
      "epoch": 2.2224478399918777,
      "grad_norm": 0.42325690388679504,
      "learning_rate": 5.183681066720815e-06,
      "loss": 0.0298,
      "step": 43780
    },
    {
      "epoch": 2.2227016599827403,
      "grad_norm": 0.2115492969751358,
      "learning_rate": 5.181988933448399e-06,
      "loss": 0.0328,
      "step": 43785
    },
    {
      "epoch": 2.222955479973603,
      "grad_norm": 0.3463805913925171,
      "learning_rate": 5.1802968001759825e-06,
      "loss": 0.0396,
      "step": 43790
    },
    {
      "epoch": 2.223209299964465,
      "grad_norm": 0.5982481837272644,
      "learning_rate": 5.178604666903566e-06,
      "loss": 0.0276,
      "step": 43795
    },
    {
      "epoch": 2.2234631199553276,
      "grad_norm": 0.4598280191421509,
      "learning_rate": 5.176912533631149e-06,
      "loss": 0.0321,
      "step": 43800
    },
    {
      "epoch": 2.2237169399461902,
      "grad_norm": 0.33554238080978394,
      "learning_rate": 5.175220400358732e-06,
      "loss": 0.0358,
      "step": 43805
    },
    {
      "epoch": 2.223970759937053,
      "grad_norm": 1.4233322143554688,
      "learning_rate": 5.173528267086316e-06,
      "loss": 0.037,
      "step": 43810
    },
    {
      "epoch": 2.224224579927915,
      "grad_norm": 0.27777615189552307,
      "learning_rate": 5.1718361338139e-06,
      "loss": 0.0287,
      "step": 43815
    },
    {
      "epoch": 2.2244783999187776,
      "grad_norm": 0.2759939134120941,
      "learning_rate": 5.170144000541483e-06,
      "loss": 0.0329,
      "step": 43820
    },
    {
      "epoch": 2.22473221990964,
      "grad_norm": 0.35867956280708313,
      "learning_rate": 5.1684518672690665e-06,
      "loss": 0.0311,
      "step": 43825
    },
    {
      "epoch": 2.2249860399005024,
      "grad_norm": 0.39028945565223694,
      "learning_rate": 5.16675973399665e-06,
      "loss": 0.0382,
      "step": 43830
    },
    {
      "epoch": 2.225239859891365,
      "grad_norm": 0.1933843493461609,
      "learning_rate": 5.165067600724234e-06,
      "loss": 0.0234,
      "step": 43835
    },
    {
      "epoch": 2.2254936798822276,
      "grad_norm": 0.35192665457725525,
      "learning_rate": 5.163375467451816e-06,
      "loss": 0.0352,
      "step": 43840
    },
    {
      "epoch": 2.22574749987309,
      "grad_norm": 0.3243769109249115,
      "learning_rate": 5.161683334179401e-06,
      "loss": 0.037,
      "step": 43845
    },
    {
      "epoch": 2.2260013198639523,
      "grad_norm": 0.31889811158180237,
      "learning_rate": 5.159991200906984e-06,
      "loss": 0.0351,
      "step": 43850
    },
    {
      "epoch": 2.226255139854815,
      "grad_norm": 0.3885818421840668,
      "learning_rate": 5.158299067634567e-06,
      "loss": 0.0349,
      "step": 43855
    },
    {
      "epoch": 2.2265089598456775,
      "grad_norm": 0.31781822443008423,
      "learning_rate": 5.1566069343621505e-06,
      "loss": 0.036,
      "step": 43860
    },
    {
      "epoch": 2.22676277983654,
      "grad_norm": 0.24584543704986572,
      "learning_rate": 5.154914801089734e-06,
      "loss": 0.0322,
      "step": 43865
    },
    {
      "epoch": 2.2270165998274023,
      "grad_norm": 0.3724859952926636,
      "learning_rate": 5.153222667817318e-06,
      "loss": 0.0459,
      "step": 43870
    },
    {
      "epoch": 2.227270419818265,
      "grad_norm": 0.37444204092025757,
      "learning_rate": 5.151530534544901e-06,
      "loss": 0.034,
      "step": 43875
    },
    {
      "epoch": 2.2275242398091275,
      "grad_norm": 0.3302791714668274,
      "learning_rate": 5.149838401272485e-06,
      "loss": 0.0386,
      "step": 43880
    },
    {
      "epoch": 2.2277780597999897,
      "grad_norm": 0.331866979598999,
      "learning_rate": 5.148146268000068e-06,
      "loss": 0.033,
      "step": 43885
    },
    {
      "epoch": 2.2280318797908523,
      "grad_norm": 0.2684163451194763,
      "learning_rate": 5.146454134727652e-06,
      "loss": 0.0304,
      "step": 43890
    },
    {
      "epoch": 2.228285699781715,
      "grad_norm": 0.30631887912750244,
      "learning_rate": 5.1447620014552344e-06,
      "loss": 0.0383,
      "step": 43895
    },
    {
      "epoch": 2.2285395197725775,
      "grad_norm": 0.3266892433166504,
      "learning_rate": 5.143069868182818e-06,
      "loss": 0.0285,
      "step": 43900
    },
    {
      "epoch": 2.2287933397634396,
      "grad_norm": 0.31804943084716797,
      "learning_rate": 5.141377734910402e-06,
      "loss": 0.0307,
      "step": 43905
    },
    {
      "epoch": 2.2290471597543022,
      "grad_norm": 0.44756922125816345,
      "learning_rate": 5.139685601637986e-06,
      "loss": 0.0385,
      "step": 43910
    },
    {
      "epoch": 2.229300979745165,
      "grad_norm": 0.2774399220943451,
      "learning_rate": 5.137993468365569e-06,
      "loss": 0.0275,
      "step": 43915
    },
    {
      "epoch": 2.229554799736027,
      "grad_norm": 0.29544955492019653,
      "learning_rate": 5.136301335093152e-06,
      "loss": 0.0315,
      "step": 43920
    },
    {
      "epoch": 2.2298086197268896,
      "grad_norm": 0.2739022672176361,
      "learning_rate": 5.134609201820736e-06,
      "loss": 0.0295,
      "step": 43925
    },
    {
      "epoch": 2.230062439717752,
      "grad_norm": 0.3167058229446411,
      "learning_rate": 5.13291706854832e-06,
      "loss": 0.0367,
      "step": 43930
    },
    {
      "epoch": 2.230316259708615,
      "grad_norm": 0.352201908826828,
      "learning_rate": 5.131224935275903e-06,
      "loss": 0.0345,
      "step": 43935
    },
    {
      "epoch": 2.230570079699477,
      "grad_norm": 0.3670605719089508,
      "learning_rate": 5.129532802003486e-06,
      "loss": 0.0388,
      "step": 43940
    },
    {
      "epoch": 2.2308238996903396,
      "grad_norm": 0.38734838366508484,
      "learning_rate": 5.12784066873107e-06,
      "loss": 0.0315,
      "step": 43945
    },
    {
      "epoch": 2.231077719681202,
      "grad_norm": 0.33268794417381287,
      "learning_rate": 5.126148535458653e-06,
      "loss": 0.0336,
      "step": 43950
    },
    {
      "epoch": 2.2313315396720648,
      "grad_norm": 0.42838096618652344,
      "learning_rate": 5.124456402186236e-06,
      "loss": 0.028,
      "step": 43955
    },
    {
      "epoch": 2.231585359662927,
      "grad_norm": 0.4699762761592865,
      "learning_rate": 5.1227642689138205e-06,
      "loss": 0.0339,
      "step": 43960
    },
    {
      "epoch": 2.2318391796537895,
      "grad_norm": 0.6462127566337585,
      "learning_rate": 5.121072135641404e-06,
      "loss": 0.0318,
      "step": 43965
    },
    {
      "epoch": 2.232092999644652,
      "grad_norm": 0.2937588393688202,
      "learning_rate": 5.119380002368987e-06,
      "loss": 0.0302,
      "step": 43970
    },
    {
      "epoch": 2.2323468196355143,
      "grad_norm": 0.2300993949174881,
      "learning_rate": 5.11768786909657e-06,
      "loss": 0.0362,
      "step": 43975
    },
    {
      "epoch": 2.232600639626377,
      "grad_norm": 0.5920124053955078,
      "learning_rate": 5.115995735824154e-06,
      "loss": 0.0308,
      "step": 43980
    },
    {
      "epoch": 2.2328544596172395,
      "grad_norm": 0.23320113122463226,
      "learning_rate": 5.114303602551738e-06,
      "loss": 0.0302,
      "step": 43985
    },
    {
      "epoch": 2.233108279608102,
      "grad_norm": 0.3090137243270874,
      "learning_rate": 5.11261146927932e-06,
      "loss": 0.0429,
      "step": 43990
    },
    {
      "epoch": 2.2333620995989643,
      "grad_norm": 0.44511866569519043,
      "learning_rate": 5.1109193360069045e-06,
      "loss": 0.0458,
      "step": 43995
    },
    {
      "epoch": 2.233615919589827,
      "grad_norm": 0.22914119064807892,
      "learning_rate": 5.109227202734488e-06,
      "loss": 0.0425,
      "step": 44000
    },
    {
      "epoch": 2.2338697395806895,
      "grad_norm": 0.2924869656562805,
      "learning_rate": 5.1075350694620716e-06,
      "loss": 0.0371,
      "step": 44005
    },
    {
      "epoch": 2.234123559571552,
      "grad_norm": 0.7008922100067139,
      "learning_rate": 5.105842936189654e-06,
      "loss": 0.0303,
      "step": 44010
    },
    {
      "epoch": 2.2343773795624142,
      "grad_norm": 0.3576049208641052,
      "learning_rate": 5.104150802917238e-06,
      "loss": 0.0336,
      "step": 44015
    },
    {
      "epoch": 2.234631199553277,
      "grad_norm": 0.37319669127464294,
      "learning_rate": 5.102458669644822e-06,
      "loss": 0.0333,
      "step": 44020
    },
    {
      "epoch": 2.2348850195441394,
      "grad_norm": 0.4137890934944153,
      "learning_rate": 5.100766536372405e-06,
      "loss": 0.0373,
      "step": 44025
    },
    {
      "epoch": 2.2351388395350016,
      "grad_norm": 0.35421955585479736,
      "learning_rate": 5.0990744030999885e-06,
      "loss": 0.0333,
      "step": 44030
    },
    {
      "epoch": 2.235392659525864,
      "grad_norm": 0.2621701955795288,
      "learning_rate": 5.097382269827572e-06,
      "loss": 0.0346,
      "step": 44035
    },
    {
      "epoch": 2.235646479516727,
      "grad_norm": 0.5451110005378723,
      "learning_rate": 5.0956901365551555e-06,
      "loss": 0.0326,
      "step": 44040
    },
    {
      "epoch": 2.2359002995075894,
      "grad_norm": 0.2946808934211731,
      "learning_rate": 5.093998003282738e-06,
      "loss": 0.0284,
      "step": 44045
    },
    {
      "epoch": 2.2361541194984516,
      "grad_norm": 0.37474432587623596,
      "learning_rate": 5.092305870010323e-06,
      "loss": 0.0386,
      "step": 44050
    },
    {
      "epoch": 2.236407939489314,
      "grad_norm": 0.37870481610298157,
      "learning_rate": 5.090613736737906e-06,
      "loss": 0.0325,
      "step": 44055
    },
    {
      "epoch": 2.2366617594801768,
      "grad_norm": 0.29171672463417053,
      "learning_rate": 5.08892160346549e-06,
      "loss": 0.0339,
      "step": 44060
    },
    {
      "epoch": 2.236915579471039,
      "grad_norm": 0.4076695442199707,
      "learning_rate": 5.087229470193072e-06,
      "loss": 0.0355,
      "step": 44065
    },
    {
      "epoch": 2.2371693994619015,
      "grad_norm": 0.30842044949531555,
      "learning_rate": 5.085537336920656e-06,
      "loss": 0.0373,
      "step": 44070
    },
    {
      "epoch": 2.237423219452764,
      "grad_norm": 0.28765350580215454,
      "learning_rate": 5.08384520364824e-06,
      "loss": 0.0345,
      "step": 44075
    },
    {
      "epoch": 2.2376770394436267,
      "grad_norm": 0.28565219044685364,
      "learning_rate": 5.082153070375824e-06,
      "loss": 0.0401,
      "step": 44080
    },
    {
      "epoch": 2.237930859434489,
      "grad_norm": 0.7458963394165039,
      "learning_rate": 5.080460937103407e-06,
      "loss": 0.0345,
      "step": 44085
    },
    {
      "epoch": 2.2381846794253515,
      "grad_norm": 0.4765220284461975,
      "learning_rate": 5.07876880383099e-06,
      "loss": 0.0373,
      "step": 44090
    },
    {
      "epoch": 2.238438499416214,
      "grad_norm": 0.35971006751060486,
      "learning_rate": 5.077076670558574e-06,
      "loss": 0.0373,
      "step": 44095
    },
    {
      "epoch": 2.2386923194070767,
      "grad_norm": 0.37267395853996277,
      "learning_rate": 5.075384537286156e-06,
      "loss": 0.0402,
      "step": 44100
    },
    {
      "epoch": 2.238946139397939,
      "grad_norm": 0.30565184354782104,
      "learning_rate": 5.07369240401374e-06,
      "loss": 0.0313,
      "step": 44105
    },
    {
      "epoch": 2.2391999593888015,
      "grad_norm": 0.4393250346183777,
      "learning_rate": 5.072000270741324e-06,
      "loss": 0.0292,
      "step": 44110
    },
    {
      "epoch": 2.239453779379664,
      "grad_norm": 0.30050981044769287,
      "learning_rate": 5.070308137468908e-06,
      "loss": 0.0419,
      "step": 44115
    },
    {
      "epoch": 2.2397075993705267,
      "grad_norm": 0.43722015619277954,
      "learning_rate": 5.0686160041964906e-06,
      "loss": 0.0313,
      "step": 44120
    },
    {
      "epoch": 2.239961419361389,
      "grad_norm": 0.3736498951911926,
      "learning_rate": 5.066923870924074e-06,
      "loss": 0.0342,
      "step": 44125
    },
    {
      "epoch": 2.2402152393522514,
      "grad_norm": 0.5136289596557617,
      "learning_rate": 5.065231737651658e-06,
      "loss": 0.0369,
      "step": 44130
    },
    {
      "epoch": 2.240469059343114,
      "grad_norm": 0.2988525331020355,
      "learning_rate": 5.063539604379242e-06,
      "loss": 0.0309,
      "step": 44135
    },
    {
      "epoch": 2.240722879333976,
      "grad_norm": 0.24521800875663757,
      "learning_rate": 5.061847471106825e-06,
      "loss": 0.0296,
      "step": 44140
    },
    {
      "epoch": 2.240976699324839,
      "grad_norm": 0.5935750007629395,
      "learning_rate": 5.060155337834408e-06,
      "loss": 0.038,
      "step": 44145
    },
    {
      "epoch": 2.2412305193157014,
      "grad_norm": 0.5300387144088745,
      "learning_rate": 5.058463204561992e-06,
      "loss": 0.0314,
      "step": 44150
    },
    {
      "epoch": 2.241484339306564,
      "grad_norm": 0.32966119050979614,
      "learning_rate": 5.056771071289575e-06,
      "loss": 0.0302,
      "step": 44155
    },
    {
      "epoch": 2.241738159297426,
      "grad_norm": 0.35278815031051636,
      "learning_rate": 5.055078938017158e-06,
      "loss": 0.0346,
      "step": 44160
    },
    {
      "epoch": 2.2419919792882887,
      "grad_norm": 0.40568065643310547,
      "learning_rate": 5.0533868047447425e-06,
      "loss": 0.0375,
      "step": 44165
    },
    {
      "epoch": 2.2422457992791514,
      "grad_norm": 0.411802738904953,
      "learning_rate": 5.051694671472326e-06,
      "loss": 0.0327,
      "step": 44170
    },
    {
      "epoch": 2.2424996192700135,
      "grad_norm": 0.347493439912796,
      "learning_rate": 5.050002538199909e-06,
      "loss": 0.0342,
      "step": 44175
    },
    {
      "epoch": 2.242753439260876,
      "grad_norm": 0.4209951162338257,
      "learning_rate": 5.048310404927492e-06,
      "loss": 0.0301,
      "step": 44180
    },
    {
      "epoch": 2.2430072592517387,
      "grad_norm": 0.4024890959262848,
      "learning_rate": 5.046618271655076e-06,
      "loss": 0.0287,
      "step": 44185
    },
    {
      "epoch": 2.2432610792426013,
      "grad_norm": 0.26085108518600464,
      "learning_rate": 5.04492613838266e-06,
      "loss": 0.0278,
      "step": 44190
    },
    {
      "epoch": 2.2435148992334635,
      "grad_norm": 0.33618801832199097,
      "learning_rate": 5.043234005110242e-06,
      "loss": 0.0264,
      "step": 44195
    },
    {
      "epoch": 2.243768719224326,
      "grad_norm": 0.31238695979118347,
      "learning_rate": 5.041541871837826e-06,
      "loss": 0.0311,
      "step": 44200
    },
    {
      "epoch": 2.2440225392151887,
      "grad_norm": 0.29754072427749634,
      "learning_rate": 5.03984973856541e-06,
      "loss": 0.0297,
      "step": 44205
    },
    {
      "epoch": 2.244276359206051,
      "grad_norm": 0.2515721917152405,
      "learning_rate": 5.0381576052929935e-06,
      "loss": 0.0347,
      "step": 44210
    },
    {
      "epoch": 2.2445301791969134,
      "grad_norm": 0.40768948197364807,
      "learning_rate": 5.036465472020576e-06,
      "loss": 0.0414,
      "step": 44215
    },
    {
      "epoch": 2.244783999187776,
      "grad_norm": 0.34544774889945984,
      "learning_rate": 5.03477333874816e-06,
      "loss": 0.0377,
      "step": 44220
    },
    {
      "epoch": 2.2450378191786386,
      "grad_norm": 0.34288665652275085,
      "learning_rate": 5.033081205475744e-06,
      "loss": 0.0355,
      "step": 44225
    },
    {
      "epoch": 2.245291639169501,
      "grad_norm": 0.36351481080055237,
      "learning_rate": 5.031389072203328e-06,
      "loss": 0.0385,
      "step": 44230
    },
    {
      "epoch": 2.2455454591603634,
      "grad_norm": 0.2494805008172989,
      "learning_rate": 5.02969693893091e-06,
      "loss": 0.0334,
      "step": 44235
    },
    {
      "epoch": 2.245799279151226,
      "grad_norm": 0.3414851725101471,
      "learning_rate": 5.028004805658494e-06,
      "loss": 0.0328,
      "step": 44240
    },
    {
      "epoch": 2.2460530991420886,
      "grad_norm": 0.3112623989582062,
      "learning_rate": 5.0263126723860775e-06,
      "loss": 0.0365,
      "step": 44245
    },
    {
      "epoch": 2.2463069191329508,
      "grad_norm": 0.4333570897579193,
      "learning_rate": 5.024620539113662e-06,
      "loss": 0.0383,
      "step": 44250
    },
    {
      "epoch": 2.2465607391238134,
      "grad_norm": 0.49220526218414307,
      "learning_rate": 5.0229284058412446e-06,
      "loss": 0.0297,
      "step": 44255
    },
    {
      "epoch": 2.246814559114676,
      "grad_norm": 0.2645159959793091,
      "learning_rate": 5.021236272568828e-06,
      "loss": 0.0296,
      "step": 44260
    },
    {
      "epoch": 2.2470683791055386,
      "grad_norm": 0.36111289262771606,
      "learning_rate": 5.019544139296412e-06,
      "loss": 0.0316,
      "step": 44265
    },
    {
      "epoch": 2.2473221990964007,
      "grad_norm": 0.3091657757759094,
      "learning_rate": 5.017852006023994e-06,
      "loss": 0.0355,
      "step": 44270
    },
    {
      "epoch": 2.2475760190872633,
      "grad_norm": 0.29647141695022583,
      "learning_rate": 5.016159872751578e-06,
      "loss": 0.028,
      "step": 44275
    },
    {
      "epoch": 2.247829839078126,
      "grad_norm": 0.5130562782287598,
      "learning_rate": 5.014467739479162e-06,
      "loss": 0.0334,
      "step": 44280
    },
    {
      "epoch": 2.248083659068988,
      "grad_norm": 0.21307742595672607,
      "learning_rate": 5.012775606206746e-06,
      "loss": 0.0275,
      "step": 44285
    },
    {
      "epoch": 2.2483374790598507,
      "grad_norm": 0.3160382807254791,
      "learning_rate": 5.0110834729343285e-06,
      "loss": 0.0381,
      "step": 44290
    },
    {
      "epoch": 2.2485912990507133,
      "grad_norm": 0.4054400324821472,
      "learning_rate": 5.009391339661912e-06,
      "loss": 0.0332,
      "step": 44295
    },
    {
      "epoch": 2.248845119041576,
      "grad_norm": 0.39842960238456726,
      "learning_rate": 5.007699206389496e-06,
      "loss": 0.0335,
      "step": 44300
    },
    {
      "epoch": 2.249098939032438,
      "grad_norm": 0.5165933966636658,
      "learning_rate": 5.006007073117079e-06,
      "loss": 0.0328,
      "step": 44305
    },
    {
      "epoch": 2.2493527590233007,
      "grad_norm": 0.42537662386894226,
      "learning_rate": 5.004314939844662e-06,
      "loss": 0.0329,
      "step": 44310
    },
    {
      "epoch": 2.2496065790141633,
      "grad_norm": 0.4120660126209259,
      "learning_rate": 5.002622806572246e-06,
      "loss": 0.03,
      "step": 44315
    },
    {
      "epoch": 2.2498603990050254,
      "grad_norm": 0.30993524193763733,
      "learning_rate": 5.00093067329983e-06,
      "loss": 0.0316,
      "step": 44320
    },
    {
      "epoch": 2.250114218995888,
      "grad_norm": 0.4453542232513428,
      "learning_rate": 4.999238540027413e-06,
      "loss": 0.0436,
      "step": 44325
    },
    {
      "epoch": 2.2503680389867506,
      "grad_norm": 0.40560978651046753,
      "learning_rate": 4.997546406754997e-06,
      "loss": 0.0419,
      "step": 44330
    },
    {
      "epoch": 2.2506218589776132,
      "grad_norm": 0.3937535285949707,
      "learning_rate": 4.99585427348258e-06,
      "loss": 0.0367,
      "step": 44335
    },
    {
      "epoch": 2.2508756789684754,
      "grad_norm": 0.31118619441986084,
      "learning_rate": 4.994162140210163e-06,
      "loss": 0.0356,
      "step": 44340
    },
    {
      "epoch": 2.251129498959338,
      "grad_norm": 0.34185513854026794,
      "learning_rate": 4.992470006937747e-06,
      "loss": 0.0341,
      "step": 44345
    },
    {
      "epoch": 2.2513833189502006,
      "grad_norm": 0.21117255091667175,
      "learning_rate": 4.99077787366533e-06,
      "loss": 0.0283,
      "step": 44350
    },
    {
      "epoch": 2.2516371389410628,
      "grad_norm": 0.2540547847747803,
      "learning_rate": 4.989085740392914e-06,
      "loss": 0.0348,
      "step": 44355
    },
    {
      "epoch": 2.2518909589319254,
      "grad_norm": 0.2739291489124298,
      "learning_rate": 4.987393607120497e-06,
      "loss": 0.0227,
      "step": 44360
    },
    {
      "epoch": 2.252144778922788,
      "grad_norm": 0.32483914494514465,
      "learning_rate": 4.985701473848081e-06,
      "loss": 0.0335,
      "step": 44365
    },
    {
      "epoch": 2.2523985989136506,
      "grad_norm": 0.36847808957099915,
      "learning_rate": 4.984009340575664e-06,
      "loss": 0.0349,
      "step": 44370
    },
    {
      "epoch": 2.2526524189045127,
      "grad_norm": 0.5985268354415894,
      "learning_rate": 4.982317207303248e-06,
      "loss": 0.0317,
      "step": 44375
    },
    {
      "epoch": 2.2529062388953753,
      "grad_norm": 0.3975210189819336,
      "learning_rate": 4.980625074030831e-06,
      "loss": 0.033,
      "step": 44380
    },
    {
      "epoch": 2.253160058886238,
      "grad_norm": 0.27625125646591187,
      "learning_rate": 4.978932940758415e-06,
      "loss": 0.0316,
      "step": 44385
    },
    {
      "epoch": 2.2534138788771005,
      "grad_norm": 0.2550783157348633,
      "learning_rate": 4.977240807485998e-06,
      "loss": 0.0329,
      "step": 44390
    },
    {
      "epoch": 2.2536676988679627,
      "grad_norm": 0.5757700204849243,
      "learning_rate": 4.975548674213581e-06,
      "loss": 0.0353,
      "step": 44395
    },
    {
      "epoch": 2.2539215188588253,
      "grad_norm": 0.2964254319667816,
      "learning_rate": 4.973856540941165e-06,
      "loss": 0.0286,
      "step": 44400
    },
    {
      "epoch": 2.254175338849688,
      "grad_norm": 0.3224160075187683,
      "learning_rate": 4.972164407668748e-06,
      "loss": 0.0305,
      "step": 44405
    },
    {
      "epoch": 2.2544291588405505,
      "grad_norm": 0.3136308789253235,
      "learning_rate": 4.970472274396332e-06,
      "loss": 0.0342,
      "step": 44410
    },
    {
      "epoch": 2.2546829788314127,
      "grad_norm": 0.30831313133239746,
      "learning_rate": 4.9687801411239155e-06,
      "loss": 0.0303,
      "step": 44415
    },
    {
      "epoch": 2.2549367988222753,
      "grad_norm": 0.3761250078678131,
      "learning_rate": 4.967088007851499e-06,
      "loss": 0.0355,
      "step": 44420
    },
    {
      "epoch": 2.255190618813138,
      "grad_norm": 0.4608723521232605,
      "learning_rate": 4.965395874579082e-06,
      "loss": 0.0349,
      "step": 44425
    },
    {
      "epoch": 2.255444438804,
      "grad_norm": 0.35675951838493347,
      "learning_rate": 4.963703741306666e-06,
      "loss": 0.0417,
      "step": 44430
    },
    {
      "epoch": 2.2556982587948626,
      "grad_norm": 0.3406544029712677,
      "learning_rate": 4.962011608034249e-06,
      "loss": 0.0341,
      "step": 44435
    },
    {
      "epoch": 2.2559520787857252,
      "grad_norm": 0.4340391755104065,
      "learning_rate": 4.960319474761832e-06,
      "loss": 0.0235,
      "step": 44440
    },
    {
      "epoch": 2.256205898776588,
      "grad_norm": 0.2823371887207031,
      "learning_rate": 4.958627341489416e-06,
      "loss": 0.029,
      "step": 44445
    },
    {
      "epoch": 2.25645971876745,
      "grad_norm": 0.2538403272628784,
      "learning_rate": 4.956935208216999e-06,
      "loss": 0.0271,
      "step": 44450
    },
    {
      "epoch": 2.2567135387583126,
      "grad_norm": 0.3238656520843506,
      "learning_rate": 4.955243074944583e-06,
      "loss": 0.0282,
      "step": 44455
    },
    {
      "epoch": 2.256967358749175,
      "grad_norm": 0.4126463234424591,
      "learning_rate": 4.9535509416721665e-06,
      "loss": 0.0276,
      "step": 44460
    },
    {
      "epoch": 2.2572211787400374,
      "grad_norm": 0.6289737820625305,
      "learning_rate": 4.95185880839975e-06,
      "loss": 0.0316,
      "step": 44465
    },
    {
      "epoch": 2.2574749987309,
      "grad_norm": 0.3341315686702728,
      "learning_rate": 4.950166675127334e-06,
      "loss": 0.0298,
      "step": 44470
    },
    {
      "epoch": 2.2577288187217626,
      "grad_norm": 0.44910258054733276,
      "learning_rate": 4.948474541854917e-06,
      "loss": 0.035,
      "step": 44475
    },
    {
      "epoch": 2.257982638712625,
      "grad_norm": 0.261214941740036,
      "learning_rate": 4.946782408582501e-06,
      "loss": 0.0373,
      "step": 44480
    },
    {
      "epoch": 2.2582364587034873,
      "grad_norm": 0.44271722435951233,
      "learning_rate": 4.945090275310083e-06,
      "loss": 0.0282,
      "step": 44485
    },
    {
      "epoch": 2.25849027869435,
      "grad_norm": 0.34975486993789673,
      "learning_rate": 4.943398142037668e-06,
      "loss": 0.0289,
      "step": 44490
    },
    {
      "epoch": 2.2587440986852125,
      "grad_norm": 0.6106913089752197,
      "learning_rate": 4.9417060087652505e-06,
      "loss": 0.034,
      "step": 44495
    },
    {
      "epoch": 2.2589979186760747,
      "grad_norm": 0.25883862376213074,
      "learning_rate": 4.940013875492834e-06,
      "loss": 0.038,
      "step": 44500
    },
    {
      "epoch": 2.2592517386669373,
      "grad_norm": 0.39152991771698,
      "learning_rate": 4.9383217422204176e-06,
      "loss": 0.0349,
      "step": 44505
    },
    {
      "epoch": 2.2595055586578,
      "grad_norm": 0.3380178213119507,
      "learning_rate": 4.936629608948001e-06,
      "loss": 0.0342,
      "step": 44510
    },
    {
      "epoch": 2.2597593786486625,
      "grad_norm": 0.36176571249961853,
      "learning_rate": 4.934937475675585e-06,
      "loss": 0.0309,
      "step": 44515
    },
    {
      "epoch": 2.2600131986395247,
      "grad_norm": 0.6492758989334106,
      "learning_rate": 4.933245342403168e-06,
      "loss": 0.0348,
      "step": 44520
    },
    {
      "epoch": 2.2602670186303873,
      "grad_norm": 0.3978842794895172,
      "learning_rate": 4.931553209130752e-06,
      "loss": 0.0379,
      "step": 44525
    },
    {
      "epoch": 2.26052083862125,
      "grad_norm": 0.40266185998916626,
      "learning_rate": 4.9298610758583344e-06,
      "loss": 0.0331,
      "step": 44530
    },
    {
      "epoch": 2.2607746586121125,
      "grad_norm": 0.262500524520874,
      "learning_rate": 4.928168942585919e-06,
      "loss": 0.0352,
      "step": 44535
    },
    {
      "epoch": 2.2610284786029746,
      "grad_norm": 0.389607697725296,
      "learning_rate": 4.9264768093135015e-06,
      "loss": 0.0291,
      "step": 44540
    },
    {
      "epoch": 2.2612822985938372,
      "grad_norm": 0.24380935728549957,
      "learning_rate": 4.924784676041086e-06,
      "loss": 0.0312,
      "step": 44545
    },
    {
      "epoch": 2.2615361185847,
      "grad_norm": 0.5458943247795105,
      "learning_rate": 4.923092542768669e-06,
      "loss": 0.0284,
      "step": 44550
    },
    {
      "epoch": 2.2617899385755624,
      "grad_norm": 1.1159203052520752,
      "learning_rate": 4.921400409496252e-06,
      "loss": 0.0305,
      "step": 44555
    },
    {
      "epoch": 2.2620437585664246,
      "grad_norm": 0.3896821141242981,
      "learning_rate": 4.919708276223836e-06,
      "loss": 0.0387,
      "step": 44560
    },
    {
      "epoch": 2.262297578557287,
      "grad_norm": 0.3972178101539612,
      "learning_rate": 4.918016142951419e-06,
      "loss": 0.035,
      "step": 44565
    },
    {
      "epoch": 2.26255139854815,
      "grad_norm": 0.3106682598590851,
      "learning_rate": 4.916324009679003e-06,
      "loss": 0.0276,
      "step": 44570
    },
    {
      "epoch": 2.262805218539012,
      "grad_norm": 0.24602703750133514,
      "learning_rate": 4.914631876406586e-06,
      "loss": 0.0354,
      "step": 44575
    },
    {
      "epoch": 2.2630590385298746,
      "grad_norm": 0.26132622361183167,
      "learning_rate": 4.91293974313417e-06,
      "loss": 0.0274,
      "step": 44580
    },
    {
      "epoch": 2.263312858520737,
      "grad_norm": 0.34328407049179077,
      "learning_rate": 4.911247609861753e-06,
      "loss": 0.029,
      "step": 44585
    },
    {
      "epoch": 2.2635666785115998,
      "grad_norm": 0.3113946318626404,
      "learning_rate": 4.909555476589337e-06,
      "loss": 0.034,
      "step": 44590
    },
    {
      "epoch": 2.263820498502462,
      "grad_norm": 0.24702712893486023,
      "learning_rate": 4.90786334331692e-06,
      "loss": 0.033,
      "step": 44595
    },
    {
      "epoch": 2.2640743184933245,
      "grad_norm": 0.4547874629497528,
      "learning_rate": 4.906171210044503e-06,
      "loss": 0.0334,
      "step": 44600
    },
    {
      "epoch": 2.264328138484187,
      "grad_norm": 0.31980690360069275,
      "learning_rate": 4.904479076772087e-06,
      "loss": 0.0291,
      "step": 44605
    },
    {
      "epoch": 2.2645819584750493,
      "grad_norm": 0.28869301080703735,
      "learning_rate": 4.90278694349967e-06,
      "loss": 0.0295,
      "step": 44610
    },
    {
      "epoch": 2.264835778465912,
      "grad_norm": 0.3857107162475586,
      "learning_rate": 4.901094810227254e-06,
      "loss": 0.0398,
      "step": 44615
    },
    {
      "epoch": 2.2650895984567745,
      "grad_norm": 0.25656676292419434,
      "learning_rate": 4.899402676954837e-06,
      "loss": 0.032,
      "step": 44620
    },
    {
      "epoch": 2.265343418447637,
      "grad_norm": 0.24995790421962738,
      "learning_rate": 4.897710543682421e-06,
      "loss": 0.0271,
      "step": 44625
    },
    {
      "epoch": 2.2655972384384993,
      "grad_norm": 0.4084688723087311,
      "learning_rate": 4.8960184104100045e-06,
      "loss": 0.0327,
      "step": 44630
    },
    {
      "epoch": 2.265851058429362,
      "grad_norm": 0.2487572878599167,
      "learning_rate": 4.894326277137588e-06,
      "loss": 0.0308,
      "step": 44635
    },
    {
      "epoch": 2.2661048784202245,
      "grad_norm": 0.6064757108688354,
      "learning_rate": 4.8926341438651716e-06,
      "loss": 0.0287,
      "step": 44640
    },
    {
      "epoch": 2.2663586984110866,
      "grad_norm": 0.37674620747566223,
      "learning_rate": 4.890942010592754e-06,
      "loss": 0.0289,
      "step": 44645
    },
    {
      "epoch": 2.266612518401949,
      "grad_norm": 0.27238699793815613,
      "learning_rate": 4.889249877320339e-06,
      "loss": 0.0316,
      "step": 44650
    },
    {
      "epoch": 2.266866338392812,
      "grad_norm": 0.3761901557445526,
      "learning_rate": 4.887557744047921e-06,
      "loss": 0.0374,
      "step": 44655
    },
    {
      "epoch": 2.2671201583836744,
      "grad_norm": 0.4535421133041382,
      "learning_rate": 4.885865610775505e-06,
      "loss": 0.0327,
      "step": 44660
    },
    {
      "epoch": 2.267373978374537,
      "grad_norm": 0.3541247248649597,
      "learning_rate": 4.8841734775030884e-06,
      "loss": 0.032,
      "step": 44665
    },
    {
      "epoch": 2.267627798365399,
      "grad_norm": 0.2639147639274597,
      "learning_rate": 4.882481344230672e-06,
      "loss": 0.029,
      "step": 44670
    },
    {
      "epoch": 2.267881618356262,
      "grad_norm": 0.3500845432281494,
      "learning_rate": 4.8807892109582555e-06,
      "loss": 0.0335,
      "step": 44675
    },
    {
      "epoch": 2.2681354383471244,
      "grad_norm": 0.22339092195034027,
      "learning_rate": 4.879097077685839e-06,
      "loss": 0.0304,
      "step": 44680
    },
    {
      "epoch": 2.2683892583379865,
      "grad_norm": 0.2681799829006195,
      "learning_rate": 4.877404944413423e-06,
      "loss": 0.0327,
      "step": 44685
    },
    {
      "epoch": 2.268643078328849,
      "grad_norm": 0.35728010535240173,
      "learning_rate": 4.875712811141005e-06,
      "loss": 0.0318,
      "step": 44690
    },
    {
      "epoch": 2.2688968983197118,
      "grad_norm": 0.5533677935600281,
      "learning_rate": 4.87402067786859e-06,
      "loss": 0.0297,
      "step": 44695
    },
    {
      "epoch": 2.2691507183105744,
      "grad_norm": 0.3111599087715149,
      "learning_rate": 4.872328544596172e-06,
      "loss": 0.032,
      "step": 44700
    },
    {
      "epoch": 2.2694045383014365,
      "grad_norm": 0.5662405490875244,
      "learning_rate": 4.870636411323757e-06,
      "loss": 0.0449,
      "step": 44705
    },
    {
      "epoch": 2.269658358292299,
      "grad_norm": 0.33681607246398926,
      "learning_rate": 4.8689442780513395e-06,
      "loss": 0.0352,
      "step": 44710
    },
    {
      "epoch": 2.2699121782831617,
      "grad_norm": 0.33438214659690857,
      "learning_rate": 4.867252144778923e-06,
      "loss": 0.033,
      "step": 44715
    },
    {
      "epoch": 2.270165998274024,
      "grad_norm": 0.49922633171081543,
      "learning_rate": 4.865560011506507e-06,
      "loss": 0.035,
      "step": 44720
    },
    {
      "epoch": 2.2704198182648865,
      "grad_norm": 0.32202664017677307,
      "learning_rate": 4.86386787823409e-06,
      "loss": 0.0364,
      "step": 44725
    },
    {
      "epoch": 2.270673638255749,
      "grad_norm": 0.41406023502349854,
      "learning_rate": 4.862175744961674e-06,
      "loss": 0.0374,
      "step": 44730
    },
    {
      "epoch": 2.2709274582466117,
      "grad_norm": 0.404519647359848,
      "learning_rate": 4.860483611689257e-06,
      "loss": 0.0337,
      "step": 44735
    },
    {
      "epoch": 2.271181278237474,
      "grad_norm": 0.3270086348056793,
      "learning_rate": 4.858791478416841e-06,
      "loss": 0.0401,
      "step": 44740
    },
    {
      "epoch": 2.2714350982283364,
      "grad_norm": 0.3245806097984314,
      "learning_rate": 4.8570993451444235e-06,
      "loss": 0.0311,
      "step": 44745
    },
    {
      "epoch": 2.271688918219199,
      "grad_norm": 0.1995822787284851,
      "learning_rate": 4.855407211872008e-06,
      "loss": 0.032,
      "step": 44750
    },
    {
      "epoch": 2.271942738210061,
      "grad_norm": 0.3266526758670807,
      "learning_rate": 4.8537150785995906e-06,
      "loss": 0.0243,
      "step": 44755
    },
    {
      "epoch": 2.272196558200924,
      "grad_norm": 0.26783430576324463,
      "learning_rate": 4.852022945327174e-06,
      "loss": 0.0273,
      "step": 44760
    },
    {
      "epoch": 2.2724503781917864,
      "grad_norm": 0.4026534855365753,
      "learning_rate": 4.850330812054758e-06,
      "loss": 0.0275,
      "step": 44765
    },
    {
      "epoch": 2.272704198182649,
      "grad_norm": 0.33564287424087524,
      "learning_rate": 4.848638678782341e-06,
      "loss": 0.0362,
      "step": 44770
    },
    {
      "epoch": 2.272958018173511,
      "grad_norm": 0.57597815990448,
      "learning_rate": 4.846946545509925e-06,
      "loss": 0.0407,
      "step": 44775
    },
    {
      "epoch": 2.2732118381643738,
      "grad_norm": 0.3466244637966156,
      "learning_rate": 4.845254412237508e-06,
      "loss": 0.0361,
      "step": 44780
    },
    {
      "epoch": 2.2734656581552364,
      "grad_norm": 0.21853549778461456,
      "learning_rate": 4.843562278965092e-06,
      "loss": 0.0273,
      "step": 44785
    },
    {
      "epoch": 2.2737194781460985,
      "grad_norm": 0.32469242811203003,
      "learning_rate": 4.841870145692675e-06,
      "loss": 0.0289,
      "step": 44790
    },
    {
      "epoch": 2.273973298136961,
      "grad_norm": 0.330661416053772,
      "learning_rate": 4.840178012420259e-06,
      "loss": 0.0345,
      "step": 44795
    },
    {
      "epoch": 2.2742271181278237,
      "grad_norm": 0.28515830636024475,
      "learning_rate": 4.8384858791478425e-06,
      "loss": 0.0297,
      "step": 44800
    },
    {
      "epoch": 2.2744809381186863,
      "grad_norm": 0.4901011288166046,
      "learning_rate": 4.836793745875425e-06,
      "loss": 0.031,
      "step": 44805
    },
    {
      "epoch": 2.274734758109549,
      "grad_norm": 0.31480327248573303,
      "learning_rate": 4.8351016126030095e-06,
      "loss": 0.0382,
      "step": 44810
    },
    {
      "epoch": 2.274988578100411,
      "grad_norm": 0.3091256618499756,
      "learning_rate": 4.833409479330592e-06,
      "loss": 0.0379,
      "step": 44815
    },
    {
      "epoch": 2.2752423980912737,
      "grad_norm": 0.2778155207633972,
      "learning_rate": 4.831717346058176e-06,
      "loss": 0.035,
      "step": 44820
    },
    {
      "epoch": 2.2754962180821363,
      "grad_norm": 0.41067826747894287,
      "learning_rate": 4.830025212785759e-06,
      "loss": 0.033,
      "step": 44825
    },
    {
      "epoch": 2.2757500380729985,
      "grad_norm": 0.49526798725128174,
      "learning_rate": 4.828333079513343e-06,
      "loss": 0.0311,
      "step": 44830
    },
    {
      "epoch": 2.276003858063861,
      "grad_norm": 0.3333423137664795,
      "learning_rate": 4.826640946240926e-06,
      "loss": 0.028,
      "step": 44835
    },
    {
      "epoch": 2.2762576780547237,
      "grad_norm": 0.386203795671463,
      "learning_rate": 4.82494881296851e-06,
      "loss": 0.0348,
      "step": 44840
    },
    {
      "epoch": 2.2765114980455863,
      "grad_norm": 0.3671344816684723,
      "learning_rate": 4.8232566796960935e-06,
      "loss": 0.0344,
      "step": 44845
    },
    {
      "epoch": 2.2767653180364484,
      "grad_norm": 0.50922030210495,
      "learning_rate": 4.821564546423676e-06,
      "loss": 0.0308,
      "step": 44850
    },
    {
      "epoch": 2.277019138027311,
      "grad_norm": 0.228525772690773,
      "learning_rate": 4.819872413151261e-06,
      "loss": 0.036,
      "step": 44855
    },
    {
      "epoch": 2.2772729580181736,
      "grad_norm": 0.41872110962867737,
      "learning_rate": 4.818180279878843e-06,
      "loss": 0.0392,
      "step": 44860
    },
    {
      "epoch": 2.277526778009036,
      "grad_norm": 0.32197269797325134,
      "learning_rate": 4.816488146606428e-06,
      "loss": 0.0307,
      "step": 44865
    },
    {
      "epoch": 2.2777805979998984,
      "grad_norm": 0.2782743573188782,
      "learning_rate": 4.81479601333401e-06,
      "loss": 0.0312,
      "step": 44870
    },
    {
      "epoch": 2.278034417990761,
      "grad_norm": 0.40119466185569763,
      "learning_rate": 4.813103880061594e-06,
      "loss": 0.027,
      "step": 44875
    },
    {
      "epoch": 2.2782882379816236,
      "grad_norm": 0.34341195225715637,
      "learning_rate": 4.8114117467891775e-06,
      "loss": 0.0322,
      "step": 44880
    },
    {
      "epoch": 2.2785420579724858,
      "grad_norm": 0.4127197563648224,
      "learning_rate": 4.809719613516761e-06,
      "loss": 0.028,
      "step": 44885
    },
    {
      "epoch": 2.2787958779633484,
      "grad_norm": 0.4768372178077698,
      "learning_rate": 4.8080274802443446e-06,
      "loss": 0.0325,
      "step": 44890
    },
    {
      "epoch": 2.279049697954211,
      "grad_norm": 0.30066752433776855,
      "learning_rate": 4.806335346971928e-06,
      "loss": 0.0342,
      "step": 44895
    },
    {
      "epoch": 2.279303517945073,
      "grad_norm": 0.3694850504398346,
      "learning_rate": 4.804643213699512e-06,
      "loss": 0.0417,
      "step": 44900
    },
    {
      "epoch": 2.2795573379359357,
      "grad_norm": 0.3389255404472351,
      "learning_rate": 4.802951080427094e-06,
      "loss": 0.0328,
      "step": 44905
    },
    {
      "epoch": 2.2798111579267983,
      "grad_norm": 0.32273051142692566,
      "learning_rate": 4.801258947154679e-06,
      "loss": 0.0302,
      "step": 44910
    },
    {
      "epoch": 2.280064977917661,
      "grad_norm": 0.2565813660621643,
      "learning_rate": 4.7995668138822614e-06,
      "loss": 0.0373,
      "step": 44915
    },
    {
      "epoch": 2.280318797908523,
      "grad_norm": 0.17176474630832672,
      "learning_rate": 4.797874680609845e-06,
      "loss": 0.0308,
      "step": 44920
    },
    {
      "epoch": 2.2805726178993857,
      "grad_norm": 0.28532516956329346,
      "learning_rate": 4.7961825473374285e-06,
      "loss": 0.0294,
      "step": 44925
    },
    {
      "epoch": 2.2808264378902483,
      "grad_norm": 0.3874659836292267,
      "learning_rate": 4.794490414065012e-06,
      "loss": 0.0294,
      "step": 44930
    },
    {
      "epoch": 2.281080257881111,
      "grad_norm": 0.3868589997291565,
      "learning_rate": 4.792798280792596e-06,
      "loss": 0.0309,
      "step": 44935
    },
    {
      "epoch": 2.281334077871973,
      "grad_norm": 0.37089648842811584,
      "learning_rate": 4.791106147520179e-06,
      "loss": 0.032,
      "step": 44940
    },
    {
      "epoch": 2.2815878978628357,
      "grad_norm": 0.28405794501304626,
      "learning_rate": 4.789414014247763e-06,
      "loss": 0.0304,
      "step": 44945
    },
    {
      "epoch": 2.2818417178536983,
      "grad_norm": 0.316731721162796,
      "learning_rate": 4.787721880975346e-06,
      "loss": 0.0359,
      "step": 44950
    },
    {
      "epoch": 2.282095537844561,
      "grad_norm": 0.6578758358955383,
      "learning_rate": 4.78602974770293e-06,
      "loss": 0.0352,
      "step": 44955
    },
    {
      "epoch": 2.282349357835423,
      "grad_norm": 0.41155996918678284,
      "learning_rate": 4.784337614430513e-06,
      "loss": 0.0313,
      "step": 44960
    },
    {
      "epoch": 2.2826031778262856,
      "grad_norm": 0.33022230863571167,
      "learning_rate": 4.782645481158096e-06,
      "loss": 0.0244,
      "step": 44965
    },
    {
      "epoch": 2.2828569978171482,
      "grad_norm": 0.285152792930603,
      "learning_rate": 4.7809533478856804e-06,
      "loss": 0.0289,
      "step": 44970
    },
    {
      "epoch": 2.2831108178080104,
      "grad_norm": 0.30267083644866943,
      "learning_rate": 4.779261214613263e-06,
      "loss": 0.0341,
      "step": 44975
    },
    {
      "epoch": 2.283364637798873,
      "grad_norm": 1.3061234951019287,
      "learning_rate": 4.777569081340847e-06,
      "loss": 0.0311,
      "step": 44980
    },
    {
      "epoch": 2.2836184577897356,
      "grad_norm": 0.2621577978134155,
      "learning_rate": 4.77587694806843e-06,
      "loss": 0.0288,
      "step": 44985
    },
    {
      "epoch": 2.283872277780598,
      "grad_norm": 0.5385904312133789,
      "learning_rate": 4.774184814796014e-06,
      "loss": 0.037,
      "step": 44990
    },
    {
      "epoch": 2.2841260977714604,
      "grad_norm": 0.42236170172691345,
      "learning_rate": 4.772492681523597e-06,
      "loss": 0.0367,
      "step": 44995
    },
    {
      "epoch": 2.284379917762323,
      "grad_norm": 0.35834038257598877,
      "learning_rate": 4.770800548251181e-06,
      "loss": 0.0349,
      "step": 45000
    },
    {
      "epoch": 2.2846337377531856,
      "grad_norm": 0.39529144763946533,
      "learning_rate": 4.769108414978764e-06,
      "loss": 0.0393,
      "step": 45005
    },
    {
      "epoch": 2.2848875577440477,
      "grad_norm": 0.3461845815181732,
      "learning_rate": 4.767416281706347e-06,
      "loss": 0.0359,
      "step": 45010
    },
    {
      "epoch": 2.2851413777349103,
      "grad_norm": 0.2342214435338974,
      "learning_rate": 4.7657241484339315e-06,
      "loss": 0.0258,
      "step": 45015
    },
    {
      "epoch": 2.285395197725773,
      "grad_norm": 0.4008984863758087,
      "learning_rate": 4.764032015161514e-06,
      "loss": 0.0345,
      "step": 45020
    },
    {
      "epoch": 2.2856490177166355,
      "grad_norm": 0.3111340403556824,
      "learning_rate": 4.762339881889098e-06,
      "loss": 0.0428,
      "step": 45025
    },
    {
      "epoch": 2.2859028377074977,
      "grad_norm": 0.2649993300437927,
      "learning_rate": 4.760647748616681e-06,
      "loss": 0.027,
      "step": 45030
    },
    {
      "epoch": 2.2861566576983603,
      "grad_norm": 0.8399127721786499,
      "learning_rate": 4.758955615344265e-06,
      "loss": 0.0369,
      "step": 45035
    },
    {
      "epoch": 2.286410477689223,
      "grad_norm": 0.3271864354610443,
      "learning_rate": 4.757263482071848e-06,
      "loss": 0.0268,
      "step": 45040
    },
    {
      "epoch": 2.286664297680085,
      "grad_norm": 0.35605019330978394,
      "learning_rate": 4.755571348799432e-06,
      "loss": 0.0318,
      "step": 45045
    },
    {
      "epoch": 2.2869181176709477,
      "grad_norm": 0.25064730644226074,
      "learning_rate": 4.7538792155270154e-06,
      "loss": 0.026,
      "step": 45050
    },
    {
      "epoch": 2.2871719376618103,
      "grad_norm": 0.2588244676589966,
      "learning_rate": 4.752187082254599e-06,
      "loss": 0.0283,
      "step": 45055
    },
    {
      "epoch": 2.287425757652673,
      "grad_norm": 0.4669940769672394,
      "learning_rate": 4.7504949489821825e-06,
      "loss": 0.0341,
      "step": 45060
    },
    {
      "epoch": 2.287679577643535,
      "grad_norm": 0.2035631388425827,
      "learning_rate": 4.748802815709765e-06,
      "loss": 0.0323,
      "step": 45065
    },
    {
      "epoch": 2.2879333976343976,
      "grad_norm": 0.29153090715408325,
      "learning_rate": 4.747110682437349e-06,
      "loss": 0.0329,
      "step": 45070
    },
    {
      "epoch": 2.2881872176252602,
      "grad_norm": 0.5664026737213135,
      "learning_rate": 4.745418549164932e-06,
      "loss": 0.0354,
      "step": 45075
    },
    {
      "epoch": 2.288441037616123,
      "grad_norm": 0.31028592586517334,
      "learning_rate": 4.743726415892516e-06,
      "loss": 0.0361,
      "step": 45080
    },
    {
      "epoch": 2.288694857606985,
      "grad_norm": 0.29820579290390015,
      "learning_rate": 4.742034282620099e-06,
      "loss": 0.0323,
      "step": 45085
    },
    {
      "epoch": 2.2889486775978476,
      "grad_norm": 0.3673454523086548,
      "learning_rate": 4.740342149347683e-06,
      "loss": 0.032,
      "step": 45090
    },
    {
      "epoch": 2.28920249758871,
      "grad_norm": 0.35142046213150024,
      "learning_rate": 4.7386500160752665e-06,
      "loss": 0.0338,
      "step": 45095
    },
    {
      "epoch": 2.289456317579573,
      "grad_norm": 0.293951153755188,
      "learning_rate": 4.73695788280285e-06,
      "loss": 0.0291,
      "step": 45100
    },
    {
      "epoch": 2.289710137570435,
      "grad_norm": 0.3154962956905365,
      "learning_rate": 4.735265749530434e-06,
      "loss": 0.0289,
      "step": 45105
    },
    {
      "epoch": 2.2899639575612976,
      "grad_norm": 0.3382570147514343,
      "learning_rate": 4.733573616258017e-06,
      "loss": 0.0348,
      "step": 45110
    },
    {
      "epoch": 2.29021777755216,
      "grad_norm": 0.3350144624710083,
      "learning_rate": 4.7318814829856e-06,
      "loss": 0.0279,
      "step": 45115
    },
    {
      "epoch": 2.2904715975430223,
      "grad_norm": 0.5217757821083069,
      "learning_rate": 4.730189349713184e-06,
      "loss": 0.0328,
      "step": 45120
    },
    {
      "epoch": 2.290725417533885,
      "grad_norm": 0.34693512320518494,
      "learning_rate": 4.728497216440767e-06,
      "loss": 0.0329,
      "step": 45125
    },
    {
      "epoch": 2.2909792375247475,
      "grad_norm": 0.41887202858924866,
      "learning_rate": 4.726805083168351e-06,
      "loss": 0.0348,
      "step": 45130
    },
    {
      "epoch": 2.29123305751561,
      "grad_norm": 0.3702602684497833,
      "learning_rate": 4.725112949895934e-06,
      "loss": 0.0312,
      "step": 45135
    },
    {
      "epoch": 2.2914868775064723,
      "grad_norm": 0.4091658294200897,
      "learning_rate": 4.7234208166235176e-06,
      "loss": 0.0283,
      "step": 45140
    },
    {
      "epoch": 2.291740697497335,
      "grad_norm": 0.46928733587265015,
      "learning_rate": 4.721728683351101e-06,
      "loss": 0.0297,
      "step": 45145
    },
    {
      "epoch": 2.2919945174881975,
      "grad_norm": 0.31767764687538147,
      "learning_rate": 4.720036550078685e-06,
      "loss": 0.0268,
      "step": 45150
    },
    {
      "epoch": 2.2922483374790596,
      "grad_norm": 0.2629365622997284,
      "learning_rate": 4.718344416806268e-06,
      "loss": 0.0359,
      "step": 45155
    },
    {
      "epoch": 2.2925021574699223,
      "grad_norm": 0.40257173776626587,
      "learning_rate": 4.716652283533851e-06,
      "loss": 0.0369,
      "step": 45160
    },
    {
      "epoch": 2.292755977460785,
      "grad_norm": 0.29946520924568176,
      "learning_rate": 4.714960150261435e-06,
      "loss": 0.0307,
      "step": 45165
    },
    {
      "epoch": 2.2930097974516475,
      "grad_norm": 0.4067082107067108,
      "learning_rate": 4.713268016989018e-06,
      "loss": 0.0354,
      "step": 45170
    },
    {
      "epoch": 2.2932636174425096,
      "grad_norm": 1.0462335348129272,
      "learning_rate": 4.711575883716602e-06,
      "loss": 0.0276,
      "step": 45175
    },
    {
      "epoch": 2.293517437433372,
      "grad_norm": 0.5343073606491089,
      "learning_rate": 4.709883750444185e-06,
      "loss": 0.0288,
      "step": 45180
    },
    {
      "epoch": 2.293771257424235,
      "grad_norm": 0.601951539516449,
      "learning_rate": 4.708191617171769e-06,
      "loss": 0.0341,
      "step": 45185
    },
    {
      "epoch": 2.294025077415097,
      "grad_norm": 0.3987095355987549,
      "learning_rate": 4.706499483899352e-06,
      "loss": 0.0273,
      "step": 45190
    },
    {
      "epoch": 2.2942788974059596,
      "grad_norm": 0.45923149585723877,
      "learning_rate": 4.704807350626936e-06,
      "loss": 0.0414,
      "step": 45195
    },
    {
      "epoch": 2.294532717396822,
      "grad_norm": 0.3649718165397644,
      "learning_rate": 4.703115217354519e-06,
      "loss": 0.0213,
      "step": 45200
    },
    {
      "epoch": 2.294786537387685,
      "grad_norm": 0.24480339884757996,
      "learning_rate": 4.701423084082103e-06,
      "loss": 0.0281,
      "step": 45205
    },
    {
      "epoch": 2.295040357378547,
      "grad_norm": 0.36629045009613037,
      "learning_rate": 4.699730950809686e-06,
      "loss": 0.0317,
      "step": 45210
    },
    {
      "epoch": 2.2952941773694095,
      "grad_norm": 0.3395744264125824,
      "learning_rate": 4.69803881753727e-06,
      "loss": 0.0334,
      "step": 45215
    },
    {
      "epoch": 2.295547997360272,
      "grad_norm": 0.7202277779579163,
      "learning_rate": 4.696346684264853e-06,
      "loss": 0.0329,
      "step": 45220
    },
    {
      "epoch": 2.2958018173511348,
      "grad_norm": 0.3674011528491974,
      "learning_rate": 4.694654550992436e-06,
      "loss": 0.0273,
      "step": 45225
    },
    {
      "epoch": 2.296055637341997,
      "grad_norm": 0.3968465030193329,
      "learning_rate": 4.69296241772002e-06,
      "loss": 0.0245,
      "step": 45230
    },
    {
      "epoch": 2.2963094573328595,
      "grad_norm": 0.371276319026947,
      "learning_rate": 4.691270284447603e-06,
      "loss": 0.0281,
      "step": 45235
    },
    {
      "epoch": 2.296563277323722,
      "grad_norm": 0.5648274421691895,
      "learning_rate": 4.689578151175187e-06,
      "loss": 0.0409,
      "step": 45240
    },
    {
      "epoch": 2.2968170973145847,
      "grad_norm": 0.7816662788391113,
      "learning_rate": 4.68788601790277e-06,
      "loss": 0.0299,
      "step": 45245
    },
    {
      "epoch": 2.297070917305447,
      "grad_norm": 0.35681354999542236,
      "learning_rate": 4.686193884630354e-06,
      "loss": 0.0302,
      "step": 45250
    },
    {
      "epoch": 2.2973247372963095,
      "grad_norm": 0.3457604646682739,
      "learning_rate": 4.684501751357937e-06,
      "loss": 0.0344,
      "step": 45255
    },
    {
      "epoch": 2.297578557287172,
      "grad_norm": 0.7468198537826538,
      "learning_rate": 4.682809618085521e-06,
      "loss": 0.0352,
      "step": 45260
    },
    {
      "epoch": 2.2978323772780342,
      "grad_norm": 0.2634367048740387,
      "learning_rate": 4.6811174848131045e-06,
      "loss": 0.0321,
      "step": 45265
    },
    {
      "epoch": 2.298086197268897,
      "grad_norm": 0.22003506124019623,
      "learning_rate": 4.679425351540688e-06,
      "loss": 0.0338,
      "step": 45270
    },
    {
      "epoch": 2.2983400172597594,
      "grad_norm": 0.17040963470935822,
      "learning_rate": 4.677733218268271e-06,
      "loss": 0.0281,
      "step": 45275
    },
    {
      "epoch": 2.298593837250622,
      "grad_norm": 0.27573078870773315,
      "learning_rate": 4.676041084995855e-06,
      "loss": 0.0304,
      "step": 45280
    },
    {
      "epoch": 2.298847657241484,
      "grad_norm": 0.9302784204483032,
      "learning_rate": 4.674348951723438e-06,
      "loss": 0.0329,
      "step": 45285
    },
    {
      "epoch": 2.299101477232347,
      "grad_norm": 0.3319515585899353,
      "learning_rate": 4.672656818451022e-06,
      "loss": 0.0423,
      "step": 45290
    },
    {
      "epoch": 2.2993552972232094,
      "grad_norm": 0.7237670421600342,
      "learning_rate": 4.670964685178605e-06,
      "loss": 0.033,
      "step": 45295
    },
    {
      "epoch": 2.2996091172140716,
      "grad_norm": 0.3120134174823761,
      "learning_rate": 4.6692725519061884e-06,
      "loss": 0.0396,
      "step": 45300
    },
    {
      "epoch": 2.299862937204934,
      "grad_norm": 0.47648027539253235,
      "learning_rate": 4.667580418633772e-06,
      "loss": 0.03,
      "step": 45305
    },
    {
      "epoch": 2.300116757195797,
      "grad_norm": 0.24086786806583405,
      "learning_rate": 4.6658882853613555e-06,
      "loss": 0.0379,
      "step": 45310
    },
    {
      "epoch": 2.3003705771866594,
      "grad_norm": 0.2652145326137543,
      "learning_rate": 4.664196152088939e-06,
      "loss": 0.0355,
      "step": 45315
    },
    {
      "epoch": 2.3006243971775215,
      "grad_norm": 0.2888307273387909,
      "learning_rate": 4.662504018816522e-06,
      "loss": 0.0265,
      "step": 45320
    },
    {
      "epoch": 2.300878217168384,
      "grad_norm": 0.3059805929660797,
      "learning_rate": 4.660811885544106e-06,
      "loss": 0.0322,
      "step": 45325
    },
    {
      "epoch": 2.3011320371592467,
      "grad_norm": 0.555613100528717,
      "learning_rate": 4.659119752271689e-06,
      "loss": 0.031,
      "step": 45330
    },
    {
      "epoch": 2.301385857150109,
      "grad_norm": 0.31799572706222534,
      "learning_rate": 4.657427618999273e-06,
      "loss": 0.0326,
      "step": 45335
    },
    {
      "epoch": 2.3016396771409715,
      "grad_norm": 0.2731345295906067,
      "learning_rate": 4.655735485726856e-06,
      "loss": 0.0367,
      "step": 45340
    },
    {
      "epoch": 2.301893497131834,
      "grad_norm": 0.23575860261917114,
      "learning_rate": 4.6540433524544395e-06,
      "loss": 0.024,
      "step": 45345
    },
    {
      "epoch": 2.3021473171226967,
      "grad_norm": 0.43875861167907715,
      "learning_rate": 4.652351219182023e-06,
      "loss": 0.0288,
      "step": 45350
    },
    {
      "epoch": 2.302401137113559,
      "grad_norm": 0.3125607669353485,
      "learning_rate": 4.650659085909607e-06,
      "loss": 0.032,
      "step": 45355
    },
    {
      "epoch": 2.3026549571044215,
      "grad_norm": 0.2842237949371338,
      "learning_rate": 4.64896695263719e-06,
      "loss": 0.0329,
      "step": 45360
    },
    {
      "epoch": 2.302908777095284,
      "grad_norm": 0.3345993459224701,
      "learning_rate": 4.647274819364774e-06,
      "loss": 0.0352,
      "step": 45365
    },
    {
      "epoch": 2.3031625970861467,
      "grad_norm": 0.3049146831035614,
      "learning_rate": 4.645582686092357e-06,
      "loss": 0.0305,
      "step": 45370
    },
    {
      "epoch": 2.303416417077009,
      "grad_norm": 0.3029148280620575,
      "learning_rate": 4.643890552819941e-06,
      "loss": 0.0304,
      "step": 45375
    },
    {
      "epoch": 2.3036702370678714,
      "grad_norm": 0.2953786253929138,
      "learning_rate": 4.642198419547524e-06,
      "loss": 0.027,
      "step": 45380
    },
    {
      "epoch": 2.303924057058734,
      "grad_norm": 0.5618043541908264,
      "learning_rate": 4.640506286275107e-06,
      "loss": 0.036,
      "step": 45385
    },
    {
      "epoch": 2.3041778770495966,
      "grad_norm": 0.36698117852211,
      "learning_rate": 4.6388141530026905e-06,
      "loss": 0.0356,
      "step": 45390
    },
    {
      "epoch": 2.304431697040459,
      "grad_norm": 0.30508795380592346,
      "learning_rate": 4.637122019730274e-06,
      "loss": 0.0278,
      "step": 45395
    },
    {
      "epoch": 2.3046855170313214,
      "grad_norm": 0.39213359355926514,
      "learning_rate": 4.635429886457858e-06,
      "loss": 0.0371,
      "step": 45400
    },
    {
      "epoch": 2.304939337022184,
      "grad_norm": 0.32886233925819397,
      "learning_rate": 4.633737753185441e-06,
      "loss": 0.0336,
      "step": 45405
    },
    {
      "epoch": 2.305193157013046,
      "grad_norm": 0.34936273097991943,
      "learning_rate": 4.632045619913025e-06,
      "loss": 0.0377,
      "step": 45410
    },
    {
      "epoch": 2.3054469770039088,
      "grad_norm": 0.3372868299484253,
      "learning_rate": 4.630353486640608e-06,
      "loss": 0.0334,
      "step": 45415
    },
    {
      "epoch": 2.3057007969947714,
      "grad_norm": 0.29870179295539856,
      "learning_rate": 4.628661353368192e-06,
      "loss": 0.0376,
      "step": 45420
    },
    {
      "epoch": 2.305954616985634,
      "grad_norm": 0.2501830756664276,
      "learning_rate": 4.626969220095775e-06,
      "loss": 0.0287,
      "step": 45425
    },
    {
      "epoch": 2.306208436976496,
      "grad_norm": 0.33491966128349304,
      "learning_rate": 4.625277086823359e-06,
      "loss": 0.0283,
      "step": 45430
    },
    {
      "epoch": 2.3064622569673587,
      "grad_norm": 0.30442988872528076,
      "learning_rate": 4.623584953550942e-06,
      "loss": 0.0347,
      "step": 45435
    },
    {
      "epoch": 2.3067160769582213,
      "grad_norm": 0.44757723808288574,
      "learning_rate": 4.621892820278526e-06,
      "loss": 0.0331,
      "step": 45440
    },
    {
      "epoch": 2.3069698969490835,
      "grad_norm": 0.27261823415756226,
      "learning_rate": 4.620200687006109e-06,
      "loss": 0.0305,
      "step": 45445
    },
    {
      "epoch": 2.307223716939946,
      "grad_norm": 0.27286872267723083,
      "learning_rate": 4.618508553733693e-06,
      "loss": 0.0272,
      "step": 45450
    },
    {
      "epoch": 2.3074775369308087,
      "grad_norm": 0.23978255689144135,
      "learning_rate": 4.616816420461276e-06,
      "loss": 0.034,
      "step": 45455
    },
    {
      "epoch": 2.3077313569216713,
      "grad_norm": 0.406010240316391,
      "learning_rate": 4.615124287188859e-06,
      "loss": 0.0358,
      "step": 45460
    },
    {
      "epoch": 2.3079851769125335,
      "grad_norm": 0.34851107001304626,
      "learning_rate": 4.613432153916443e-06,
      "loss": 0.0336,
      "step": 45465
    },
    {
      "epoch": 2.308238996903396,
      "grad_norm": 0.38284942507743835,
      "learning_rate": 4.611740020644026e-06,
      "loss": 0.0297,
      "step": 45470
    },
    {
      "epoch": 2.3084928168942587,
      "grad_norm": 0.34113410115242004,
      "learning_rate": 4.61004788737161e-06,
      "loss": 0.0345,
      "step": 45475
    },
    {
      "epoch": 2.308746636885121,
      "grad_norm": 0.3386651575565338,
      "learning_rate": 4.608355754099193e-06,
      "loss": 0.0318,
      "step": 45480
    },
    {
      "epoch": 2.3090004568759834,
      "grad_norm": 0.4475891590118408,
      "learning_rate": 4.606663620826777e-06,
      "loss": 0.0384,
      "step": 45485
    },
    {
      "epoch": 2.309254276866846,
      "grad_norm": 0.39710333943367004,
      "learning_rate": 4.60497148755436e-06,
      "loss": 0.0338,
      "step": 45490
    },
    {
      "epoch": 2.3095080968577086,
      "grad_norm": 0.5366309881210327,
      "learning_rate": 4.603279354281944e-06,
      "loss": 0.0312,
      "step": 45495
    },
    {
      "epoch": 2.3097619168485712,
      "grad_norm": 0.5256260633468628,
      "learning_rate": 4.601587221009527e-06,
      "loss": 0.0373,
      "step": 45500
    },
    {
      "epoch": 2.3100157368394334,
      "grad_norm": 0.23877793550491333,
      "learning_rate": 4.59989508773711e-06,
      "loss": 0.0307,
      "step": 45505
    },
    {
      "epoch": 2.310269556830296,
      "grad_norm": 0.24250926077365875,
      "learning_rate": 4.598202954464694e-06,
      "loss": 0.0384,
      "step": 45510
    },
    {
      "epoch": 2.3105233768211586,
      "grad_norm": 0.3884424567222595,
      "learning_rate": 4.5965108211922775e-06,
      "loss": 0.0455,
      "step": 45515
    },
    {
      "epoch": 2.3107771968120208,
      "grad_norm": 0.3509715795516968,
      "learning_rate": 4.594818687919861e-06,
      "loss": 0.0321,
      "step": 45520
    },
    {
      "epoch": 2.3110310168028834,
      "grad_norm": 0.3542064428329468,
      "learning_rate": 4.5931265546474446e-06,
      "loss": 0.0336,
      "step": 45525
    },
    {
      "epoch": 2.311284836793746,
      "grad_norm": 0.43271404504776,
      "learning_rate": 4.591434421375028e-06,
      "loss": 0.037,
      "step": 45530
    },
    {
      "epoch": 2.3115386567846086,
      "grad_norm": 0.5255846381187439,
      "learning_rate": 4.589742288102612e-06,
      "loss": 0.0345,
      "step": 45535
    },
    {
      "epoch": 2.3117924767754707,
      "grad_norm": 0.36676549911499023,
      "learning_rate": 4.588050154830195e-06,
      "loss": 0.029,
      "step": 45540
    },
    {
      "epoch": 2.3120462967663333,
      "grad_norm": 0.3285497725009918,
      "learning_rate": 4.586358021557779e-06,
      "loss": 0.0299,
      "step": 45545
    },
    {
      "epoch": 2.312300116757196,
      "grad_norm": 0.4070643186569214,
      "learning_rate": 4.5846658882853614e-06,
      "loss": 0.0324,
      "step": 45550
    },
    {
      "epoch": 2.312553936748058,
      "grad_norm": 0.22612564265727997,
      "learning_rate": 4.582973755012945e-06,
      "loss": 0.0295,
      "step": 45555
    },
    {
      "epoch": 2.3128077567389207,
      "grad_norm": 0.527091383934021,
      "learning_rate": 4.5812816217405285e-06,
      "loss": 0.0292,
      "step": 45560
    },
    {
      "epoch": 2.3130615767297833,
      "grad_norm": 0.36508095264434814,
      "learning_rate": 4.579589488468112e-06,
      "loss": 0.0351,
      "step": 45565
    },
    {
      "epoch": 2.313315396720646,
      "grad_norm": 0.2954444885253906,
      "learning_rate": 4.577897355195696e-06,
      "loss": 0.0373,
      "step": 45570
    },
    {
      "epoch": 2.313569216711508,
      "grad_norm": 0.48034361004829407,
      "learning_rate": 4.576205221923279e-06,
      "loss": 0.039,
      "step": 45575
    },
    {
      "epoch": 2.3138230367023707,
      "grad_norm": 0.35288605093955994,
      "learning_rate": 4.574513088650863e-06,
      "loss": 0.0271,
      "step": 45580
    },
    {
      "epoch": 2.3140768566932333,
      "grad_norm": 0.24266214668750763,
      "learning_rate": 4.572820955378446e-06,
      "loss": 0.0267,
      "step": 45585
    },
    {
      "epoch": 2.3143306766840954,
      "grad_norm": 0.2815217077732086,
      "learning_rate": 4.57112882210603e-06,
      "loss": 0.0294,
      "step": 45590
    },
    {
      "epoch": 2.314584496674958,
      "grad_norm": 0.27903756499290466,
      "learning_rate": 4.5694366888336125e-06,
      "loss": 0.0332,
      "step": 45595
    },
    {
      "epoch": 2.3148383166658206,
      "grad_norm": 0.43626129627227783,
      "learning_rate": 4.567744555561197e-06,
      "loss": 0.0358,
      "step": 45600
    },
    {
      "epoch": 2.3150921366566832,
      "grad_norm": 0.3123166561126709,
      "learning_rate": 4.5660524222887796e-06,
      "loss": 0.0352,
      "step": 45605
    },
    {
      "epoch": 2.3153459566475454,
      "grad_norm": 0.3405829966068268,
      "learning_rate": 4.564360289016363e-06,
      "loss": 0.0346,
      "step": 45610
    },
    {
      "epoch": 2.315599776638408,
      "grad_norm": 0.5032159090042114,
      "learning_rate": 4.562668155743947e-06,
      "loss": 0.0311,
      "step": 45615
    },
    {
      "epoch": 2.3158535966292706,
      "grad_norm": 0.36772871017456055,
      "learning_rate": 4.56097602247153e-06,
      "loss": 0.0308,
      "step": 45620
    },
    {
      "epoch": 2.3161074166201328,
      "grad_norm": 0.49046018719673157,
      "learning_rate": 4.559283889199114e-06,
      "loss": 0.0309,
      "step": 45625
    },
    {
      "epoch": 2.3163612366109954,
      "grad_norm": 0.379315584897995,
      "learning_rate": 4.557591755926697e-06,
      "loss": 0.0326,
      "step": 45630
    },
    {
      "epoch": 2.316615056601858,
      "grad_norm": 0.38047313690185547,
      "learning_rate": 4.555899622654281e-06,
      "loss": 0.0437,
      "step": 45635
    },
    {
      "epoch": 2.3168688765927206,
      "grad_norm": 1.0208139419555664,
      "learning_rate": 4.5542074893818635e-06,
      "loss": 0.028,
      "step": 45640
    },
    {
      "epoch": 2.317122696583583,
      "grad_norm": 0.3524598479270935,
      "learning_rate": 4.552515356109448e-06,
      "loss": 0.0403,
      "step": 45645
    },
    {
      "epoch": 2.3173765165744453,
      "grad_norm": 0.3423077166080475,
      "learning_rate": 4.550823222837031e-06,
      "loss": 0.0431,
      "step": 45650
    },
    {
      "epoch": 2.317630336565308,
      "grad_norm": 0.72367924451828,
      "learning_rate": 4.549131089564614e-06,
      "loss": 0.0309,
      "step": 45655
    },
    {
      "epoch": 2.3178841565561705,
      "grad_norm": 0.5329000949859619,
      "learning_rate": 4.547438956292198e-06,
      "loss": 0.0427,
      "step": 45660
    },
    {
      "epoch": 2.3181379765470327,
      "grad_norm": 0.2928794026374817,
      "learning_rate": 4.545746823019781e-06,
      "loss": 0.0353,
      "step": 45665
    },
    {
      "epoch": 2.3183917965378953,
      "grad_norm": 0.3894771933555603,
      "learning_rate": 4.544054689747365e-06,
      "loss": 0.031,
      "step": 45670
    },
    {
      "epoch": 2.318645616528758,
      "grad_norm": 0.40756314992904663,
      "learning_rate": 4.542362556474948e-06,
      "loss": 0.0284,
      "step": 45675
    },
    {
      "epoch": 2.3188994365196205,
      "grad_norm": 0.2886298894882202,
      "learning_rate": 4.540670423202532e-06,
      "loss": 0.0348,
      "step": 45680
    },
    {
      "epoch": 2.3191532565104827,
      "grad_norm": 0.4084828495979309,
      "learning_rate": 4.5389782899301154e-06,
      "loss": 0.0368,
      "step": 45685
    },
    {
      "epoch": 2.3194070765013453,
      "grad_norm": 0.6314843893051147,
      "learning_rate": 4.537286156657699e-06,
      "loss": 0.0375,
      "step": 45690
    },
    {
      "epoch": 2.319660896492208,
      "grad_norm": 0.21973839402198792,
      "learning_rate": 4.5355940233852825e-06,
      "loss": 0.0317,
      "step": 45695
    },
    {
      "epoch": 2.31991471648307,
      "grad_norm": 0.2902248799800873,
      "learning_rate": 4.533901890112865e-06,
      "loss": 0.0299,
      "step": 45700
    },
    {
      "epoch": 2.3201685364739326,
      "grad_norm": 0.3448190987110138,
      "learning_rate": 4.53220975684045e-06,
      "loss": 0.0449,
      "step": 45705
    },
    {
      "epoch": 2.320422356464795,
      "grad_norm": 0.2905779480934143,
      "learning_rate": 4.530517623568032e-06,
      "loss": 0.0309,
      "step": 45710
    },
    {
      "epoch": 2.320676176455658,
      "grad_norm": 0.3386252820491791,
      "learning_rate": 4.528825490295616e-06,
      "loss": 0.0302,
      "step": 45715
    },
    {
      "epoch": 2.32092999644652,
      "grad_norm": 0.31150320172309875,
      "learning_rate": 4.527133357023199e-06,
      "loss": 0.0295,
      "step": 45720
    },
    {
      "epoch": 2.3211838164373826,
      "grad_norm": 0.20965662598609924,
      "learning_rate": 4.525441223750783e-06,
      "loss": 0.0283,
      "step": 45725
    },
    {
      "epoch": 2.321437636428245,
      "grad_norm": 0.31546589732170105,
      "learning_rate": 4.5237490904783665e-06,
      "loss": 0.0312,
      "step": 45730
    },
    {
      "epoch": 2.3216914564191073,
      "grad_norm": 0.4126461148262024,
      "learning_rate": 4.52205695720595e-06,
      "loss": 0.0347,
      "step": 45735
    },
    {
      "epoch": 2.32194527640997,
      "grad_norm": 0.3679712116718292,
      "learning_rate": 4.520364823933534e-06,
      "loss": 0.029,
      "step": 45740
    },
    {
      "epoch": 2.3221990964008326,
      "grad_norm": 0.7221439480781555,
      "learning_rate": 4.518672690661116e-06,
      "loss": 0.0276,
      "step": 45745
    },
    {
      "epoch": 2.322452916391695,
      "grad_norm": 0.4402199387550354,
      "learning_rate": 4.516980557388701e-06,
      "loss": 0.0336,
      "step": 45750
    },
    {
      "epoch": 2.3227067363825573,
      "grad_norm": 0.32814040780067444,
      "learning_rate": 4.515288424116283e-06,
      "loss": 0.0336,
      "step": 45755
    },
    {
      "epoch": 2.32296055637342,
      "grad_norm": 0.4378836154937744,
      "learning_rate": 4.513596290843868e-06,
      "loss": 0.0465,
      "step": 45760
    },
    {
      "epoch": 2.3232143763642825,
      "grad_norm": 0.38695916533470154,
      "learning_rate": 4.5119041575714505e-06,
      "loss": 0.0375,
      "step": 45765
    },
    {
      "epoch": 2.323468196355145,
      "grad_norm": 0.7761498689651489,
      "learning_rate": 4.510212024299034e-06,
      "loss": 0.0357,
      "step": 45770
    },
    {
      "epoch": 2.3237220163460073,
      "grad_norm": 0.4433533549308777,
      "learning_rate": 4.5085198910266175e-06,
      "loss": 0.0413,
      "step": 45775
    },
    {
      "epoch": 2.32397583633687,
      "grad_norm": 0.2922021746635437,
      "learning_rate": 4.506827757754201e-06,
      "loss": 0.0351,
      "step": 45780
    },
    {
      "epoch": 2.3242296563277325,
      "grad_norm": 0.32604724168777466,
      "learning_rate": 4.505135624481785e-06,
      "loss": 0.0337,
      "step": 45785
    },
    {
      "epoch": 2.324483476318595,
      "grad_norm": 0.3584740459918976,
      "learning_rate": 4.503443491209367e-06,
      "loss": 0.0372,
      "step": 45790
    },
    {
      "epoch": 2.3247372963094572,
      "grad_norm": 0.5668825507164001,
      "learning_rate": 4.501751357936952e-06,
      "loss": 0.0413,
      "step": 45795
    },
    {
      "epoch": 2.32499111630032,
      "grad_norm": 0.3352181613445282,
      "learning_rate": 4.500059224664534e-06,
      "loss": 0.0373,
      "step": 45800
    },
    {
      "epoch": 2.3252449362911825,
      "grad_norm": 0.32401424646377563,
      "learning_rate": 4.498367091392119e-06,
      "loss": 0.0341,
      "step": 45805
    },
    {
      "epoch": 2.3254987562820446,
      "grad_norm": 0.3033709228038788,
      "learning_rate": 4.4966749581197015e-06,
      "loss": 0.0288,
      "step": 45810
    },
    {
      "epoch": 2.325752576272907,
      "grad_norm": 0.30588579177856445,
      "learning_rate": 4.494982824847285e-06,
      "loss": 0.0268,
      "step": 45815
    },
    {
      "epoch": 2.32600639626377,
      "grad_norm": 0.40908050537109375,
      "learning_rate": 4.493290691574869e-06,
      "loss": 0.0363,
      "step": 45820
    },
    {
      "epoch": 2.3262602162546324,
      "grad_norm": 0.5321323871612549,
      "learning_rate": 4.491598558302452e-06,
      "loss": 0.033,
      "step": 45825
    },
    {
      "epoch": 2.3265140362454946,
      "grad_norm": 0.20381450653076172,
      "learning_rate": 4.489906425030036e-06,
      "loss": 0.0328,
      "step": 45830
    },
    {
      "epoch": 2.326767856236357,
      "grad_norm": 0.9036016464233398,
      "learning_rate": 4.488214291757619e-06,
      "loss": 0.0495,
      "step": 45835
    },
    {
      "epoch": 2.32702167622722,
      "grad_norm": 0.2598854601383209,
      "learning_rate": 4.486522158485203e-06,
      "loss": 0.0328,
      "step": 45840
    },
    {
      "epoch": 2.327275496218082,
      "grad_norm": 0.2837057411670685,
      "learning_rate": 4.484830025212786e-06,
      "loss": 0.0267,
      "step": 45845
    },
    {
      "epoch": 2.3275293162089445,
      "grad_norm": 0.33842211961746216,
      "learning_rate": 4.48313789194037e-06,
      "loss": 0.0313,
      "step": 45850
    },
    {
      "epoch": 2.327783136199807,
      "grad_norm": 0.3597102761268616,
      "learning_rate": 4.481445758667953e-06,
      "loss": 0.034,
      "step": 45855
    },
    {
      "epoch": 2.3280369561906697,
      "grad_norm": 0.33887141942977905,
      "learning_rate": 4.479753625395536e-06,
      "loss": 0.0269,
      "step": 45860
    },
    {
      "epoch": 2.328290776181532,
      "grad_norm": 0.24716730415821075,
      "learning_rate": 4.4780614921231205e-06,
      "loss": 0.03,
      "step": 45865
    },
    {
      "epoch": 2.3285445961723945,
      "grad_norm": 0.4096173048019409,
      "learning_rate": 4.476369358850703e-06,
      "loss": 0.0298,
      "step": 45870
    },
    {
      "epoch": 2.328798416163257,
      "grad_norm": 0.34281015396118164,
      "learning_rate": 4.474677225578287e-06,
      "loss": 0.0224,
      "step": 45875
    },
    {
      "epoch": 2.3290522361541193,
      "grad_norm": 0.2912023067474365,
      "learning_rate": 4.47298509230587e-06,
      "loss": 0.0281,
      "step": 45880
    },
    {
      "epoch": 2.329306056144982,
      "grad_norm": 0.3161762058734894,
      "learning_rate": 4.471292959033454e-06,
      "loss": 0.0327,
      "step": 45885
    },
    {
      "epoch": 2.3295598761358445,
      "grad_norm": 0.3971143364906311,
      "learning_rate": 4.469600825761037e-06,
      "loss": 0.0245,
      "step": 45890
    },
    {
      "epoch": 2.329813696126707,
      "grad_norm": 0.24481990933418274,
      "learning_rate": 4.467908692488621e-06,
      "loss": 0.0262,
      "step": 45895
    },
    {
      "epoch": 2.3300675161175692,
      "grad_norm": 0.4387982189655304,
      "learning_rate": 4.4662165592162045e-06,
      "loss": 0.028,
      "step": 45900
    },
    {
      "epoch": 2.330321336108432,
      "grad_norm": 0.3262830674648285,
      "learning_rate": 4.464524425943787e-06,
      "loss": 0.0432,
      "step": 45905
    },
    {
      "epoch": 2.3305751560992944,
      "grad_norm": 0.5117439031600952,
      "learning_rate": 4.4628322926713716e-06,
      "loss": 0.0405,
      "step": 45910
    },
    {
      "epoch": 2.330828976090157,
      "grad_norm": 0.4031218886375427,
      "learning_rate": 4.461140159398954e-06,
      "loss": 0.0318,
      "step": 45915
    },
    {
      "epoch": 2.331082796081019,
      "grad_norm": 0.31962868571281433,
      "learning_rate": 4.459448026126539e-06,
      "loss": 0.0303,
      "step": 45920
    },
    {
      "epoch": 2.331336616071882,
      "grad_norm": 0.3103834390640259,
      "learning_rate": 4.457755892854121e-06,
      "loss": 0.0375,
      "step": 45925
    },
    {
      "epoch": 2.3315904360627444,
      "grad_norm": 0.41599103808403015,
      "learning_rate": 4.456063759581705e-06,
      "loss": 0.0323,
      "step": 45930
    },
    {
      "epoch": 2.331844256053607,
      "grad_norm": 0.38975197076797485,
      "learning_rate": 4.4543716263092884e-06,
      "loss": 0.0345,
      "step": 45935
    },
    {
      "epoch": 2.332098076044469,
      "grad_norm": 0.22098346054553986,
      "learning_rate": 4.452679493036872e-06,
      "loss": 0.0255,
      "step": 45940
    },
    {
      "epoch": 2.3323518960353318,
      "grad_norm": 0.26242735981941223,
      "learning_rate": 4.4509873597644555e-06,
      "loss": 0.0331,
      "step": 45945
    },
    {
      "epoch": 2.3326057160261944,
      "grad_norm": 0.2703067660331726,
      "learning_rate": 4.449295226492038e-06,
      "loss": 0.0303,
      "step": 45950
    },
    {
      "epoch": 2.3328595360170565,
      "grad_norm": 0.42793992161750793,
      "learning_rate": 4.447603093219623e-06,
      "loss": 0.042,
      "step": 45955
    },
    {
      "epoch": 2.333113356007919,
      "grad_norm": 0.28537246584892273,
      "learning_rate": 4.445910959947205e-06,
      "loss": 0.0342,
      "step": 45960
    },
    {
      "epoch": 2.3333671759987817,
      "grad_norm": 0.338586688041687,
      "learning_rate": 4.44421882667479e-06,
      "loss": 0.0304,
      "step": 45965
    },
    {
      "epoch": 2.3336209959896443,
      "grad_norm": 0.2869323790073395,
      "learning_rate": 4.442526693402372e-06,
      "loss": 0.0335,
      "step": 45970
    },
    {
      "epoch": 2.3338748159805065,
      "grad_norm": 0.21080097556114197,
      "learning_rate": 4.440834560129956e-06,
      "loss": 0.0275,
      "step": 45975
    },
    {
      "epoch": 2.334128635971369,
      "grad_norm": 0.8009614944458008,
      "learning_rate": 4.4391424268575395e-06,
      "loss": 0.0291,
      "step": 45980
    },
    {
      "epoch": 2.3343824559622317,
      "grad_norm": 0.4224371910095215,
      "learning_rate": 4.437450293585123e-06,
      "loss": 0.0277,
      "step": 45985
    },
    {
      "epoch": 2.334636275953094,
      "grad_norm": 0.3391774296760559,
      "learning_rate": 4.4357581603127066e-06,
      "loss": 0.0242,
      "step": 45990
    },
    {
      "epoch": 2.3348900959439565,
      "grad_norm": 0.27695131301879883,
      "learning_rate": 4.43406602704029e-06,
      "loss": 0.0334,
      "step": 45995
    },
    {
      "epoch": 2.335143915934819,
      "grad_norm": 0.2666962742805481,
      "learning_rate": 4.432373893767874e-06,
      "loss": 0.031,
      "step": 46000
    },
    {
      "epoch": 2.3353977359256817,
      "grad_norm": 0.3789578378200531,
      "learning_rate": 4.430681760495457e-06,
      "loss": 0.0369,
      "step": 46005
    },
    {
      "epoch": 2.335651555916544,
      "grad_norm": 0.37819820642471313,
      "learning_rate": 4.428989627223041e-06,
      "loss": 0.0306,
      "step": 46010
    },
    {
      "epoch": 2.3359053759074064,
      "grad_norm": 0.4716276526451111,
      "learning_rate": 4.427297493950624e-06,
      "loss": 0.0484,
      "step": 46015
    },
    {
      "epoch": 2.336159195898269,
      "grad_norm": 0.3425740897655487,
      "learning_rate": 4.425605360678207e-06,
      "loss": 0.0333,
      "step": 46020
    },
    {
      "epoch": 2.336413015889131,
      "grad_norm": 0.41194647550582886,
      "learning_rate": 4.423913227405791e-06,
      "loss": 0.0341,
      "step": 46025
    },
    {
      "epoch": 2.336666835879994,
      "grad_norm": 0.3530094027519226,
      "learning_rate": 4.422221094133374e-06,
      "loss": 0.0293,
      "step": 46030
    },
    {
      "epoch": 2.3369206558708564,
      "grad_norm": 0.31580179929733276,
      "learning_rate": 4.420528960860958e-06,
      "loss": 0.0295,
      "step": 46035
    },
    {
      "epoch": 2.337174475861719,
      "grad_norm": 0.35331493616104126,
      "learning_rate": 4.418836827588541e-06,
      "loss": 0.0366,
      "step": 46040
    },
    {
      "epoch": 2.337428295852581,
      "grad_norm": 0.47498664259910583,
      "learning_rate": 4.417144694316125e-06,
      "loss": 0.0372,
      "step": 46045
    },
    {
      "epoch": 2.3376821158434438,
      "grad_norm": 0.31873664259910583,
      "learning_rate": 4.415452561043708e-06,
      "loss": 0.0284,
      "step": 46050
    },
    {
      "epoch": 2.3379359358343064,
      "grad_norm": 0.41014498472213745,
      "learning_rate": 4.413760427771292e-06,
      "loss": 0.0403,
      "step": 46055
    },
    {
      "epoch": 2.338189755825169,
      "grad_norm": 0.43682634830474854,
      "learning_rate": 4.412068294498875e-06,
      "loss": 0.0445,
      "step": 46060
    },
    {
      "epoch": 2.338443575816031,
      "grad_norm": 0.5572406649589539,
      "learning_rate": 4.410376161226458e-06,
      "loss": 0.0274,
      "step": 46065
    },
    {
      "epoch": 2.3386973958068937,
      "grad_norm": 0.4906730353832245,
      "learning_rate": 4.4086840279540424e-06,
      "loss": 0.0288,
      "step": 46070
    },
    {
      "epoch": 2.3389512157977563,
      "grad_norm": 0.26693737506866455,
      "learning_rate": 4.406991894681625e-06,
      "loss": 0.0349,
      "step": 46075
    },
    {
      "epoch": 2.339205035788619,
      "grad_norm": 0.2562685012817383,
      "learning_rate": 4.4052997614092095e-06,
      "loss": 0.0286,
      "step": 46080
    },
    {
      "epoch": 2.339458855779481,
      "grad_norm": 0.3544439375400543,
      "learning_rate": 4.403607628136792e-06,
      "loss": 0.0291,
      "step": 46085
    },
    {
      "epoch": 2.3397126757703437,
      "grad_norm": 0.5881972908973694,
      "learning_rate": 4.401915494864376e-06,
      "loss": 0.0264,
      "step": 46090
    },
    {
      "epoch": 2.3399664957612063,
      "grad_norm": 0.304453045129776,
      "learning_rate": 4.400223361591959e-06,
      "loss": 0.0334,
      "step": 46095
    },
    {
      "epoch": 2.3402203157520685,
      "grad_norm": 0.26709556579589844,
      "learning_rate": 4.398531228319543e-06,
      "loss": 0.0366,
      "step": 46100
    },
    {
      "epoch": 2.340474135742931,
      "grad_norm": 0.4727105498313904,
      "learning_rate": 4.396839095047126e-06,
      "loss": 0.0335,
      "step": 46105
    },
    {
      "epoch": 2.3407279557337937,
      "grad_norm": 0.27335217595100403,
      "learning_rate": 4.39514696177471e-06,
      "loss": 0.0283,
      "step": 46110
    },
    {
      "epoch": 2.3409817757246563,
      "grad_norm": 0.544344961643219,
      "learning_rate": 4.3934548285022935e-06,
      "loss": 0.0305,
      "step": 46115
    },
    {
      "epoch": 2.3412355957155184,
      "grad_norm": 0.2982689440250397,
      "learning_rate": 4.391762695229876e-06,
      "loss": 0.0301,
      "step": 46120
    },
    {
      "epoch": 2.341489415706381,
      "grad_norm": 0.4149956703186035,
      "learning_rate": 4.390070561957461e-06,
      "loss": 0.0338,
      "step": 46125
    },
    {
      "epoch": 2.3417432356972436,
      "grad_norm": 0.28789716958999634,
      "learning_rate": 4.388378428685043e-06,
      "loss": 0.0364,
      "step": 46130
    },
    {
      "epoch": 2.341997055688106,
      "grad_norm": 0.32526326179504395,
      "learning_rate": 4.386686295412627e-06,
      "loss": 0.0333,
      "step": 46135
    },
    {
      "epoch": 2.3422508756789684,
      "grad_norm": 0.3240688145160675,
      "learning_rate": 4.38499416214021e-06,
      "loss": 0.0303,
      "step": 46140
    },
    {
      "epoch": 2.342504695669831,
      "grad_norm": 0.3309798836708069,
      "learning_rate": 4.383302028867794e-06,
      "loss": 0.0346,
      "step": 46145
    },
    {
      "epoch": 2.3427585156606936,
      "grad_norm": 0.4500001072883606,
      "learning_rate": 4.3816098955953775e-06,
      "loss": 0.0369,
      "step": 46150
    },
    {
      "epoch": 2.3430123356515558,
      "grad_norm": 0.32434019446372986,
      "learning_rate": 4.379917762322961e-06,
      "loss": 0.0362,
      "step": 46155
    },
    {
      "epoch": 2.3432661556424184,
      "grad_norm": 0.40813419222831726,
      "learning_rate": 4.3782256290505445e-06,
      "loss": 0.0308,
      "step": 46160
    },
    {
      "epoch": 2.343519975633281,
      "grad_norm": 0.3180595636367798,
      "learning_rate": 4.376533495778128e-06,
      "loss": 0.0303,
      "step": 46165
    },
    {
      "epoch": 2.343773795624143,
      "grad_norm": 0.3424380123615265,
      "learning_rate": 4.374841362505712e-06,
      "loss": 0.0317,
      "step": 46170
    },
    {
      "epoch": 2.3440276156150057,
      "grad_norm": 0.4235934019088745,
      "learning_rate": 4.373149229233295e-06,
      "loss": 0.0366,
      "step": 46175
    },
    {
      "epoch": 2.3442814356058683,
      "grad_norm": 0.29761865735054016,
      "learning_rate": 4.371457095960878e-06,
      "loss": 0.0312,
      "step": 46180
    },
    {
      "epoch": 2.344535255596731,
      "grad_norm": 0.34761884808540344,
      "learning_rate": 4.369764962688462e-06,
      "loss": 0.0273,
      "step": 46185
    },
    {
      "epoch": 2.344789075587593,
      "grad_norm": 0.7427986860275269,
      "learning_rate": 4.368072829416045e-06,
      "loss": 0.0321,
      "step": 46190
    },
    {
      "epoch": 2.3450428955784557,
      "grad_norm": 0.42068642377853394,
      "learning_rate": 4.3663806961436285e-06,
      "loss": 0.0328,
      "step": 46195
    },
    {
      "epoch": 2.3452967155693183,
      "grad_norm": 0.23314493894577026,
      "learning_rate": 4.364688562871212e-06,
      "loss": 0.0296,
      "step": 46200
    },
    {
      "epoch": 2.345550535560181,
      "grad_norm": 0.561964750289917,
      "learning_rate": 4.362996429598796e-06,
      "loss": 0.0316,
      "step": 46205
    },
    {
      "epoch": 2.345804355551043,
      "grad_norm": 0.2341952621936798,
      "learning_rate": 4.361304296326379e-06,
      "loss": 0.0307,
      "step": 46210
    },
    {
      "epoch": 2.3460581755419057,
      "grad_norm": 0.3044934868812561,
      "learning_rate": 4.359612163053963e-06,
      "loss": 0.0316,
      "step": 46215
    },
    {
      "epoch": 2.3463119955327683,
      "grad_norm": 0.29033297300338745,
      "learning_rate": 4.357920029781546e-06,
      "loss": 0.0257,
      "step": 46220
    },
    {
      "epoch": 2.346565815523631,
      "grad_norm": 0.548446774482727,
      "learning_rate": 4.356227896509129e-06,
      "loss": 0.0297,
      "step": 46225
    },
    {
      "epoch": 2.346819635514493,
      "grad_norm": 0.43785858154296875,
      "learning_rate": 4.354535763236713e-06,
      "loss": 0.0323,
      "step": 46230
    },
    {
      "epoch": 2.3470734555053556,
      "grad_norm": 0.303646445274353,
      "learning_rate": 4.352843629964296e-06,
      "loss": 0.0285,
      "step": 46235
    },
    {
      "epoch": 2.3473272754962182,
      "grad_norm": 0.32536157965660095,
      "learning_rate": 4.3511514966918796e-06,
      "loss": 0.0359,
      "step": 46240
    },
    {
      "epoch": 2.3475810954870804,
      "grad_norm": 0.22959794104099274,
      "learning_rate": 4.349459363419463e-06,
      "loss": 0.0364,
      "step": 46245
    },
    {
      "epoch": 2.347834915477943,
      "grad_norm": 0.24991437792778015,
      "learning_rate": 4.347767230147047e-06,
      "loss": 0.0268,
      "step": 46250
    },
    {
      "epoch": 2.3480887354688056,
      "grad_norm": 0.4207878112792969,
      "learning_rate": 4.34607509687463e-06,
      "loss": 0.0363,
      "step": 46255
    },
    {
      "epoch": 2.348342555459668,
      "grad_norm": 0.2433720827102661,
      "learning_rate": 4.344382963602214e-06,
      "loss": 0.0365,
      "step": 46260
    },
    {
      "epoch": 2.3485963754505303,
      "grad_norm": 0.4478245675563812,
      "learning_rate": 4.342690830329797e-06,
      "loss": 0.0351,
      "step": 46265
    },
    {
      "epoch": 2.348850195441393,
      "grad_norm": 0.3620702922344208,
      "learning_rate": 4.340998697057381e-06,
      "loss": 0.0285,
      "step": 46270
    },
    {
      "epoch": 2.3491040154322556,
      "grad_norm": 0.46670037508010864,
      "learning_rate": 4.339306563784964e-06,
      "loss": 0.0309,
      "step": 46275
    },
    {
      "epoch": 2.3493578354231177,
      "grad_norm": 0.5231664180755615,
      "learning_rate": 4.337614430512547e-06,
      "loss": 0.0299,
      "step": 46280
    },
    {
      "epoch": 2.3496116554139803,
      "grad_norm": 0.32671064138412476,
      "learning_rate": 4.335922297240131e-06,
      "loss": 0.0333,
      "step": 46285
    },
    {
      "epoch": 2.349865475404843,
      "grad_norm": 0.6618512272834778,
      "learning_rate": 4.334230163967714e-06,
      "loss": 0.0348,
      "step": 46290
    },
    {
      "epoch": 2.3501192953957055,
      "grad_norm": 0.2867075800895691,
      "learning_rate": 4.332538030695298e-06,
      "loss": 0.0317,
      "step": 46295
    },
    {
      "epoch": 2.3503731153865677,
      "grad_norm": 0.31707561016082764,
      "learning_rate": 4.330845897422881e-06,
      "loss": 0.0347,
      "step": 46300
    },
    {
      "epoch": 2.3506269353774303,
      "grad_norm": 0.21336521208286285,
      "learning_rate": 4.329153764150465e-06,
      "loss": 0.0283,
      "step": 46305
    },
    {
      "epoch": 2.350880755368293,
      "grad_norm": 0.24199140071868896,
      "learning_rate": 4.327461630878048e-06,
      "loss": 0.0398,
      "step": 46310
    },
    {
      "epoch": 2.351134575359155,
      "grad_norm": 0.390618234872818,
      "learning_rate": 4.325769497605632e-06,
      "loss": 0.0352,
      "step": 46315
    },
    {
      "epoch": 2.3513883953500176,
      "grad_norm": 0.32350775599479675,
      "learning_rate": 4.3240773643332154e-06,
      "loss": 0.0362,
      "step": 46320
    },
    {
      "epoch": 2.3516422153408802,
      "grad_norm": 0.3590441346168518,
      "learning_rate": 4.322385231060799e-06,
      "loss": 0.0316,
      "step": 46325
    },
    {
      "epoch": 2.351896035331743,
      "grad_norm": 0.33314502239227295,
      "learning_rate": 4.320693097788382e-06,
      "loss": 0.0355,
      "step": 46330
    },
    {
      "epoch": 2.3521498553226055,
      "grad_norm": 0.3136577606201172,
      "learning_rate": 4.319000964515966e-06,
      "loss": 0.0377,
      "step": 46335
    },
    {
      "epoch": 2.3524036753134676,
      "grad_norm": 0.29925400018692017,
      "learning_rate": 4.317308831243549e-06,
      "loss": 0.0308,
      "step": 46340
    },
    {
      "epoch": 2.35265749530433,
      "grad_norm": 0.37032827734947205,
      "learning_rate": 4.315616697971133e-06,
      "loss": 0.0302,
      "step": 46345
    },
    {
      "epoch": 2.352911315295193,
      "grad_norm": 0.2775189280509949,
      "learning_rate": 4.313924564698716e-06,
      "loss": 0.0332,
      "step": 46350
    },
    {
      "epoch": 2.353165135286055,
      "grad_norm": 0.507088840007782,
      "learning_rate": 4.312232431426299e-06,
      "loss": 0.0291,
      "step": 46355
    },
    {
      "epoch": 2.3534189552769176,
      "grad_norm": 0.4810927212238312,
      "learning_rate": 4.310540298153883e-06,
      "loss": 0.0331,
      "step": 46360
    },
    {
      "epoch": 2.35367277526778,
      "grad_norm": 0.3024055063724518,
      "learning_rate": 4.3088481648814665e-06,
      "loss": 0.0303,
      "step": 46365
    },
    {
      "epoch": 2.353926595258643,
      "grad_norm": 0.3359050452709198,
      "learning_rate": 4.30715603160905e-06,
      "loss": 0.0359,
      "step": 46370
    },
    {
      "epoch": 2.354180415249505,
      "grad_norm": 0.4089173376560211,
      "learning_rate": 4.305463898336633e-06,
      "loss": 0.0406,
      "step": 46375
    },
    {
      "epoch": 2.3544342352403675,
      "grad_norm": 0.2997671067714691,
      "learning_rate": 4.303771765064217e-06,
      "loss": 0.0391,
      "step": 46380
    },
    {
      "epoch": 2.35468805523123,
      "grad_norm": 0.2722918391227722,
      "learning_rate": 4.3020796317918e-06,
      "loss": 0.0277,
      "step": 46385
    },
    {
      "epoch": 2.3549418752220923,
      "grad_norm": 0.29539498686790466,
      "learning_rate": 4.300387498519384e-06,
      "loss": 0.026,
      "step": 46390
    },
    {
      "epoch": 2.355195695212955,
      "grad_norm": 0.30494311451911926,
      "learning_rate": 4.298695365246967e-06,
      "loss": 0.0283,
      "step": 46395
    },
    {
      "epoch": 2.3554495152038175,
      "grad_norm": 0.3142552971839905,
      "learning_rate": 4.2970032319745504e-06,
      "loss": 0.031,
      "step": 46400
    },
    {
      "epoch": 2.35570333519468,
      "grad_norm": 0.3581981658935547,
      "learning_rate": 4.295311098702134e-06,
      "loss": 0.0342,
      "step": 46405
    },
    {
      "epoch": 2.3559571551855423,
      "grad_norm": 0.2926599979400635,
      "learning_rate": 4.2936189654297175e-06,
      "loss": 0.028,
      "step": 46410
    },
    {
      "epoch": 2.356210975176405,
      "grad_norm": 0.44851022958755493,
      "learning_rate": 4.291926832157301e-06,
      "loss": 0.0327,
      "step": 46415
    },
    {
      "epoch": 2.3564647951672675,
      "grad_norm": 0.5071258544921875,
      "learning_rate": 4.290234698884885e-06,
      "loss": 0.026,
      "step": 46420
    },
    {
      "epoch": 2.3567186151581296,
      "grad_norm": 0.48489269614219666,
      "learning_rate": 4.288542565612468e-06,
      "loss": 0.0395,
      "step": 46425
    },
    {
      "epoch": 2.3569724351489922,
      "grad_norm": 0.34989824891090393,
      "learning_rate": 4.286850432340052e-06,
      "loss": 0.0262,
      "step": 46430
    },
    {
      "epoch": 2.357226255139855,
      "grad_norm": 0.2624622881412506,
      "learning_rate": 4.285158299067635e-06,
      "loss": 0.0319,
      "step": 46435
    },
    {
      "epoch": 2.3574800751307174,
      "grad_norm": 0.2850838005542755,
      "learning_rate": 4.283466165795218e-06,
      "loss": 0.0328,
      "step": 46440
    },
    {
      "epoch": 2.3577338951215796,
      "grad_norm": 0.38306334614753723,
      "learning_rate": 4.2817740325228015e-06,
      "loss": 0.0286,
      "step": 46445
    },
    {
      "epoch": 2.357987715112442,
      "grad_norm": 0.2267332822084427,
      "learning_rate": 4.280081899250385e-06,
      "loss": 0.0293,
      "step": 46450
    },
    {
      "epoch": 2.358241535103305,
      "grad_norm": 0.4027699828147888,
      "learning_rate": 4.278389765977969e-06,
      "loss": 0.0368,
      "step": 46455
    },
    {
      "epoch": 2.358495355094167,
      "grad_norm": 0.2977030575275421,
      "learning_rate": 4.276697632705552e-06,
      "loss": 0.0298,
      "step": 46460
    },
    {
      "epoch": 2.3587491750850296,
      "grad_norm": 0.3098295331001282,
      "learning_rate": 4.275005499433136e-06,
      "loss": 0.0395,
      "step": 46465
    },
    {
      "epoch": 2.359002995075892,
      "grad_norm": 0.3363555371761322,
      "learning_rate": 4.273313366160719e-06,
      "loss": 0.0312,
      "step": 46470
    },
    {
      "epoch": 2.3592568150667548,
      "grad_norm": 1.3011178970336914,
      "learning_rate": 4.271621232888303e-06,
      "loss": 0.0378,
      "step": 46475
    },
    {
      "epoch": 2.3595106350576174,
      "grad_norm": 0.28810590505599976,
      "learning_rate": 4.269929099615886e-06,
      "loss": 0.0379,
      "step": 46480
    },
    {
      "epoch": 2.3597644550484795,
      "grad_norm": 0.36974385380744934,
      "learning_rate": 4.26823696634347e-06,
      "loss": 0.0277,
      "step": 46485
    },
    {
      "epoch": 2.360018275039342,
      "grad_norm": 0.2768723666667938,
      "learning_rate": 4.2665448330710526e-06,
      "loss": 0.0357,
      "step": 46490
    },
    {
      "epoch": 2.3602720950302047,
      "grad_norm": 0.25840070843696594,
      "learning_rate": 4.264852699798637e-06,
      "loss": 0.0305,
      "step": 46495
    },
    {
      "epoch": 2.360525915021067,
      "grad_norm": 0.5833351612091064,
      "learning_rate": 4.26316056652622e-06,
      "loss": 0.0462,
      "step": 46500
    },
    {
      "epoch": 2.3607797350119295,
      "grad_norm": 0.30810680985450745,
      "learning_rate": 4.261468433253804e-06,
      "loss": 0.0297,
      "step": 46505
    },
    {
      "epoch": 2.361033555002792,
      "grad_norm": 0.27955737709999084,
      "learning_rate": 4.259776299981387e-06,
      "loss": 0.0311,
      "step": 46510
    },
    {
      "epoch": 2.3612873749936547,
      "grad_norm": 0.2240092009305954,
      "learning_rate": 4.25808416670897e-06,
      "loss": 0.0276,
      "step": 46515
    },
    {
      "epoch": 2.361541194984517,
      "grad_norm": 0.5027244091033936,
      "learning_rate": 4.256392033436554e-06,
      "loss": 0.0285,
      "step": 46520
    },
    {
      "epoch": 2.3617950149753795,
      "grad_norm": 0.32359835505485535,
      "learning_rate": 4.254699900164137e-06,
      "loss": 0.033,
      "step": 46525
    },
    {
      "epoch": 2.362048834966242,
      "grad_norm": 0.3147777318954468,
      "learning_rate": 4.253007766891721e-06,
      "loss": 0.0302,
      "step": 46530
    },
    {
      "epoch": 2.3623026549571042,
      "grad_norm": 0.36554408073425293,
      "learning_rate": 4.251315633619304e-06,
      "loss": 0.0303,
      "step": 46535
    },
    {
      "epoch": 2.362556474947967,
      "grad_norm": 0.7193531394004822,
      "learning_rate": 4.249623500346888e-06,
      "loss": 0.035,
      "step": 46540
    },
    {
      "epoch": 2.3628102949388294,
      "grad_norm": 0.16339777410030365,
      "learning_rate": 4.247931367074471e-06,
      "loss": 0.0345,
      "step": 46545
    },
    {
      "epoch": 2.363064114929692,
      "grad_norm": 0.294882208108902,
      "learning_rate": 4.246239233802055e-06,
      "loss": 0.0387,
      "step": 46550
    },
    {
      "epoch": 2.363317934920554,
      "grad_norm": 0.3476087749004364,
      "learning_rate": 4.244547100529638e-06,
      "loss": 0.0319,
      "step": 46555
    },
    {
      "epoch": 2.363571754911417,
      "grad_norm": 0.43848177790641785,
      "learning_rate": 4.242854967257221e-06,
      "loss": 0.0282,
      "step": 46560
    },
    {
      "epoch": 2.3638255749022794,
      "grad_norm": 0.30569490790367126,
      "learning_rate": 4.241162833984805e-06,
      "loss": 0.0381,
      "step": 46565
    },
    {
      "epoch": 2.3640793948931416,
      "grad_norm": 0.39467477798461914,
      "learning_rate": 4.239470700712388e-06,
      "loss": 0.0347,
      "step": 46570
    },
    {
      "epoch": 2.364333214884004,
      "grad_norm": 0.25366389751434326,
      "learning_rate": 4.237778567439972e-06,
      "loss": 0.0332,
      "step": 46575
    },
    {
      "epoch": 2.3645870348748668,
      "grad_norm": 0.31514278054237366,
      "learning_rate": 4.2360864341675555e-06,
      "loss": 0.0361,
      "step": 46580
    },
    {
      "epoch": 2.3648408548657294,
      "grad_norm": 0.51214200258255,
      "learning_rate": 4.234394300895139e-06,
      "loss": 0.0289,
      "step": 46585
    },
    {
      "epoch": 2.3650946748565915,
      "grad_norm": 0.25476548075675964,
      "learning_rate": 4.232702167622723e-06,
      "loss": 0.0329,
      "step": 46590
    },
    {
      "epoch": 2.365348494847454,
      "grad_norm": 0.30458933115005493,
      "learning_rate": 4.231010034350306e-06,
      "loss": 0.0289,
      "step": 46595
    },
    {
      "epoch": 2.3656023148383167,
      "grad_norm": 0.2518400251865387,
      "learning_rate": 4.229317901077889e-06,
      "loss": 0.0251,
      "step": 46600
    },
    {
      "epoch": 2.3658561348291793,
      "grad_norm": 0.23019066452980042,
      "learning_rate": 4.227625767805472e-06,
      "loss": 0.0323,
      "step": 46605
    },
    {
      "epoch": 2.3661099548200415,
      "grad_norm": 0.2931053042411804,
      "learning_rate": 4.225933634533056e-06,
      "loss": 0.0279,
      "step": 46610
    },
    {
      "epoch": 2.366363774810904,
      "grad_norm": 0.8957230448722839,
      "learning_rate": 4.2242415012606395e-06,
      "loss": 0.0303,
      "step": 46615
    },
    {
      "epoch": 2.3666175948017667,
      "grad_norm": 0.35307177901268005,
      "learning_rate": 4.222549367988223e-06,
      "loss": 0.0255,
      "step": 46620
    },
    {
      "epoch": 2.3668714147926293,
      "grad_norm": 0.3644431233406067,
      "learning_rate": 4.2208572347158066e-06,
      "loss": 0.0366,
      "step": 46625
    },
    {
      "epoch": 2.3671252347834915,
      "grad_norm": 0.2834033668041229,
      "learning_rate": 4.21916510144339e-06,
      "loss": 0.0338,
      "step": 46630
    },
    {
      "epoch": 2.367379054774354,
      "grad_norm": 0.29104098677635193,
      "learning_rate": 4.217472968170974e-06,
      "loss": 0.0363,
      "step": 46635
    },
    {
      "epoch": 2.3676328747652167,
      "grad_norm": 0.26519590616226196,
      "learning_rate": 4.215780834898557e-06,
      "loss": 0.0341,
      "step": 46640
    },
    {
      "epoch": 2.367886694756079,
      "grad_norm": 0.6125626564025879,
      "learning_rate": 4.214088701626141e-06,
      "loss": 0.0253,
      "step": 46645
    },
    {
      "epoch": 2.3681405147469414,
      "grad_norm": 0.3606280982494354,
      "learning_rate": 4.2123965683537234e-06,
      "loss": 0.0269,
      "step": 46650
    },
    {
      "epoch": 2.368394334737804,
      "grad_norm": 0.4279687702655792,
      "learning_rate": 4.210704435081308e-06,
      "loss": 0.0278,
      "step": 46655
    },
    {
      "epoch": 2.3686481547286666,
      "grad_norm": 0.6018189191818237,
      "learning_rate": 4.2090123018088905e-06,
      "loss": 0.0353,
      "step": 46660
    },
    {
      "epoch": 2.368901974719529,
      "grad_norm": 0.38729801774024963,
      "learning_rate": 4.207320168536475e-06,
      "loss": 0.0284,
      "step": 46665
    },
    {
      "epoch": 2.3691557947103914,
      "grad_norm": 0.35448938608169556,
      "learning_rate": 4.205628035264058e-06,
      "loss": 0.0268,
      "step": 46670
    },
    {
      "epoch": 2.369409614701254,
      "grad_norm": 0.25390028953552246,
      "learning_rate": 4.203935901991641e-06,
      "loss": 0.025,
      "step": 46675
    },
    {
      "epoch": 2.369663434692116,
      "grad_norm": 0.22552552819252014,
      "learning_rate": 4.202243768719225e-06,
      "loss": 0.0268,
      "step": 46680
    },
    {
      "epoch": 2.3699172546829788,
      "grad_norm": 0.6019819378852844,
      "learning_rate": 4.200551635446808e-06,
      "loss": 0.0412,
      "step": 46685
    },
    {
      "epoch": 2.3701710746738414,
      "grad_norm": 0.3028351068496704,
      "learning_rate": 4.198859502174392e-06,
      "loss": 0.0335,
      "step": 46690
    },
    {
      "epoch": 2.370424894664704,
      "grad_norm": 0.4909636378288269,
      "learning_rate": 4.1971673689019745e-06,
      "loss": 0.038,
      "step": 46695
    },
    {
      "epoch": 2.370678714655566,
      "grad_norm": 0.37978753447532654,
      "learning_rate": 4.195475235629559e-06,
      "loss": 0.0307,
      "step": 46700
    },
    {
      "epoch": 2.3709325346464287,
      "grad_norm": 0.16427743434906006,
      "learning_rate": 4.193783102357142e-06,
      "loss": 0.0276,
      "step": 46705
    },
    {
      "epoch": 2.3711863546372913,
      "grad_norm": 0.4129177927970886,
      "learning_rate": 4.192090969084726e-06,
      "loss": 0.0313,
      "step": 46710
    },
    {
      "epoch": 2.3714401746281535,
      "grad_norm": 0.5559883117675781,
      "learning_rate": 4.190398835812309e-06,
      "loss": 0.0342,
      "step": 46715
    },
    {
      "epoch": 2.371693994619016,
      "grad_norm": 0.32237133383750916,
      "learning_rate": 4.188706702539892e-06,
      "loss": 0.0245,
      "step": 46720
    },
    {
      "epoch": 2.3719478146098787,
      "grad_norm": 0.2823083996772766,
      "learning_rate": 4.187014569267476e-06,
      "loss": 0.0255,
      "step": 46725
    },
    {
      "epoch": 2.3722016346007413,
      "grad_norm": 0.5001686215400696,
      "learning_rate": 4.185322435995059e-06,
      "loss": 0.0336,
      "step": 46730
    },
    {
      "epoch": 2.3724554545916035,
      "grad_norm": 0.26875582337379456,
      "learning_rate": 4.183630302722643e-06,
      "loss": 0.0395,
      "step": 46735
    },
    {
      "epoch": 2.372709274582466,
      "grad_norm": 0.4248732030391693,
      "learning_rate": 4.181938169450226e-06,
      "loss": 0.0278,
      "step": 46740
    },
    {
      "epoch": 2.3729630945733287,
      "grad_norm": 0.3200513422489166,
      "learning_rate": 4.18024603617781e-06,
      "loss": 0.0329,
      "step": 46745
    },
    {
      "epoch": 2.3732169145641913,
      "grad_norm": 0.38520458340644836,
      "learning_rate": 4.1785539029053935e-06,
      "loss": 0.0367,
      "step": 46750
    },
    {
      "epoch": 2.3734707345550534,
      "grad_norm": 0.3649415671825409,
      "learning_rate": 4.176861769632977e-06,
      "loss": 0.033,
      "step": 46755
    },
    {
      "epoch": 2.373724554545916,
      "grad_norm": 0.38120341300964355,
      "learning_rate": 4.17516963636056e-06,
      "loss": 0.0311,
      "step": 46760
    },
    {
      "epoch": 2.3739783745367786,
      "grad_norm": 0.2855028212070465,
      "learning_rate": 4.173477503088143e-06,
      "loss": 0.036,
      "step": 46765
    },
    {
      "epoch": 2.3742321945276412,
      "grad_norm": 0.4050988554954529,
      "learning_rate": 4.171785369815727e-06,
      "loss": 0.0362,
      "step": 46770
    },
    {
      "epoch": 2.3744860145185034,
      "grad_norm": 0.6171091198921204,
      "learning_rate": 4.17009323654331e-06,
      "loss": 0.0407,
      "step": 46775
    },
    {
      "epoch": 2.374739834509366,
      "grad_norm": 0.29305702447891235,
      "learning_rate": 4.168401103270894e-06,
      "loss": 0.0277,
      "step": 46780
    },
    {
      "epoch": 2.3749936545002286,
      "grad_norm": 0.3541305363178253,
      "learning_rate": 4.1667089699984774e-06,
      "loss": 0.033,
      "step": 46785
    },
    {
      "epoch": 2.3752474744910907,
      "grad_norm": 0.34138309955596924,
      "learning_rate": 4.165016836726061e-06,
      "loss": 0.0317,
      "step": 46790
    },
    {
      "epoch": 2.3755012944819534,
      "grad_norm": 0.34759989380836487,
      "learning_rate": 4.1633247034536445e-06,
      "loss": 0.0289,
      "step": 46795
    },
    {
      "epoch": 2.375755114472816,
      "grad_norm": 0.4288697838783264,
      "learning_rate": 4.161632570181228e-06,
      "loss": 0.0336,
      "step": 46800
    },
    {
      "epoch": 2.3760089344636786,
      "grad_norm": 0.31496793031692505,
      "learning_rate": 4.159940436908812e-06,
      "loss": 0.0328,
      "step": 46805
    },
    {
      "epoch": 2.3762627544545407,
      "grad_norm": 0.3088779151439667,
      "learning_rate": 4.158248303636394e-06,
      "loss": 0.0293,
      "step": 46810
    },
    {
      "epoch": 2.3765165744454033,
      "grad_norm": 0.34354227781295776,
      "learning_rate": 4.156556170363979e-06,
      "loss": 0.0322,
      "step": 46815
    },
    {
      "epoch": 2.376770394436266,
      "grad_norm": 0.3034856617450714,
      "learning_rate": 4.154864037091561e-06,
      "loss": 0.0389,
      "step": 46820
    },
    {
      "epoch": 2.377024214427128,
      "grad_norm": 0.47851690649986267,
      "learning_rate": 4.153171903819145e-06,
      "loss": 0.0375,
      "step": 46825
    },
    {
      "epoch": 2.3772780344179907,
      "grad_norm": 0.3652953505516052,
      "learning_rate": 4.1514797705467285e-06,
      "loss": 0.0355,
      "step": 46830
    },
    {
      "epoch": 2.3775318544088533,
      "grad_norm": 0.5073241591453552,
      "learning_rate": 4.149787637274312e-06,
      "loss": 0.027,
      "step": 46835
    },
    {
      "epoch": 2.377785674399716,
      "grad_norm": 0.27668410539627075,
      "learning_rate": 4.148095504001896e-06,
      "loss": 0.0298,
      "step": 46840
    },
    {
      "epoch": 2.378039494390578,
      "grad_norm": 0.47308704257011414,
      "learning_rate": 4.146403370729479e-06,
      "loss": 0.0376,
      "step": 46845
    },
    {
      "epoch": 2.3782933143814406,
      "grad_norm": 0.35060396790504456,
      "learning_rate": 4.144711237457063e-06,
      "loss": 0.0356,
      "step": 46850
    },
    {
      "epoch": 2.3785471343723033,
      "grad_norm": 0.28105148673057556,
      "learning_rate": 4.143019104184645e-06,
      "loss": 0.0359,
      "step": 46855
    },
    {
      "epoch": 2.3788009543631654,
      "grad_norm": 0.39512941241264343,
      "learning_rate": 4.14132697091223e-06,
      "loss": 0.0326,
      "step": 46860
    },
    {
      "epoch": 2.379054774354028,
      "grad_norm": 0.36061257123947144,
      "learning_rate": 4.1396348376398125e-06,
      "loss": 0.0334,
      "step": 46865
    },
    {
      "epoch": 2.3793085943448906,
      "grad_norm": 1.4042720794677734,
      "learning_rate": 4.137942704367396e-06,
      "loss": 0.0331,
      "step": 46870
    },
    {
      "epoch": 2.379562414335753,
      "grad_norm": 0.33557531237602234,
      "learning_rate": 4.1362505710949796e-06,
      "loss": 0.0289,
      "step": 46875
    },
    {
      "epoch": 2.3798162343266154,
      "grad_norm": 0.2567538619041443,
      "learning_rate": 4.134558437822563e-06,
      "loss": 0.0302,
      "step": 46880
    },
    {
      "epoch": 2.380070054317478,
      "grad_norm": 0.48120400309562683,
      "learning_rate": 4.132866304550147e-06,
      "loss": 0.0302,
      "step": 46885
    },
    {
      "epoch": 2.3803238743083406,
      "grad_norm": 0.30264732241630554,
      "learning_rate": 4.13117417127773e-06,
      "loss": 0.0312,
      "step": 46890
    },
    {
      "epoch": 2.380577694299203,
      "grad_norm": 0.2624591588973999,
      "learning_rate": 4.129482038005314e-06,
      "loss": 0.0309,
      "step": 46895
    },
    {
      "epoch": 2.3808315142900653,
      "grad_norm": 0.2809188663959503,
      "learning_rate": 4.127789904732897e-06,
      "loss": 0.0273,
      "step": 46900
    },
    {
      "epoch": 2.381085334280928,
      "grad_norm": 0.5473712086677551,
      "learning_rate": 4.126097771460481e-06,
      "loss": 0.0314,
      "step": 46905
    },
    {
      "epoch": 2.3813391542717905,
      "grad_norm": 0.3612068295478821,
      "learning_rate": 4.124405638188064e-06,
      "loss": 0.0324,
      "step": 46910
    },
    {
      "epoch": 2.381592974262653,
      "grad_norm": 0.30368199944496155,
      "learning_rate": 4.122713504915647e-06,
      "loss": 0.0367,
      "step": 46915
    },
    {
      "epoch": 2.3818467942535153,
      "grad_norm": 0.35154810547828674,
      "learning_rate": 4.121021371643231e-06,
      "loss": 0.0321,
      "step": 46920
    },
    {
      "epoch": 2.382100614244378,
      "grad_norm": 0.48841744661331177,
      "learning_rate": 4.119329238370814e-06,
      "loss": 0.0326,
      "step": 46925
    },
    {
      "epoch": 2.3823544342352405,
      "grad_norm": 0.3970104455947876,
      "learning_rate": 4.117637105098398e-06,
      "loss": 0.0288,
      "step": 46930
    },
    {
      "epoch": 2.3826082542261027,
      "grad_norm": 0.32299691438674927,
      "learning_rate": 4.115944971825981e-06,
      "loss": 0.0282,
      "step": 46935
    },
    {
      "epoch": 2.3828620742169653,
      "grad_norm": 0.41886699199676514,
      "learning_rate": 4.114252838553565e-06,
      "loss": 0.0368,
      "step": 46940
    },
    {
      "epoch": 2.383115894207828,
      "grad_norm": 0.36832237243652344,
      "learning_rate": 4.112560705281148e-06,
      "loss": 0.033,
      "step": 46945
    },
    {
      "epoch": 2.3833697141986905,
      "grad_norm": 0.48107612133026123,
      "learning_rate": 4.110868572008732e-06,
      "loss": 0.0363,
      "step": 46950
    },
    {
      "epoch": 2.3836235341895526,
      "grad_norm": 0.31277942657470703,
      "learning_rate": 4.109176438736315e-06,
      "loss": 0.0266,
      "step": 46955
    },
    {
      "epoch": 2.3838773541804152,
      "grad_norm": 0.28694868087768555,
      "learning_rate": 4.107484305463898e-06,
      "loss": 0.032,
      "step": 46960
    },
    {
      "epoch": 2.384131174171278,
      "grad_norm": 0.3679179847240448,
      "learning_rate": 4.1057921721914825e-06,
      "loss": 0.0279,
      "step": 46965
    },
    {
      "epoch": 2.38438499416214,
      "grad_norm": 0.19425620138645172,
      "learning_rate": 4.104100038919065e-06,
      "loss": 0.0261,
      "step": 46970
    },
    {
      "epoch": 2.3846388141530026,
      "grad_norm": 0.46411842107772827,
      "learning_rate": 4.10240790564665e-06,
      "loss": 0.0361,
      "step": 46975
    },
    {
      "epoch": 2.384892634143865,
      "grad_norm": 0.6477605700492859,
      "learning_rate": 4.100715772374232e-06,
      "loss": 0.0352,
      "step": 46980
    },
    {
      "epoch": 2.385146454134728,
      "grad_norm": 0.43195345997810364,
      "learning_rate": 4.099023639101816e-06,
      "loss": 0.029,
      "step": 46985
    },
    {
      "epoch": 2.38540027412559,
      "grad_norm": 0.2981319725513458,
      "learning_rate": 4.097331505829399e-06,
      "loss": 0.0301,
      "step": 46990
    },
    {
      "epoch": 2.3856540941164526,
      "grad_norm": 0.2841643989086151,
      "learning_rate": 4.095639372556983e-06,
      "loss": 0.0283,
      "step": 46995
    },
    {
      "epoch": 2.385907914107315,
      "grad_norm": 0.37677624821662903,
      "learning_rate": 4.0939472392845665e-06,
      "loss": 0.0302,
      "step": 47000
    },
    {
      "epoch": 2.3861617340981773,
      "grad_norm": 0.2917637825012207,
      "learning_rate": 4.092255106012149e-06,
      "loss": 0.0326,
      "step": 47005
    },
    {
      "epoch": 2.38641555408904,
      "grad_norm": 0.23505131900310516,
      "learning_rate": 4.0905629727397336e-06,
      "loss": 0.0268,
      "step": 47010
    },
    {
      "epoch": 2.3866693740799025,
      "grad_norm": 0.36129844188690186,
      "learning_rate": 4.088870839467316e-06,
      "loss": 0.0317,
      "step": 47015
    },
    {
      "epoch": 2.386923194070765,
      "grad_norm": 0.4052008092403412,
      "learning_rate": 4.087178706194901e-06,
      "loss": 0.026,
      "step": 47020
    },
    {
      "epoch": 2.3871770140616273,
      "grad_norm": 0.35589584708213806,
      "learning_rate": 4.085486572922483e-06,
      "loss": 0.0349,
      "step": 47025
    },
    {
      "epoch": 2.38743083405249,
      "grad_norm": 0.19920487701892853,
      "learning_rate": 4.083794439650067e-06,
      "loss": 0.0345,
      "step": 47030
    },
    {
      "epoch": 2.3876846540433525,
      "grad_norm": 0.7072770595550537,
      "learning_rate": 4.0821023063776504e-06,
      "loss": 0.0264,
      "step": 47035
    },
    {
      "epoch": 2.387938474034215,
      "grad_norm": 0.5042538046836853,
      "learning_rate": 4.080410173105234e-06,
      "loss": 0.0352,
      "step": 47040
    },
    {
      "epoch": 2.3881922940250773,
      "grad_norm": 0.3139657974243164,
      "learning_rate": 4.0787180398328175e-06,
      "loss": 0.0353,
      "step": 47045
    },
    {
      "epoch": 2.38844611401594,
      "grad_norm": 0.7741235494613647,
      "learning_rate": 4.077025906560401e-06,
      "loss": 0.0345,
      "step": 47050
    },
    {
      "epoch": 2.3886999340068025,
      "grad_norm": 0.26117607951164246,
      "learning_rate": 4.075333773287985e-06,
      "loss": 0.0259,
      "step": 47055
    },
    {
      "epoch": 2.388953753997665,
      "grad_norm": 0.28457748889923096,
      "learning_rate": 4.073641640015568e-06,
      "loss": 0.0353,
      "step": 47060
    },
    {
      "epoch": 2.3892075739885272,
      "grad_norm": 0.8835075497627258,
      "learning_rate": 4.071949506743152e-06,
      "loss": 0.0281,
      "step": 47065
    },
    {
      "epoch": 2.38946139397939,
      "grad_norm": 0.5735437870025635,
      "learning_rate": 4.070257373470735e-06,
      "loss": 0.0321,
      "step": 47070
    },
    {
      "epoch": 2.3897152139702524,
      "grad_norm": 0.3915027379989624,
      "learning_rate": 4.068565240198318e-06,
      "loss": 0.0358,
      "step": 47075
    },
    {
      "epoch": 2.3899690339611146,
      "grad_norm": 0.2943165600299835,
      "learning_rate": 4.066873106925902e-06,
      "loss": 0.0302,
      "step": 47080
    },
    {
      "epoch": 2.390222853951977,
      "grad_norm": 0.3288983106613159,
      "learning_rate": 4.065180973653485e-06,
      "loss": 0.0345,
      "step": 47085
    },
    {
      "epoch": 2.39047667394284,
      "grad_norm": 0.23327219486236572,
      "learning_rate": 4.063488840381069e-06,
      "loss": 0.0308,
      "step": 47090
    },
    {
      "epoch": 2.3907304939337024,
      "grad_norm": 0.3964608907699585,
      "learning_rate": 4.061796707108652e-06,
      "loss": 0.0268,
      "step": 47095
    },
    {
      "epoch": 2.3909843139245646,
      "grad_norm": 0.4795733094215393,
      "learning_rate": 4.060104573836236e-06,
      "loss": 0.0289,
      "step": 47100
    },
    {
      "epoch": 2.391238133915427,
      "grad_norm": 0.39220699667930603,
      "learning_rate": 4.058412440563819e-06,
      "loss": 0.0257,
      "step": 47105
    },
    {
      "epoch": 2.3914919539062898,
      "grad_norm": 0.26814162731170654,
      "learning_rate": 4.056720307291403e-06,
      "loss": 0.0325,
      "step": 47110
    },
    {
      "epoch": 2.391745773897152,
      "grad_norm": 0.3248618543148041,
      "learning_rate": 4.055028174018986e-06,
      "loss": 0.0287,
      "step": 47115
    },
    {
      "epoch": 2.3919995938880145,
      "grad_norm": 0.2833445966243744,
      "learning_rate": 4.053336040746569e-06,
      "loss": 0.0299,
      "step": 47120
    },
    {
      "epoch": 2.392253413878877,
      "grad_norm": 0.34968307614326477,
      "learning_rate": 4.051643907474153e-06,
      "loss": 0.0329,
      "step": 47125
    },
    {
      "epoch": 2.3925072338697397,
      "grad_norm": 0.34798914194107056,
      "learning_rate": 4.049951774201736e-06,
      "loss": 0.0359,
      "step": 47130
    },
    {
      "epoch": 2.392761053860602,
      "grad_norm": 0.4476436674594879,
      "learning_rate": 4.0482596409293205e-06,
      "loss": 0.0342,
      "step": 47135
    },
    {
      "epoch": 2.3930148738514645,
      "grad_norm": 0.366957426071167,
      "learning_rate": 4.046567507656903e-06,
      "loss": 0.0357,
      "step": 47140
    },
    {
      "epoch": 2.393268693842327,
      "grad_norm": 0.24836216866970062,
      "learning_rate": 4.044875374384487e-06,
      "loss": 0.0335,
      "step": 47145
    },
    {
      "epoch": 2.3935225138331893,
      "grad_norm": 0.3659680187702179,
      "learning_rate": 4.04318324111207e-06,
      "loss": 0.0311,
      "step": 47150
    },
    {
      "epoch": 2.393776333824052,
      "grad_norm": 0.19439701735973358,
      "learning_rate": 4.041491107839654e-06,
      "loss": 0.0325,
      "step": 47155
    },
    {
      "epoch": 2.3940301538149145,
      "grad_norm": 0.3122596740722656,
      "learning_rate": 4.039798974567237e-06,
      "loss": 0.0356,
      "step": 47160
    },
    {
      "epoch": 2.394283973805777,
      "grad_norm": 0.2920304834842682,
      "learning_rate": 4.03810684129482e-06,
      "loss": 0.0359,
      "step": 47165
    },
    {
      "epoch": 2.3945377937966397,
      "grad_norm": 0.2468995451927185,
      "learning_rate": 4.0364147080224044e-06,
      "loss": 0.0234,
      "step": 47170
    },
    {
      "epoch": 2.394791613787502,
      "grad_norm": 0.4460321068763733,
      "learning_rate": 4.034722574749987e-06,
      "loss": 0.0314,
      "step": 47175
    },
    {
      "epoch": 2.3950454337783644,
      "grad_norm": 0.40437963604927063,
      "learning_rate": 4.0330304414775715e-06,
      "loss": 0.033,
      "step": 47180
    },
    {
      "epoch": 2.395299253769227,
      "grad_norm": 0.30205151438713074,
      "learning_rate": 4.031338308205154e-06,
      "loss": 0.0348,
      "step": 47185
    },
    {
      "epoch": 2.395553073760089,
      "grad_norm": 0.2941863238811493,
      "learning_rate": 4.029646174932738e-06,
      "loss": 0.0267,
      "step": 47190
    },
    {
      "epoch": 2.395806893750952,
      "grad_norm": 0.3476633131504059,
      "learning_rate": 4.027954041660321e-06,
      "loss": 0.0342,
      "step": 47195
    },
    {
      "epoch": 2.3960607137418144,
      "grad_norm": 0.21167103946208954,
      "learning_rate": 4.026261908387905e-06,
      "loss": 0.0364,
      "step": 47200
    },
    {
      "epoch": 2.396314533732677,
      "grad_norm": 0.30532538890838623,
      "learning_rate": 4.024569775115488e-06,
      "loss": 0.0387,
      "step": 47205
    },
    {
      "epoch": 2.396568353723539,
      "grad_norm": 0.3205692172050476,
      "learning_rate": 4.022877641843072e-06,
      "loss": 0.0312,
      "step": 47210
    },
    {
      "epoch": 2.3968221737144018,
      "grad_norm": 0.428493857383728,
      "learning_rate": 4.0211855085706555e-06,
      "loss": 0.0291,
      "step": 47215
    },
    {
      "epoch": 2.3970759937052644,
      "grad_norm": 0.3529970943927765,
      "learning_rate": 4.019493375298239e-06,
      "loss": 0.0329,
      "step": 47220
    },
    {
      "epoch": 2.3973298136961265,
      "grad_norm": 0.24430900812149048,
      "learning_rate": 4.017801242025823e-06,
      "loss": 0.0268,
      "step": 47225
    },
    {
      "epoch": 2.397583633686989,
      "grad_norm": 0.2590664029121399,
      "learning_rate": 4.016109108753406e-06,
      "loss": 0.036,
      "step": 47230
    },
    {
      "epoch": 2.3978374536778517,
      "grad_norm": 0.3526705503463745,
      "learning_rate": 4.014416975480989e-06,
      "loss": 0.0345,
      "step": 47235
    },
    {
      "epoch": 2.3980912736687143,
      "grad_norm": 0.7975666522979736,
      "learning_rate": 4.012724842208573e-06,
      "loss": 0.0364,
      "step": 47240
    },
    {
      "epoch": 2.3983450936595765,
      "grad_norm": 0.23875541985034943,
      "learning_rate": 4.011032708936156e-06,
      "loss": 0.0347,
      "step": 47245
    },
    {
      "epoch": 2.398598913650439,
      "grad_norm": 0.23341935873031616,
      "learning_rate": 4.0093405756637395e-06,
      "loss": 0.0313,
      "step": 47250
    },
    {
      "epoch": 2.3988527336413017,
      "grad_norm": 0.2588426470756531,
      "learning_rate": 4.007648442391323e-06,
      "loss": 0.0337,
      "step": 47255
    },
    {
      "epoch": 2.399106553632164,
      "grad_norm": 0.4820748269557953,
      "learning_rate": 4.0059563091189066e-06,
      "loss": 0.0295,
      "step": 47260
    },
    {
      "epoch": 2.3993603736230265,
      "grad_norm": 0.623077929019928,
      "learning_rate": 4.00426417584649e-06,
      "loss": 0.0366,
      "step": 47265
    },
    {
      "epoch": 2.399614193613889,
      "grad_norm": 0.6706781387329102,
      "learning_rate": 4.002572042574074e-06,
      "loss": 0.0342,
      "step": 47270
    },
    {
      "epoch": 2.3998680136047517,
      "grad_norm": 0.32776203751564026,
      "learning_rate": 4.000879909301657e-06,
      "loss": 0.0244,
      "step": 47275
    },
    {
      "epoch": 2.400121833595614,
      "grad_norm": 0.29026415944099426,
      "learning_rate": 3.99918777602924e-06,
      "loss": 0.0299,
      "step": 47280
    },
    {
      "epoch": 2.4003756535864764,
      "grad_norm": 0.32502275705337524,
      "learning_rate": 3.997495642756824e-06,
      "loss": 0.0291,
      "step": 47285
    },
    {
      "epoch": 2.400629473577339,
      "grad_norm": 0.2750031352043152,
      "learning_rate": 3.995803509484407e-06,
      "loss": 0.0318,
      "step": 47290
    },
    {
      "epoch": 2.400883293568201,
      "grad_norm": 0.28516432642936707,
      "learning_rate": 3.994111376211991e-06,
      "loss": 0.0301,
      "step": 47295
    },
    {
      "epoch": 2.401137113559064,
      "grad_norm": 0.3644152283668518,
      "learning_rate": 3.992419242939574e-06,
      "loss": 0.0382,
      "step": 47300
    },
    {
      "epoch": 2.4013909335499264,
      "grad_norm": 0.3614419996738434,
      "learning_rate": 3.990727109667158e-06,
      "loss": 0.0358,
      "step": 47305
    },
    {
      "epoch": 2.401644753540789,
      "grad_norm": 0.3342830538749695,
      "learning_rate": 3.989034976394741e-06,
      "loss": 0.036,
      "step": 47310
    },
    {
      "epoch": 2.4018985735316516,
      "grad_norm": 0.26829349994659424,
      "learning_rate": 3.987342843122325e-06,
      "loss": 0.0273,
      "step": 47315
    },
    {
      "epoch": 2.4021523935225138,
      "grad_norm": 0.21632443368434906,
      "learning_rate": 3.985650709849908e-06,
      "loss": 0.028,
      "step": 47320
    },
    {
      "epoch": 2.4024062135133764,
      "grad_norm": 0.3586091697216034,
      "learning_rate": 3.983958576577491e-06,
      "loss": 0.0285,
      "step": 47325
    },
    {
      "epoch": 2.402660033504239,
      "grad_norm": 0.3348862826824188,
      "learning_rate": 3.982266443305075e-06,
      "loss": 0.0307,
      "step": 47330
    },
    {
      "epoch": 2.402913853495101,
      "grad_norm": 0.24888646602630615,
      "learning_rate": 3.980574310032658e-06,
      "loss": 0.0339,
      "step": 47335
    },
    {
      "epoch": 2.4031676734859637,
      "grad_norm": 0.5736678838729858,
      "learning_rate": 3.978882176760242e-06,
      "loss": 0.0384,
      "step": 47340
    },
    {
      "epoch": 2.4034214934768263,
      "grad_norm": 0.43109336495399475,
      "learning_rate": 3.977190043487825e-06,
      "loss": 0.0296,
      "step": 47345
    },
    {
      "epoch": 2.403675313467689,
      "grad_norm": 0.26424750685691833,
      "learning_rate": 3.975497910215409e-06,
      "loss": 0.0339,
      "step": 47350
    },
    {
      "epoch": 2.403929133458551,
      "grad_norm": 0.3025932312011719,
      "learning_rate": 3.973805776942992e-06,
      "loss": 0.0265,
      "step": 47355
    },
    {
      "epoch": 2.4041829534494137,
      "grad_norm": 0.5657501220703125,
      "learning_rate": 3.972113643670576e-06,
      "loss": 0.0304,
      "step": 47360
    },
    {
      "epoch": 2.4044367734402763,
      "grad_norm": 0.2696693539619446,
      "learning_rate": 3.970421510398159e-06,
      "loss": 0.0253,
      "step": 47365
    },
    {
      "epoch": 2.4046905934311384,
      "grad_norm": 0.2921593487262726,
      "learning_rate": 3.968729377125743e-06,
      "loss": 0.0307,
      "step": 47370
    },
    {
      "epoch": 2.404944413422001,
      "grad_norm": 0.4239245057106018,
      "learning_rate": 3.967037243853326e-06,
      "loss": 0.0326,
      "step": 47375
    },
    {
      "epoch": 2.4051982334128637,
      "grad_norm": 0.28727543354034424,
      "learning_rate": 3.96534511058091e-06,
      "loss": 0.034,
      "step": 47380
    },
    {
      "epoch": 2.4054520534037263,
      "grad_norm": 0.2851596176624298,
      "learning_rate": 3.9636529773084935e-06,
      "loss": 0.0341,
      "step": 47385
    },
    {
      "epoch": 2.4057058733945884,
      "grad_norm": 0.3607095777988434,
      "learning_rate": 3.961960844036077e-06,
      "loss": 0.0344,
      "step": 47390
    },
    {
      "epoch": 2.405959693385451,
      "grad_norm": 0.35408905148506165,
      "learning_rate": 3.96026871076366e-06,
      "loss": 0.0355,
      "step": 47395
    },
    {
      "epoch": 2.4062135133763136,
      "grad_norm": 0.23963472247123718,
      "learning_rate": 3.958576577491244e-06,
      "loss": 0.0281,
      "step": 47400
    },
    {
      "epoch": 2.4064673333671758,
      "grad_norm": 0.5606070756912231,
      "learning_rate": 3.956884444218827e-06,
      "loss": 0.0296,
      "step": 47405
    },
    {
      "epoch": 2.4067211533580384,
      "grad_norm": 0.25117459893226624,
      "learning_rate": 3.95519231094641e-06,
      "loss": 0.0283,
      "step": 47410
    },
    {
      "epoch": 2.406974973348901,
      "grad_norm": 0.2868806719779968,
      "learning_rate": 3.953500177673994e-06,
      "loss": 0.0323,
      "step": 47415
    },
    {
      "epoch": 2.4072287933397636,
      "grad_norm": 0.30753445625305176,
      "learning_rate": 3.9518080444015774e-06,
      "loss": 0.0343,
      "step": 47420
    },
    {
      "epoch": 2.4074826133306257,
      "grad_norm": 0.5894188284873962,
      "learning_rate": 3.950115911129161e-06,
      "loss": 0.0341,
      "step": 47425
    },
    {
      "epoch": 2.4077364333214883,
      "grad_norm": 0.35693567991256714,
      "learning_rate": 3.9484237778567445e-06,
      "loss": 0.0314,
      "step": 47430
    },
    {
      "epoch": 2.407990253312351,
      "grad_norm": 0.3354618549346924,
      "learning_rate": 3.946731644584328e-06,
      "loss": 0.0307,
      "step": 47435
    },
    {
      "epoch": 2.4082440733032136,
      "grad_norm": 0.276893675327301,
      "learning_rate": 3.945039511311911e-06,
      "loss": 0.0289,
      "step": 47440
    },
    {
      "epoch": 2.4084978932940757,
      "grad_norm": 0.29656293988227844,
      "learning_rate": 3.943347378039495e-06,
      "loss": 0.0303,
      "step": 47445
    },
    {
      "epoch": 2.4087517132849383,
      "grad_norm": 0.33579137921333313,
      "learning_rate": 3.941655244767078e-06,
      "loss": 0.0248,
      "step": 47450
    },
    {
      "epoch": 2.409005533275801,
      "grad_norm": 0.25143346190452576,
      "learning_rate": 3.939963111494661e-06,
      "loss": 0.0272,
      "step": 47455
    },
    {
      "epoch": 2.4092593532666635,
      "grad_norm": 0.3469618856906891,
      "learning_rate": 3.938270978222245e-06,
      "loss": 0.0239,
      "step": 47460
    },
    {
      "epoch": 2.4095131732575257,
      "grad_norm": 0.27013543248176575,
      "learning_rate": 3.9365788449498285e-06,
      "loss": 0.026,
      "step": 47465
    },
    {
      "epoch": 2.4097669932483883,
      "grad_norm": 0.28849682211875916,
      "learning_rate": 3.934886711677412e-06,
      "loss": 0.0268,
      "step": 47470
    },
    {
      "epoch": 2.410020813239251,
      "grad_norm": 0.21555496752262115,
      "learning_rate": 3.933194578404996e-06,
      "loss": 0.0282,
      "step": 47475
    },
    {
      "epoch": 2.410274633230113,
      "grad_norm": 0.2733899652957916,
      "learning_rate": 3.931502445132579e-06,
      "loss": 0.0375,
      "step": 47480
    },
    {
      "epoch": 2.4105284532209756,
      "grad_norm": 0.9940813183784485,
      "learning_rate": 3.929810311860162e-06,
      "loss": 0.0307,
      "step": 47485
    },
    {
      "epoch": 2.4107822732118382,
      "grad_norm": 0.3628968596458435,
      "learning_rate": 3.928118178587746e-06,
      "loss": 0.0269,
      "step": 47490
    },
    {
      "epoch": 2.411036093202701,
      "grad_norm": 0.2938300371170044,
      "learning_rate": 3.926426045315329e-06,
      "loss": 0.031,
      "step": 47495
    },
    {
      "epoch": 2.411289913193563,
      "grad_norm": 0.42575541138648987,
      "learning_rate": 3.9247339120429125e-06,
      "loss": 0.034,
      "step": 47500
    },
    {
      "epoch": 2.4115437331844256,
      "grad_norm": 0.2942411005496979,
      "learning_rate": 3.923041778770496e-06,
      "loss": 0.0319,
      "step": 47505
    },
    {
      "epoch": 2.411797553175288,
      "grad_norm": 0.533082127571106,
      "learning_rate": 3.9213496454980795e-06,
      "loss": 0.026,
      "step": 47510
    },
    {
      "epoch": 2.4120513731661504,
      "grad_norm": 0.4928203523159027,
      "learning_rate": 3.919657512225663e-06,
      "loss": 0.0393,
      "step": 47515
    },
    {
      "epoch": 2.412305193157013,
      "grad_norm": 0.28853780031204224,
      "learning_rate": 3.917965378953247e-06,
      "loss": 0.0328,
      "step": 47520
    },
    {
      "epoch": 2.4125590131478756,
      "grad_norm": 0.3255179524421692,
      "learning_rate": 3.91627324568083e-06,
      "loss": 0.0334,
      "step": 47525
    },
    {
      "epoch": 2.412812833138738,
      "grad_norm": 0.2753036618232727,
      "learning_rate": 3.914581112408414e-06,
      "loss": 0.0364,
      "step": 47530
    },
    {
      "epoch": 2.4130666531296003,
      "grad_norm": 0.43244901299476624,
      "learning_rate": 3.912888979135997e-06,
      "loss": 0.026,
      "step": 47535
    },
    {
      "epoch": 2.413320473120463,
      "grad_norm": 0.2927841246128082,
      "learning_rate": 3.911196845863581e-06,
      "loss": 0.0327,
      "step": 47540
    },
    {
      "epoch": 2.4135742931113255,
      "grad_norm": 0.4211539626121521,
      "learning_rate": 3.9095047125911635e-06,
      "loss": 0.0251,
      "step": 47545
    },
    {
      "epoch": 2.4138281131021877,
      "grad_norm": 0.28814446926116943,
      "learning_rate": 3.907812579318748e-06,
      "loss": 0.0291,
      "step": 47550
    },
    {
      "epoch": 2.4140819330930503,
      "grad_norm": 0.42922118306159973,
      "learning_rate": 3.906120446046331e-06,
      "loss": 0.0336,
      "step": 47555
    },
    {
      "epoch": 2.414335753083913,
      "grad_norm": 0.33254024386405945,
      "learning_rate": 3.904428312773915e-06,
      "loss": 0.0341,
      "step": 47560
    },
    {
      "epoch": 2.4145895730747755,
      "grad_norm": 0.40885141491889954,
      "learning_rate": 3.902736179501498e-06,
      "loss": 0.0278,
      "step": 47565
    },
    {
      "epoch": 2.4148433930656377,
      "grad_norm": 0.3493020534515381,
      "learning_rate": 3.901044046229081e-06,
      "loss": 0.0309,
      "step": 47570
    },
    {
      "epoch": 2.4150972130565003,
      "grad_norm": 0.2720143496990204,
      "learning_rate": 3.899351912956665e-06,
      "loss": 0.0309,
      "step": 47575
    },
    {
      "epoch": 2.415351033047363,
      "grad_norm": 0.44273534417152405,
      "learning_rate": 3.897659779684248e-06,
      "loss": 0.0346,
      "step": 47580
    },
    {
      "epoch": 2.4156048530382255,
      "grad_norm": 0.3117941617965698,
      "learning_rate": 3.895967646411832e-06,
      "loss": 0.0265,
      "step": 47585
    },
    {
      "epoch": 2.4158586730290876,
      "grad_norm": 0.29840338230133057,
      "learning_rate": 3.8942755131394146e-06,
      "loss": 0.031,
      "step": 47590
    },
    {
      "epoch": 2.4161124930199502,
      "grad_norm": 0.33411332964897156,
      "learning_rate": 3.892583379866999e-06,
      "loss": 0.025,
      "step": 47595
    },
    {
      "epoch": 2.416366313010813,
      "grad_norm": 0.4123666286468506,
      "learning_rate": 3.890891246594582e-06,
      "loss": 0.038,
      "step": 47600
    },
    {
      "epoch": 2.4166201330016754,
      "grad_norm": 0.3992561101913452,
      "learning_rate": 3.889199113322166e-06,
      "loss": 0.033,
      "step": 47605
    },
    {
      "epoch": 2.4168739529925376,
      "grad_norm": 0.26997870206832886,
      "learning_rate": 3.887506980049749e-06,
      "loss": 0.033,
      "step": 47610
    },
    {
      "epoch": 2.4171277729834,
      "grad_norm": 0.24809785187244415,
      "learning_rate": 3.885814846777332e-06,
      "loss": 0.0288,
      "step": 47615
    },
    {
      "epoch": 2.417381592974263,
      "grad_norm": 0.27336522936820984,
      "learning_rate": 3.884122713504916e-06,
      "loss": 0.0205,
      "step": 47620
    },
    {
      "epoch": 2.417635412965125,
      "grad_norm": 0.2975972890853882,
      "learning_rate": 3.882430580232499e-06,
      "loss": 0.027,
      "step": 47625
    },
    {
      "epoch": 2.4178892329559876,
      "grad_norm": 0.23915809392929077,
      "learning_rate": 3.880738446960083e-06,
      "loss": 0.036,
      "step": 47630
    },
    {
      "epoch": 2.41814305294685,
      "grad_norm": 0.4225624203681946,
      "learning_rate": 3.8790463136876665e-06,
      "loss": 0.033,
      "step": 47635
    },
    {
      "epoch": 2.4183968729377128,
      "grad_norm": 0.2970333397388458,
      "learning_rate": 3.87735418041525e-06,
      "loss": 0.0337,
      "step": 47640
    },
    {
      "epoch": 2.418650692928575,
      "grad_norm": 0.2330479919910431,
      "learning_rate": 3.8756620471428336e-06,
      "loss": 0.0286,
      "step": 47645
    },
    {
      "epoch": 2.4189045129194375,
      "grad_norm": 0.23988056182861328,
      "learning_rate": 3.873969913870417e-06,
      "loss": 0.0239,
      "step": 47650
    },
    {
      "epoch": 2.4191583329103,
      "grad_norm": 0.2401759773492813,
      "learning_rate": 3.872277780598e-06,
      "loss": 0.0339,
      "step": 47655
    },
    {
      "epoch": 2.4194121529011623,
      "grad_norm": 0.2513987720012665,
      "learning_rate": 3.870585647325583e-06,
      "loss": 0.0331,
      "step": 47660
    },
    {
      "epoch": 2.419665972892025,
      "grad_norm": 0.28474968671798706,
      "learning_rate": 3.868893514053167e-06,
      "loss": 0.0278,
      "step": 47665
    },
    {
      "epoch": 2.4199197928828875,
      "grad_norm": 0.35355323553085327,
      "learning_rate": 3.8672013807807504e-06,
      "loss": 0.0262,
      "step": 47670
    },
    {
      "epoch": 2.42017361287375,
      "grad_norm": 0.43043777346611023,
      "learning_rate": 3.865509247508334e-06,
      "loss": 0.0339,
      "step": 47675
    },
    {
      "epoch": 2.4204274328646123,
      "grad_norm": 0.4528213143348694,
      "learning_rate": 3.8638171142359175e-06,
      "loss": 0.0302,
      "step": 47680
    },
    {
      "epoch": 2.420681252855475,
      "grad_norm": 0.34391161799430847,
      "learning_rate": 3.862124980963501e-06,
      "loss": 0.028,
      "step": 47685
    },
    {
      "epoch": 2.4209350728463375,
      "grad_norm": 0.3792388141155243,
      "learning_rate": 3.860432847691085e-06,
      "loss": 0.0269,
      "step": 47690
    },
    {
      "epoch": 2.4211888928371996,
      "grad_norm": 0.3140493631362915,
      "learning_rate": 3.858740714418668e-06,
      "loss": 0.0372,
      "step": 47695
    },
    {
      "epoch": 2.4214427128280622,
      "grad_norm": 0.28535985946655273,
      "learning_rate": 3.857048581146252e-06,
      "loss": 0.0346,
      "step": 47700
    },
    {
      "epoch": 2.421696532818925,
      "grad_norm": 0.21206805109977722,
      "learning_rate": 3.855356447873834e-06,
      "loss": 0.0302,
      "step": 47705
    },
    {
      "epoch": 2.4219503528097874,
      "grad_norm": 0.252411812543869,
      "learning_rate": 3.853664314601419e-06,
      "loss": 0.0266,
      "step": 47710
    },
    {
      "epoch": 2.4222041728006496,
      "grad_norm": 0.4648708701133728,
      "learning_rate": 3.8519721813290015e-06,
      "loss": 0.0249,
      "step": 47715
    },
    {
      "epoch": 2.422457992791512,
      "grad_norm": 0.27817830443382263,
      "learning_rate": 3.850280048056586e-06,
      "loss": 0.0226,
      "step": 47720
    },
    {
      "epoch": 2.422711812782375,
      "grad_norm": 0.2477516084909439,
      "learning_rate": 3.8485879147841686e-06,
      "loss": 0.0267,
      "step": 47725
    },
    {
      "epoch": 2.4229656327732374,
      "grad_norm": 0.3148895502090454,
      "learning_rate": 3.846895781511752e-06,
      "loss": 0.0294,
      "step": 47730
    },
    {
      "epoch": 2.4232194527640996,
      "grad_norm": 0.6183810830116272,
      "learning_rate": 3.845203648239336e-06,
      "loss": 0.0287,
      "step": 47735
    },
    {
      "epoch": 2.423473272754962,
      "grad_norm": 0.20661011338233948,
      "learning_rate": 3.843511514966919e-06,
      "loss": 0.0287,
      "step": 47740
    },
    {
      "epoch": 2.4237270927458248,
      "grad_norm": 0.5037245750427246,
      "learning_rate": 3.841819381694503e-06,
      "loss": 0.0366,
      "step": 47745
    },
    {
      "epoch": 2.4239809127366874,
      "grad_norm": 0.5295654535293579,
      "learning_rate": 3.8401272484220855e-06,
      "loss": 0.0314,
      "step": 47750
    },
    {
      "epoch": 2.4242347327275495,
      "grad_norm": 0.22970163822174072,
      "learning_rate": 3.83843511514967e-06,
      "loss": 0.0298,
      "step": 47755
    },
    {
      "epoch": 2.424488552718412,
      "grad_norm": 0.3270503580570221,
      "learning_rate": 3.8367429818772525e-06,
      "loss": 0.0317,
      "step": 47760
    },
    {
      "epoch": 2.4247423727092747,
      "grad_norm": 0.3448561131954193,
      "learning_rate": 3.835050848604837e-06,
      "loss": 0.0388,
      "step": 47765
    },
    {
      "epoch": 2.424996192700137,
      "grad_norm": 0.23250630497932434,
      "learning_rate": 3.83335871533242e-06,
      "loss": 0.0291,
      "step": 47770
    },
    {
      "epoch": 2.4252500126909995,
      "grad_norm": 0.37792283296585083,
      "learning_rate": 3.831666582060003e-06,
      "loss": 0.0357,
      "step": 47775
    },
    {
      "epoch": 2.425503832681862,
      "grad_norm": 0.44455191493034363,
      "learning_rate": 3.829974448787587e-06,
      "loss": 0.0317,
      "step": 47780
    },
    {
      "epoch": 2.4257576526727247,
      "grad_norm": 0.2922888696193695,
      "learning_rate": 3.82828231551517e-06,
      "loss": 0.03,
      "step": 47785
    },
    {
      "epoch": 2.426011472663587,
      "grad_norm": 0.33817926049232483,
      "learning_rate": 3.826590182242754e-06,
      "loss": 0.0456,
      "step": 47790
    },
    {
      "epoch": 2.4262652926544495,
      "grad_norm": 0.46196913719177246,
      "learning_rate": 3.824898048970337e-06,
      "loss": 0.0399,
      "step": 47795
    },
    {
      "epoch": 2.426519112645312,
      "grad_norm": 0.27602121233940125,
      "learning_rate": 3.823205915697921e-06,
      "loss": 0.0359,
      "step": 47800
    },
    {
      "epoch": 2.426772932636174,
      "grad_norm": 0.2666013538837433,
      "learning_rate": 3.8215137824255044e-06,
      "loss": 0.0256,
      "step": 47805
    },
    {
      "epoch": 2.427026752627037,
      "grad_norm": 0.22680050134658813,
      "learning_rate": 3.819821649153088e-06,
      "loss": 0.0296,
      "step": 47810
    },
    {
      "epoch": 2.4272805726178994,
      "grad_norm": 0.3893703818321228,
      "learning_rate": 3.818129515880671e-06,
      "loss": 0.0377,
      "step": 47815
    },
    {
      "epoch": 2.427534392608762,
      "grad_norm": 0.31379884481430054,
      "learning_rate": 3.816437382608254e-06,
      "loss": 0.0307,
      "step": 47820
    },
    {
      "epoch": 2.427788212599624,
      "grad_norm": 0.32073163986206055,
      "learning_rate": 3.814745249335838e-06,
      "loss": 0.0313,
      "step": 47825
    },
    {
      "epoch": 2.428042032590487,
      "grad_norm": 0.345052570104599,
      "learning_rate": 3.8130531160634217e-06,
      "loss": 0.0328,
      "step": 47830
    },
    {
      "epoch": 2.4282958525813494,
      "grad_norm": 0.3013349175453186,
      "learning_rate": 3.811360982791005e-06,
      "loss": 0.0258,
      "step": 47835
    },
    {
      "epoch": 2.4285496725722115,
      "grad_norm": 0.4422132074832916,
      "learning_rate": 3.8096688495185884e-06,
      "loss": 0.0376,
      "step": 47840
    },
    {
      "epoch": 2.428803492563074,
      "grad_norm": 0.3477928340435028,
      "learning_rate": 3.8079767162461715e-06,
      "loss": 0.027,
      "step": 47845
    },
    {
      "epoch": 2.4290573125539368,
      "grad_norm": 0.4966389536857605,
      "learning_rate": 3.8062845829737555e-06,
      "loss": 0.0316,
      "step": 47850
    },
    {
      "epoch": 2.4293111325447994,
      "grad_norm": 0.24931129813194275,
      "learning_rate": 3.8045924497013386e-06,
      "loss": 0.0271,
      "step": 47855
    },
    {
      "epoch": 2.4295649525356615,
      "grad_norm": 0.3480713367462158,
      "learning_rate": 3.8029003164289226e-06,
      "loss": 0.0315,
      "step": 47860
    },
    {
      "epoch": 2.429818772526524,
      "grad_norm": 0.20172403752803802,
      "learning_rate": 3.8012081831565057e-06,
      "loss": 0.029,
      "step": 47865
    },
    {
      "epoch": 2.4300725925173867,
      "grad_norm": 0.22433215379714966,
      "learning_rate": 3.7995160498840892e-06,
      "loss": 0.0294,
      "step": 47870
    },
    {
      "epoch": 2.4303264125082493,
      "grad_norm": 0.42179742455482483,
      "learning_rate": 3.797823916611673e-06,
      "loss": 0.0302,
      "step": 47875
    },
    {
      "epoch": 2.4305802324991115,
      "grad_norm": 0.345768541097641,
      "learning_rate": 3.7961317833392563e-06,
      "loss": 0.0395,
      "step": 47880
    },
    {
      "epoch": 2.430834052489974,
      "grad_norm": 0.30003654956817627,
      "learning_rate": 3.7944396500668395e-06,
      "loss": 0.0304,
      "step": 47885
    },
    {
      "epoch": 2.4310878724808367,
      "grad_norm": 0.7992334365844727,
      "learning_rate": 3.7927475167944226e-06,
      "loss": 0.0282,
      "step": 47890
    },
    {
      "epoch": 2.4313416924716993,
      "grad_norm": 1.201789140701294,
      "learning_rate": 3.7910553835220065e-06,
      "loss": 0.0388,
      "step": 47895
    },
    {
      "epoch": 2.4315955124625614,
      "grad_norm": 0.2232540100812912,
      "learning_rate": 3.7893632502495897e-06,
      "loss": 0.0312,
      "step": 47900
    },
    {
      "epoch": 2.431849332453424,
      "grad_norm": 0.39150550961494446,
      "learning_rate": 3.7876711169771736e-06,
      "loss": 0.0396,
      "step": 47905
    },
    {
      "epoch": 2.4321031524442867,
      "grad_norm": 0.25751233100891113,
      "learning_rate": 3.7859789837047568e-06,
      "loss": 0.0313,
      "step": 47910
    },
    {
      "epoch": 2.432356972435149,
      "grad_norm": 0.38640815019607544,
      "learning_rate": 3.7842868504323403e-06,
      "loss": 0.0364,
      "step": 47915
    },
    {
      "epoch": 2.4326107924260114,
      "grad_norm": 0.30392131209373474,
      "learning_rate": 3.782594717159924e-06,
      "loss": 0.0295,
      "step": 47920
    },
    {
      "epoch": 2.432864612416874,
      "grad_norm": 0.36660847067832947,
      "learning_rate": 3.7809025838875074e-06,
      "loss": 0.0348,
      "step": 47925
    },
    {
      "epoch": 2.4331184324077366,
      "grad_norm": 0.31865790486335754,
      "learning_rate": 3.7792104506150905e-06,
      "loss": 0.0343,
      "step": 47930
    },
    {
      "epoch": 2.433372252398599,
      "grad_norm": 0.44969046115875244,
      "learning_rate": 3.7775183173426745e-06,
      "loss": 0.0335,
      "step": 47935
    },
    {
      "epoch": 2.4336260723894614,
      "grad_norm": 0.30257928371429443,
      "learning_rate": 3.7758261840702576e-06,
      "loss": 0.0323,
      "step": 47940
    },
    {
      "epoch": 2.433879892380324,
      "grad_norm": 0.3042431175708771,
      "learning_rate": 3.774134050797841e-06,
      "loss": 0.0323,
      "step": 47945
    },
    {
      "epoch": 2.434133712371186,
      "grad_norm": 0.302280992269516,
      "learning_rate": 3.7724419175254247e-06,
      "loss": 0.0312,
      "step": 47950
    },
    {
      "epoch": 2.4343875323620487,
      "grad_norm": 0.2679111063480377,
      "learning_rate": 3.7707497842530082e-06,
      "loss": 0.0268,
      "step": 47955
    },
    {
      "epoch": 2.4346413523529113,
      "grad_norm": 0.2837721109390259,
      "learning_rate": 3.7690576509805914e-06,
      "loss": 0.0316,
      "step": 47960
    },
    {
      "epoch": 2.434895172343774,
      "grad_norm": 0.3853626847267151,
      "learning_rate": 3.7673655177081753e-06,
      "loss": 0.0325,
      "step": 47965
    },
    {
      "epoch": 2.435148992334636,
      "grad_norm": 0.2845440208911896,
      "learning_rate": 3.7656733844357584e-06,
      "loss": 0.035,
      "step": 47970
    },
    {
      "epoch": 2.4354028123254987,
      "grad_norm": 0.34717926383018494,
      "learning_rate": 3.7639812511633416e-06,
      "loss": 0.0338,
      "step": 47975
    },
    {
      "epoch": 2.4356566323163613,
      "grad_norm": 0.1998908966779709,
      "learning_rate": 3.7622891178909255e-06,
      "loss": 0.0333,
      "step": 47980
    },
    {
      "epoch": 2.4359104523072235,
      "grad_norm": 0.19842973351478577,
      "learning_rate": 3.7605969846185087e-06,
      "loss": 0.0266,
      "step": 47985
    },
    {
      "epoch": 2.436164272298086,
      "grad_norm": 0.3331410586833954,
      "learning_rate": 3.758904851346092e-06,
      "loss": 0.0254,
      "step": 47990
    },
    {
      "epoch": 2.4364180922889487,
      "grad_norm": 0.33584150671958923,
      "learning_rate": 3.7572127180736757e-06,
      "loss": 0.0325,
      "step": 47995
    },
    {
      "epoch": 2.4366719122798113,
      "grad_norm": 0.3402474820613861,
      "learning_rate": 3.7555205848012593e-06,
      "loss": 0.0288,
      "step": 48000
    },
    {
      "epoch": 2.436925732270674,
      "grad_norm": 0.5381267666816711,
      "learning_rate": 3.7538284515288424e-06,
      "loss": 0.0346,
      "step": 48005
    },
    {
      "epoch": 2.437179552261536,
      "grad_norm": 0.447458416223526,
      "learning_rate": 3.7521363182564264e-06,
      "loss": 0.0301,
      "step": 48010
    },
    {
      "epoch": 2.4374333722523986,
      "grad_norm": 0.2832920253276825,
      "learning_rate": 3.7504441849840095e-06,
      "loss": 0.0349,
      "step": 48015
    },
    {
      "epoch": 2.4376871922432612,
      "grad_norm": 0.5114197731018066,
      "learning_rate": 3.7487520517115935e-06,
      "loss": 0.0337,
      "step": 48020
    },
    {
      "epoch": 2.4379410122341234,
      "grad_norm": 2.0609970092773438,
      "learning_rate": 3.7470599184391766e-06,
      "loss": 0.0272,
      "step": 48025
    },
    {
      "epoch": 2.438194832224986,
      "grad_norm": 0.38977518677711487,
      "learning_rate": 3.74536778516676e-06,
      "loss": 0.0323,
      "step": 48030
    },
    {
      "epoch": 2.4384486522158486,
      "grad_norm": 0.3010084331035614,
      "learning_rate": 3.7436756518943433e-06,
      "loss": 0.0305,
      "step": 48035
    },
    {
      "epoch": 2.438702472206711,
      "grad_norm": 0.2480255663394928,
      "learning_rate": 3.7419835186219272e-06,
      "loss": 0.0282,
      "step": 48040
    },
    {
      "epoch": 2.4389562921975734,
      "grad_norm": 0.18872040510177612,
      "learning_rate": 3.7402913853495103e-06,
      "loss": 0.0261,
      "step": 48045
    },
    {
      "epoch": 2.439210112188436,
      "grad_norm": 0.3619585931301117,
      "learning_rate": 3.7385992520770935e-06,
      "loss": 0.0286,
      "step": 48050
    },
    {
      "epoch": 2.4394639321792986,
      "grad_norm": 0.35356375575065613,
      "learning_rate": 3.7369071188046774e-06,
      "loss": 0.0305,
      "step": 48055
    },
    {
      "epoch": 2.4397177521701607,
      "grad_norm": 0.26403146982192993,
      "learning_rate": 3.7352149855322606e-06,
      "loss": 0.0273,
      "step": 48060
    },
    {
      "epoch": 2.4399715721610233,
      "grad_norm": 0.2659587860107422,
      "learning_rate": 3.7335228522598445e-06,
      "loss": 0.0304,
      "step": 48065
    },
    {
      "epoch": 2.440225392151886,
      "grad_norm": 0.32062849402427673,
      "learning_rate": 3.7318307189874276e-06,
      "loss": 0.0296,
      "step": 48070
    },
    {
      "epoch": 2.4404792121427485,
      "grad_norm": 0.35643866658210754,
      "learning_rate": 3.730138585715011e-06,
      "loss": 0.0291,
      "step": 48075
    },
    {
      "epoch": 2.4407330321336107,
      "grad_norm": 0.305879145860672,
      "learning_rate": 3.7284464524425943e-06,
      "loss": 0.0273,
      "step": 48080
    },
    {
      "epoch": 2.4409868521244733,
      "grad_norm": 0.275307297706604,
      "learning_rate": 3.7267543191701783e-06,
      "loss": 0.036,
      "step": 48085
    },
    {
      "epoch": 2.441240672115336,
      "grad_norm": 0.34820234775543213,
      "learning_rate": 3.7250621858977614e-06,
      "loss": 0.0297,
      "step": 48090
    },
    {
      "epoch": 2.441494492106198,
      "grad_norm": 0.25399303436279297,
      "learning_rate": 3.7233700526253454e-06,
      "loss": 0.0321,
      "step": 48095
    },
    {
      "epoch": 2.4417483120970607,
      "grad_norm": 0.27098965644836426,
      "learning_rate": 3.7216779193529285e-06,
      "loss": 0.0346,
      "step": 48100
    },
    {
      "epoch": 2.4420021320879233,
      "grad_norm": 0.3064155876636505,
      "learning_rate": 3.719985786080512e-06,
      "loss": 0.0281,
      "step": 48105
    },
    {
      "epoch": 2.442255952078786,
      "grad_norm": 0.3673038184642792,
      "learning_rate": 3.7182936528080956e-06,
      "loss": 0.0254,
      "step": 48110
    },
    {
      "epoch": 2.442509772069648,
      "grad_norm": 0.39501214027404785,
      "learning_rate": 3.716601519535679e-06,
      "loss": 0.039,
      "step": 48115
    },
    {
      "epoch": 2.4427635920605106,
      "grad_norm": 0.37488582730293274,
      "learning_rate": 3.7149093862632622e-06,
      "loss": 0.0285,
      "step": 48120
    },
    {
      "epoch": 2.4430174120513732,
      "grad_norm": 0.20910173654556274,
      "learning_rate": 3.713217252990846e-06,
      "loss": 0.0288,
      "step": 48125
    },
    {
      "epoch": 2.4432712320422354,
      "grad_norm": 0.32626721262931824,
      "learning_rate": 3.7115251197184293e-06,
      "loss": 0.0255,
      "step": 48130
    },
    {
      "epoch": 2.443525052033098,
      "grad_norm": 0.27159538865089417,
      "learning_rate": 3.7098329864460125e-06,
      "loss": 0.0283,
      "step": 48135
    },
    {
      "epoch": 2.4437788720239606,
      "grad_norm": 0.6607284545898438,
      "learning_rate": 3.7081408531735964e-06,
      "loss": 0.0311,
      "step": 48140
    },
    {
      "epoch": 2.444032692014823,
      "grad_norm": 0.23017600178718567,
      "learning_rate": 3.7064487199011795e-06,
      "loss": 0.037,
      "step": 48145
    },
    {
      "epoch": 2.444286512005686,
      "grad_norm": 0.3214363753795624,
      "learning_rate": 3.704756586628763e-06,
      "loss": 0.0278,
      "step": 48150
    },
    {
      "epoch": 2.444540331996548,
      "grad_norm": 0.3399595618247986,
      "learning_rate": 3.7030644533563466e-06,
      "loss": 0.0339,
      "step": 48155
    },
    {
      "epoch": 2.4447941519874106,
      "grad_norm": 0.277424693107605,
      "learning_rate": 3.70137232008393e-06,
      "loss": 0.0312,
      "step": 48160
    },
    {
      "epoch": 2.445047971978273,
      "grad_norm": 0.3377756178379059,
      "learning_rate": 3.6996801868115133e-06,
      "loss": 0.0319,
      "step": 48165
    },
    {
      "epoch": 2.4453017919691353,
      "grad_norm": 1.510842204093933,
      "learning_rate": 3.6979880535390973e-06,
      "loss": 0.0298,
      "step": 48170
    },
    {
      "epoch": 2.445555611959998,
      "grad_norm": 0.2741282880306244,
      "learning_rate": 3.6962959202666804e-06,
      "loss": 0.029,
      "step": 48175
    },
    {
      "epoch": 2.4458094319508605,
      "grad_norm": 0.3779626786708832,
      "learning_rate": 3.6946037869942644e-06,
      "loss": 0.0246,
      "step": 48180
    },
    {
      "epoch": 2.446063251941723,
      "grad_norm": 0.44682037830352783,
      "learning_rate": 3.6929116537218475e-06,
      "loss": 0.0309,
      "step": 48185
    },
    {
      "epoch": 2.4463170719325853,
      "grad_norm": 0.2568581700325012,
      "learning_rate": 3.691219520449431e-06,
      "loss": 0.0268,
      "step": 48190
    },
    {
      "epoch": 2.446570891923448,
      "grad_norm": 0.2952916920185089,
      "learning_rate": 3.689527387177014e-06,
      "loss": 0.028,
      "step": 48195
    },
    {
      "epoch": 2.4468247119143105,
      "grad_norm": 0.49456146359443665,
      "learning_rate": 3.687835253904598e-06,
      "loss": 0.0331,
      "step": 48200
    },
    {
      "epoch": 2.4470785319051727,
      "grad_norm": 0.2728605568408966,
      "learning_rate": 3.6861431206321812e-06,
      "loss": 0.0365,
      "step": 48205
    },
    {
      "epoch": 2.4473323518960353,
      "grad_norm": 0.37636256217956543,
      "learning_rate": 3.684450987359765e-06,
      "loss": 0.0397,
      "step": 48210
    },
    {
      "epoch": 2.447586171886898,
      "grad_norm": 0.241055428981781,
      "learning_rate": 3.6827588540873483e-06,
      "loss": 0.0403,
      "step": 48215
    },
    {
      "epoch": 2.4478399918777605,
      "grad_norm": 0.2556413412094116,
      "learning_rate": 3.6810667208149314e-06,
      "loss": 0.0308,
      "step": 48220
    },
    {
      "epoch": 2.4480938118686226,
      "grad_norm": 0.3312203288078308,
      "learning_rate": 3.6793745875425154e-06,
      "loss": 0.027,
      "step": 48225
    },
    {
      "epoch": 2.4483476318594852,
      "grad_norm": 0.34544646739959717,
      "learning_rate": 3.6776824542700985e-06,
      "loss": 0.0294,
      "step": 48230
    },
    {
      "epoch": 2.448601451850348,
      "grad_norm": 0.3323667347431183,
      "learning_rate": 3.675990320997682e-06,
      "loss": 0.0343,
      "step": 48235
    },
    {
      "epoch": 2.44885527184121,
      "grad_norm": 0.2082269936800003,
      "learning_rate": 3.674298187725265e-06,
      "loss": 0.0314,
      "step": 48240
    },
    {
      "epoch": 2.4491090918320726,
      "grad_norm": 0.2525387108325958,
      "learning_rate": 3.672606054452849e-06,
      "loss": 0.0331,
      "step": 48245
    },
    {
      "epoch": 2.449362911822935,
      "grad_norm": 0.34921693801879883,
      "learning_rate": 3.6709139211804323e-06,
      "loss": 0.0367,
      "step": 48250
    },
    {
      "epoch": 2.449616731813798,
      "grad_norm": 0.4713902175426483,
      "learning_rate": 3.6692217879080162e-06,
      "loss": 0.0332,
      "step": 48255
    },
    {
      "epoch": 2.44987055180466,
      "grad_norm": 0.5320238471031189,
      "learning_rate": 3.6675296546355994e-06,
      "loss": 0.0312,
      "step": 48260
    },
    {
      "epoch": 2.4501243717955226,
      "grad_norm": 0.3506769835948944,
      "learning_rate": 3.665837521363183e-06,
      "loss": 0.0305,
      "step": 48265
    },
    {
      "epoch": 2.450378191786385,
      "grad_norm": 0.2673940658569336,
      "learning_rate": 3.6641453880907665e-06,
      "loss": 0.0326,
      "step": 48270
    },
    {
      "epoch": 2.4506320117772478,
      "grad_norm": 0.259799599647522,
      "learning_rate": 3.66245325481835e-06,
      "loss": 0.0259,
      "step": 48275
    },
    {
      "epoch": 2.45088583176811,
      "grad_norm": 0.2402365803718567,
      "learning_rate": 3.660761121545933e-06,
      "loss": 0.0234,
      "step": 48280
    },
    {
      "epoch": 2.4511396517589725,
      "grad_norm": 0.28318873047828674,
      "learning_rate": 3.659068988273517e-06,
      "loss": 0.0363,
      "step": 48285
    },
    {
      "epoch": 2.451393471749835,
      "grad_norm": 0.218606099486351,
      "learning_rate": 3.6573768550011002e-06,
      "loss": 0.0287,
      "step": 48290
    },
    {
      "epoch": 2.4516472917406977,
      "grad_norm": 0.35069960355758667,
      "learning_rate": 3.6556847217286833e-06,
      "loss": 0.0276,
      "step": 48295
    },
    {
      "epoch": 2.45190111173156,
      "grad_norm": 0.2726610600948334,
      "learning_rate": 3.6539925884562673e-06,
      "loss": 0.0318,
      "step": 48300
    },
    {
      "epoch": 2.4521549317224225,
      "grad_norm": 0.4079586863517761,
      "learning_rate": 3.6523004551838504e-06,
      "loss": 0.0314,
      "step": 48305
    },
    {
      "epoch": 2.452408751713285,
      "grad_norm": 0.2880258858203888,
      "learning_rate": 3.650608321911434e-06,
      "loss": 0.0391,
      "step": 48310
    },
    {
      "epoch": 2.4526625717041473,
      "grad_norm": 0.3445485234260559,
      "learning_rate": 3.6489161886390175e-06,
      "loss": 0.0327,
      "step": 48315
    },
    {
      "epoch": 2.45291639169501,
      "grad_norm": 0.23641039431095123,
      "learning_rate": 3.647224055366601e-06,
      "loss": 0.0269,
      "step": 48320
    },
    {
      "epoch": 2.4531702116858725,
      "grad_norm": 0.3212410509586334,
      "learning_rate": 3.645531922094184e-06,
      "loss": 0.0325,
      "step": 48325
    },
    {
      "epoch": 2.453424031676735,
      "grad_norm": 0.2834191918373108,
      "learning_rate": 3.643839788821768e-06,
      "loss": 0.0288,
      "step": 48330
    },
    {
      "epoch": 2.453677851667597,
      "grad_norm": 0.26539257168769836,
      "learning_rate": 3.6421476555493513e-06,
      "loss": 0.0338,
      "step": 48335
    },
    {
      "epoch": 2.45393167165846,
      "grad_norm": 0.4019145369529724,
      "learning_rate": 3.640455522276935e-06,
      "loss": 0.0309,
      "step": 48340
    },
    {
      "epoch": 2.4541854916493224,
      "grad_norm": 0.6921430230140686,
      "learning_rate": 3.6387633890045184e-06,
      "loss": 0.0302,
      "step": 48345
    },
    {
      "epoch": 2.4544393116401846,
      "grad_norm": 0.3001328408718109,
      "learning_rate": 3.637071255732102e-06,
      "loss": 0.0314,
      "step": 48350
    },
    {
      "epoch": 2.454693131631047,
      "grad_norm": 0.34600576758384705,
      "learning_rate": 3.635379122459685e-06,
      "loss": 0.0354,
      "step": 48355
    },
    {
      "epoch": 2.45494695162191,
      "grad_norm": 0.3184991478919983,
      "learning_rate": 3.633686989187269e-06,
      "loss": 0.0294,
      "step": 48360
    },
    {
      "epoch": 2.4552007716127724,
      "grad_norm": 0.2975368797779083,
      "learning_rate": 3.631994855914852e-06,
      "loss": 0.0234,
      "step": 48365
    },
    {
      "epoch": 2.4554545916036346,
      "grad_norm": 0.40252795815467834,
      "learning_rate": 3.630302722642436e-06,
      "loss": 0.0308,
      "step": 48370
    },
    {
      "epoch": 2.455708411594497,
      "grad_norm": 0.2199138104915619,
      "learning_rate": 3.628610589370019e-06,
      "loss": 0.0296,
      "step": 48375
    },
    {
      "epoch": 2.4559622315853598,
      "grad_norm": 0.2915712296962738,
      "learning_rate": 3.6269184560976023e-06,
      "loss": 0.0249,
      "step": 48380
    },
    {
      "epoch": 2.456216051576222,
      "grad_norm": 0.4096558690071106,
      "learning_rate": 3.625226322825186e-06,
      "loss": 0.0281,
      "step": 48385
    },
    {
      "epoch": 2.4564698715670845,
      "grad_norm": 0.22814741730690002,
      "learning_rate": 3.6235341895527694e-06,
      "loss": 0.0354,
      "step": 48390
    },
    {
      "epoch": 2.456723691557947,
      "grad_norm": 0.2444760948419571,
      "learning_rate": 3.621842056280353e-06,
      "loss": 0.0294,
      "step": 48395
    },
    {
      "epoch": 2.4569775115488097,
      "grad_norm": 0.35308727622032166,
      "learning_rate": 3.620149923007936e-06,
      "loss": 0.0331,
      "step": 48400
    },
    {
      "epoch": 2.457231331539672,
      "grad_norm": 0.3179206848144531,
      "learning_rate": 3.61845778973552e-06,
      "loss": 0.0326,
      "step": 48405
    },
    {
      "epoch": 2.4574851515305345,
      "grad_norm": 0.2587548792362213,
      "learning_rate": 3.616765656463103e-06,
      "loss": 0.036,
      "step": 48410
    },
    {
      "epoch": 2.457738971521397,
      "grad_norm": 0.44887417554855347,
      "learning_rate": 3.615073523190687e-06,
      "loss": 0.0327,
      "step": 48415
    },
    {
      "epoch": 2.4579927915122597,
      "grad_norm": 0.6582462191581726,
      "learning_rate": 3.6133813899182703e-06,
      "loss": 0.0416,
      "step": 48420
    },
    {
      "epoch": 2.458246611503122,
      "grad_norm": 0.3380301892757416,
      "learning_rate": 3.611689256645854e-06,
      "loss": 0.0378,
      "step": 48425
    },
    {
      "epoch": 2.4585004314939845,
      "grad_norm": 0.3325098156929016,
      "learning_rate": 3.609997123373437e-06,
      "loss": 0.0411,
      "step": 48430
    },
    {
      "epoch": 2.458754251484847,
      "grad_norm": 0.4534139633178711,
      "learning_rate": 3.608304990101021e-06,
      "loss": 0.0336,
      "step": 48435
    },
    {
      "epoch": 2.4590080714757097,
      "grad_norm": 0.3075084090232849,
      "learning_rate": 3.606612856828604e-06,
      "loss": 0.0394,
      "step": 48440
    },
    {
      "epoch": 2.459261891466572,
      "grad_norm": 0.2668818533420563,
      "learning_rate": 3.604920723556188e-06,
      "loss": 0.0348,
      "step": 48445
    },
    {
      "epoch": 2.4595157114574344,
      "grad_norm": 0.2585658133029938,
      "learning_rate": 3.603228590283771e-06,
      "loss": 0.0367,
      "step": 48450
    },
    {
      "epoch": 2.459769531448297,
      "grad_norm": 0.246776282787323,
      "learning_rate": 3.6015364570113542e-06,
      "loss": 0.0284,
      "step": 48455
    },
    {
      "epoch": 2.460023351439159,
      "grad_norm": 0.36037343740463257,
      "learning_rate": 3.599844323738938e-06,
      "loss": 0.0256,
      "step": 48460
    },
    {
      "epoch": 2.460277171430022,
      "grad_norm": 0.5250141620635986,
      "learning_rate": 3.5981521904665213e-06,
      "loss": 0.035,
      "step": 48465
    },
    {
      "epoch": 2.4605309914208844,
      "grad_norm": 0.340463787317276,
      "learning_rate": 3.596460057194105e-06,
      "loss": 0.0289,
      "step": 48470
    },
    {
      "epoch": 2.460784811411747,
      "grad_norm": 0.4077812433242798,
      "learning_rate": 3.594767923921688e-06,
      "loss": 0.0278,
      "step": 48475
    },
    {
      "epoch": 2.461038631402609,
      "grad_norm": 0.32608455419540405,
      "learning_rate": 3.593075790649272e-06,
      "loss": 0.0344,
      "step": 48480
    },
    {
      "epoch": 2.4612924513934717,
      "grad_norm": 0.28923162817955017,
      "learning_rate": 3.591383657376855e-06,
      "loss": 0.0244,
      "step": 48485
    },
    {
      "epoch": 2.4615462713843344,
      "grad_norm": 0.5158969163894653,
      "learning_rate": 3.589691524104439e-06,
      "loss": 0.034,
      "step": 48490
    },
    {
      "epoch": 2.4618000913751965,
      "grad_norm": 0.24424219131469727,
      "learning_rate": 3.587999390832022e-06,
      "loss": 0.0281,
      "step": 48495
    },
    {
      "epoch": 2.462053911366059,
      "grad_norm": 0.27959030866622925,
      "learning_rate": 3.5863072575596057e-06,
      "loss": 0.0379,
      "step": 48500
    },
    {
      "epoch": 2.4623077313569217,
      "grad_norm": 0.307492196559906,
      "learning_rate": 3.5846151242871892e-06,
      "loss": 0.0323,
      "step": 48505
    },
    {
      "epoch": 2.4625615513477843,
      "grad_norm": 0.27596306800842285,
      "learning_rate": 3.5829229910147728e-06,
      "loss": 0.0291,
      "step": 48510
    },
    {
      "epoch": 2.4628153713386465,
      "grad_norm": 0.2955135107040405,
      "learning_rate": 3.581230857742356e-06,
      "loss": 0.0258,
      "step": 48515
    },
    {
      "epoch": 2.463069191329509,
      "grad_norm": 0.30613574385643005,
      "learning_rate": 3.57953872446994e-06,
      "loss": 0.0323,
      "step": 48520
    },
    {
      "epoch": 2.4633230113203717,
      "grad_norm": 0.30772191286087036,
      "learning_rate": 3.577846591197523e-06,
      "loss": 0.0337,
      "step": 48525
    },
    {
      "epoch": 2.463576831311234,
      "grad_norm": 0.32587411999702454,
      "learning_rate": 3.5761544579251065e-06,
      "loss": 0.0313,
      "step": 48530
    },
    {
      "epoch": 2.4638306513020964,
      "grad_norm": 0.5187804698944092,
      "learning_rate": 3.57446232465269e-06,
      "loss": 0.0365,
      "step": 48535
    },
    {
      "epoch": 2.464084471292959,
      "grad_norm": 0.3063710629940033,
      "learning_rate": 3.572770191380273e-06,
      "loss": 0.0365,
      "step": 48540
    },
    {
      "epoch": 2.4643382912838216,
      "grad_norm": 0.44692617654800415,
      "learning_rate": 3.5710780581078568e-06,
      "loss": 0.0304,
      "step": 48545
    },
    {
      "epoch": 2.464592111274684,
      "grad_norm": 0.2769562304019928,
      "learning_rate": 3.5693859248354403e-06,
      "loss": 0.0299,
      "step": 48550
    },
    {
      "epoch": 2.4648459312655464,
      "grad_norm": 0.3721355199813843,
      "learning_rate": 3.567693791563024e-06,
      "loss": 0.0328,
      "step": 48555
    },
    {
      "epoch": 2.465099751256409,
      "grad_norm": 0.36836931109428406,
      "learning_rate": 3.566001658290607e-06,
      "loss": 0.031,
      "step": 48560
    },
    {
      "epoch": 2.4653535712472716,
      "grad_norm": 0.3551080822944641,
      "learning_rate": 3.564309525018191e-06,
      "loss": 0.0316,
      "step": 48565
    },
    {
      "epoch": 2.4656073912381338,
      "grad_norm": 0.3343908488750458,
      "learning_rate": 3.562617391745774e-06,
      "loss": 0.0345,
      "step": 48570
    },
    {
      "epoch": 2.4658612112289964,
      "grad_norm": 0.3180460035800934,
      "learning_rate": 3.5609252584733576e-06,
      "loss": 0.0383,
      "step": 48575
    },
    {
      "epoch": 2.466115031219859,
      "grad_norm": 0.2818259298801422,
      "learning_rate": 3.559233125200941e-06,
      "loss": 0.0331,
      "step": 48580
    },
    {
      "epoch": 2.4663688512107216,
      "grad_norm": 0.1655755341053009,
      "learning_rate": 3.5575409919285247e-06,
      "loss": 0.0304,
      "step": 48585
    },
    {
      "epoch": 2.4666226712015837,
      "grad_norm": 0.7998566627502441,
      "learning_rate": 3.555848858656108e-06,
      "loss": 0.0267,
      "step": 48590
    },
    {
      "epoch": 2.4668764911924463,
      "grad_norm": 0.4926971197128296,
      "learning_rate": 3.5541567253836918e-06,
      "loss": 0.0332,
      "step": 48595
    },
    {
      "epoch": 2.467130311183309,
      "grad_norm": 0.2525290548801422,
      "learning_rate": 3.552464592111275e-06,
      "loss": 0.0444,
      "step": 48600
    },
    {
      "epoch": 2.467384131174171,
      "grad_norm": 0.3362772464752197,
      "learning_rate": 3.550772458838859e-06,
      "loss": 0.0286,
      "step": 48605
    },
    {
      "epoch": 2.4676379511650337,
      "grad_norm": 0.2079012095928192,
      "learning_rate": 3.549080325566442e-06,
      "loss": 0.0292,
      "step": 48610
    },
    {
      "epoch": 2.4678917711558963,
      "grad_norm": 0.2826537489891052,
      "learning_rate": 3.547388192294025e-06,
      "loss": 0.0311,
      "step": 48615
    },
    {
      "epoch": 2.468145591146759,
      "grad_norm": 0.5033324360847473,
      "learning_rate": 3.5456960590216086e-06,
      "loss": 0.0294,
      "step": 48620
    },
    {
      "epoch": 2.468399411137621,
      "grad_norm": 0.21514873206615448,
      "learning_rate": 3.544003925749192e-06,
      "loss": 0.0334,
      "step": 48625
    },
    {
      "epoch": 2.4686532311284837,
      "grad_norm": 0.42751839756965637,
      "learning_rate": 3.5423117924767757e-06,
      "loss": 0.0295,
      "step": 48630
    },
    {
      "epoch": 2.4689070511193463,
      "grad_norm": 0.3107595145702362,
      "learning_rate": 3.540619659204359e-06,
      "loss": 0.0312,
      "step": 48635
    },
    {
      "epoch": 2.4691608711102084,
      "grad_norm": 0.3402955234050751,
      "learning_rate": 3.538927525931943e-06,
      "loss": 0.0338,
      "step": 48640
    },
    {
      "epoch": 2.469414691101071,
      "grad_norm": 0.20256027579307556,
      "learning_rate": 3.537235392659526e-06,
      "loss": 0.0226,
      "step": 48645
    },
    {
      "epoch": 2.4696685110919336,
      "grad_norm": 0.28199639916419983,
      "learning_rate": 3.53554325938711e-06,
      "loss": 0.0284,
      "step": 48650
    },
    {
      "epoch": 2.4699223310827962,
      "grad_norm": 0.37197086215019226,
      "learning_rate": 3.533851126114693e-06,
      "loss": 0.0344,
      "step": 48655
    },
    {
      "epoch": 2.4701761510736584,
      "grad_norm": 0.5461879968643188,
      "learning_rate": 3.5321589928422766e-06,
      "loss": 0.0337,
      "step": 48660
    },
    {
      "epoch": 2.470429971064521,
      "grad_norm": 0.32721325755119324,
      "learning_rate": 3.5304668595698597e-06,
      "loss": 0.0353,
      "step": 48665
    },
    {
      "epoch": 2.4706837910553836,
      "grad_norm": 0.28904861211776733,
      "learning_rate": 3.5287747262974437e-06,
      "loss": 0.0297,
      "step": 48670
    },
    {
      "epoch": 2.4709376110462458,
      "grad_norm": 0.36164960265159607,
      "learning_rate": 3.527082593025027e-06,
      "loss": 0.0368,
      "step": 48675
    },
    {
      "epoch": 2.4711914310371084,
      "grad_norm": 0.4914800226688385,
      "learning_rate": 3.5253904597526108e-06,
      "loss": 0.0339,
      "step": 48680
    },
    {
      "epoch": 2.471445251027971,
      "grad_norm": 0.359645277261734,
      "learning_rate": 3.523698326480194e-06,
      "loss": 0.0308,
      "step": 48685
    },
    {
      "epoch": 2.4716990710188336,
      "grad_norm": 0.2739097476005554,
      "learning_rate": 3.5220061932077774e-06,
      "loss": 0.037,
      "step": 48690
    },
    {
      "epoch": 2.4719528910096957,
      "grad_norm": 0.27334341406822205,
      "learning_rate": 3.520314059935361e-06,
      "loss": 0.0206,
      "step": 48695
    },
    {
      "epoch": 2.4722067110005583,
      "grad_norm": 0.47454833984375,
      "learning_rate": 3.518621926662944e-06,
      "loss": 0.0348,
      "step": 48700
    },
    {
      "epoch": 2.472460530991421,
      "grad_norm": 0.3284585177898407,
      "learning_rate": 3.5169297933905276e-06,
      "loss": 0.032,
      "step": 48705
    },
    {
      "epoch": 2.4727143509822835,
      "grad_norm": 0.3352780044078827,
      "learning_rate": 3.5152376601181108e-06,
      "loss": 0.0308,
      "step": 48710
    },
    {
      "epoch": 2.4729681709731457,
      "grad_norm": 0.4959048628807068,
      "learning_rate": 3.5135455268456947e-06,
      "loss": 0.031,
      "step": 48715
    },
    {
      "epoch": 2.4732219909640083,
      "grad_norm": 0.3201127052307129,
      "learning_rate": 3.511853393573278e-06,
      "loss": 0.0273,
      "step": 48720
    },
    {
      "epoch": 2.473475810954871,
      "grad_norm": 0.2874223589897156,
      "learning_rate": 3.510161260300862e-06,
      "loss": 0.0346,
      "step": 48725
    },
    {
      "epoch": 2.4737296309457335,
      "grad_norm": 0.38007476925849915,
      "learning_rate": 3.508469127028445e-06,
      "loss": 0.0364,
      "step": 48730
    },
    {
      "epoch": 2.4739834509365957,
      "grad_norm": 0.7429245710372925,
      "learning_rate": 3.5067769937560285e-06,
      "loss": 0.031,
      "step": 48735
    },
    {
      "epoch": 2.4742372709274583,
      "grad_norm": 0.4520021080970764,
      "learning_rate": 3.505084860483612e-06,
      "loss": 0.0389,
      "step": 48740
    },
    {
      "epoch": 2.474491090918321,
      "grad_norm": 0.3023478388786316,
      "learning_rate": 3.5033927272111956e-06,
      "loss": 0.0335,
      "step": 48745
    },
    {
      "epoch": 2.474744910909183,
      "grad_norm": 0.28845712542533875,
      "learning_rate": 3.5017005939387787e-06,
      "loss": 0.0255,
      "step": 48750
    },
    {
      "epoch": 2.4749987309000456,
      "grad_norm": 0.22676537930965424,
      "learning_rate": 3.5000084606663627e-06,
      "loss": 0.0263,
      "step": 48755
    },
    {
      "epoch": 2.4752525508909082,
      "grad_norm": 0.2971080243587494,
      "learning_rate": 3.4983163273939458e-06,
      "loss": 0.0307,
      "step": 48760
    },
    {
      "epoch": 2.475506370881771,
      "grad_norm": 0.47433048486709595,
      "learning_rate": 3.4966241941215297e-06,
      "loss": 0.0292,
      "step": 48765
    },
    {
      "epoch": 2.475760190872633,
      "grad_norm": 0.32119104266166687,
      "learning_rate": 3.494932060849113e-06,
      "loss": 0.0315,
      "step": 48770
    },
    {
      "epoch": 2.4760140108634956,
      "grad_norm": 0.3103313446044922,
      "learning_rate": 3.4932399275766964e-06,
      "loss": 0.0354,
      "step": 48775
    },
    {
      "epoch": 2.476267830854358,
      "grad_norm": 0.2995244860649109,
      "learning_rate": 3.4915477943042795e-06,
      "loss": 0.0315,
      "step": 48780
    },
    {
      "epoch": 2.4765216508452204,
      "grad_norm": 0.36046555638313293,
      "learning_rate": 3.489855661031863e-06,
      "loss": 0.0316,
      "step": 48785
    },
    {
      "epoch": 2.476775470836083,
      "grad_norm": 0.33610787987709045,
      "learning_rate": 3.4881635277594466e-06,
      "loss": 0.032,
      "step": 48790
    },
    {
      "epoch": 2.4770292908269456,
      "grad_norm": 0.29550817608833313,
      "learning_rate": 3.4864713944870297e-06,
      "loss": 0.0359,
      "step": 48795
    },
    {
      "epoch": 2.477283110817808,
      "grad_norm": 0.26815691590309143,
      "learning_rate": 3.4847792612146137e-06,
      "loss": 0.0299,
      "step": 48800
    },
    {
      "epoch": 2.4775369308086703,
      "grad_norm": 0.2790869474411011,
      "learning_rate": 3.483087127942197e-06,
      "loss": 0.033,
      "step": 48805
    },
    {
      "epoch": 2.477790750799533,
      "grad_norm": 0.3197694718837738,
      "learning_rate": 3.481394994669781e-06,
      "loss": 0.0301,
      "step": 48810
    },
    {
      "epoch": 2.4780445707903955,
      "grad_norm": 0.521484911441803,
      "learning_rate": 3.479702861397364e-06,
      "loss": 0.0286,
      "step": 48815
    },
    {
      "epoch": 2.4782983907812577,
      "grad_norm": 0.26468929648399353,
      "learning_rate": 3.4780107281249475e-06,
      "loss": 0.0333,
      "step": 48820
    },
    {
      "epoch": 2.4785522107721203,
      "grad_norm": 0.3268936276435852,
      "learning_rate": 3.4763185948525306e-06,
      "loss": 0.026,
      "step": 48825
    },
    {
      "epoch": 2.478806030762983,
      "grad_norm": 0.7504985332489014,
      "learning_rate": 3.4746264615801146e-06,
      "loss": 0.0318,
      "step": 48830
    },
    {
      "epoch": 2.4790598507538455,
      "grad_norm": 0.25839510560035706,
      "learning_rate": 3.4729343283076977e-06,
      "loss": 0.0334,
      "step": 48835
    },
    {
      "epoch": 2.479313670744708,
      "grad_norm": 0.38872215151786804,
      "learning_rate": 3.4712421950352816e-06,
      "loss": 0.0286,
      "step": 48840
    },
    {
      "epoch": 2.4795674907355703,
      "grad_norm": 0.44475406408309937,
      "learning_rate": 3.4695500617628648e-06,
      "loss": 0.0341,
      "step": 48845
    },
    {
      "epoch": 2.479821310726433,
      "grad_norm": 0.2652853727340698,
      "learning_rate": 3.4678579284904483e-06,
      "loss": 0.0308,
      "step": 48850
    },
    {
      "epoch": 2.4800751307172955,
      "grad_norm": 0.2847198247909546,
      "learning_rate": 3.466165795218032e-06,
      "loss": 0.0342,
      "step": 48855
    },
    {
      "epoch": 2.4803289507081576,
      "grad_norm": 0.33611080050468445,
      "learning_rate": 3.464473661945615e-06,
      "loss": 0.0349,
      "step": 48860
    },
    {
      "epoch": 2.4805827706990202,
      "grad_norm": 0.3062199056148529,
      "learning_rate": 3.4627815286731985e-06,
      "loss": 0.031,
      "step": 48865
    },
    {
      "epoch": 2.480836590689883,
      "grad_norm": 0.272950142621994,
      "learning_rate": 3.4610893954007816e-06,
      "loss": 0.0328,
      "step": 48870
    },
    {
      "epoch": 2.4810904106807454,
      "grad_norm": 0.27738696336746216,
      "learning_rate": 3.4593972621283656e-06,
      "loss": 0.031,
      "step": 48875
    },
    {
      "epoch": 2.4813442306716076,
      "grad_norm": 0.3519563674926758,
      "learning_rate": 3.4577051288559487e-06,
      "loss": 0.0302,
      "step": 48880
    },
    {
      "epoch": 2.48159805066247,
      "grad_norm": 0.4059299826622009,
      "learning_rate": 3.4560129955835327e-06,
      "loss": 0.0306,
      "step": 48885
    },
    {
      "epoch": 2.481851870653333,
      "grad_norm": 0.6799148321151733,
      "learning_rate": 3.454320862311116e-06,
      "loss": 0.0355,
      "step": 48890
    },
    {
      "epoch": 2.482105690644195,
      "grad_norm": 0.6731832027435303,
      "learning_rate": 3.4526287290386994e-06,
      "loss": 0.0288,
      "step": 48895
    },
    {
      "epoch": 2.4823595106350576,
      "grad_norm": 0.42268016934394836,
      "learning_rate": 3.450936595766283e-06,
      "loss": 0.0309,
      "step": 48900
    },
    {
      "epoch": 2.48261333062592,
      "grad_norm": 0.6146883368492126,
      "learning_rate": 3.4492444624938665e-06,
      "loss": 0.0376,
      "step": 48905
    },
    {
      "epoch": 2.4828671506167828,
      "grad_norm": 0.22841694951057434,
      "learning_rate": 3.4475523292214496e-06,
      "loss": 0.0302,
      "step": 48910
    },
    {
      "epoch": 2.483120970607645,
      "grad_norm": 0.2890803813934326,
      "learning_rate": 3.4458601959490335e-06,
      "loss": 0.0251,
      "step": 48915
    },
    {
      "epoch": 2.4833747905985075,
      "grad_norm": 0.4178946912288666,
      "learning_rate": 3.4441680626766167e-06,
      "loss": 0.0378,
      "step": 48920
    },
    {
      "epoch": 2.48362861058937,
      "grad_norm": 0.43672826886177063,
      "learning_rate": 3.4424759294042e-06,
      "loss": 0.0344,
      "step": 48925
    },
    {
      "epoch": 2.4838824305802323,
      "grad_norm": 0.32364997267723083,
      "learning_rate": 3.4407837961317838e-06,
      "loss": 0.0303,
      "step": 48930
    },
    {
      "epoch": 2.484136250571095,
      "grad_norm": 0.37183573842048645,
      "learning_rate": 3.4390916628593673e-06,
      "loss": 0.0318,
      "step": 48935
    },
    {
      "epoch": 2.4843900705619575,
      "grad_norm": 0.2628357410430908,
      "learning_rate": 3.4373995295869504e-06,
      "loss": 0.0271,
      "step": 48940
    },
    {
      "epoch": 2.48464389055282,
      "grad_norm": 0.290749192237854,
      "learning_rate": 3.435707396314534e-06,
      "loss": 0.0273,
      "step": 48945
    },
    {
      "epoch": 2.4848977105436822,
      "grad_norm": 0.8906141519546509,
      "learning_rate": 3.4340152630421175e-06,
      "loss": 0.0314,
      "step": 48950
    },
    {
      "epoch": 2.485151530534545,
      "grad_norm": 0.2430388629436493,
      "learning_rate": 3.4323231297697006e-06,
      "loss": 0.0212,
      "step": 48955
    },
    {
      "epoch": 2.4854053505254075,
      "grad_norm": 0.28215649724006653,
      "learning_rate": 3.4306309964972846e-06,
      "loss": 0.0382,
      "step": 48960
    },
    {
      "epoch": 2.48565917051627,
      "grad_norm": 0.3023797869682312,
      "learning_rate": 3.4289388632248677e-06,
      "loss": 0.0352,
      "step": 48965
    },
    {
      "epoch": 2.485912990507132,
      "grad_norm": 0.34995901584625244,
      "learning_rate": 3.4272467299524513e-06,
      "loss": 0.032,
      "step": 48970
    },
    {
      "epoch": 2.486166810497995,
      "grad_norm": 0.3282700181007385,
      "learning_rate": 3.425554596680035e-06,
      "loss": 0.0299,
      "step": 48975
    },
    {
      "epoch": 2.4864206304888574,
      "grad_norm": 0.288655549287796,
      "learning_rate": 3.4238624634076183e-06,
      "loss": 0.0281,
      "step": 48980
    },
    {
      "epoch": 2.48667445047972,
      "grad_norm": 0.5877058506011963,
      "learning_rate": 3.4221703301352015e-06,
      "loss": 0.0331,
      "step": 48985
    },
    {
      "epoch": 2.486928270470582,
      "grad_norm": 0.27871277928352356,
      "learning_rate": 3.4204781968627854e-06,
      "loss": 0.0335,
      "step": 48990
    },
    {
      "epoch": 2.487182090461445,
      "grad_norm": 0.37147513031959534,
      "learning_rate": 3.4187860635903686e-06,
      "loss": 0.0345,
      "step": 48995
    },
    {
      "epoch": 2.4874359104523074,
      "grad_norm": 0.3929043710231781,
      "learning_rate": 3.4170939303179525e-06,
      "loss": 0.0441,
      "step": 49000
    },
    {
      "epoch": 2.4876897304431695,
      "grad_norm": 0.32876530289649963,
      "learning_rate": 3.4154017970455356e-06,
      "loss": 0.0284,
      "step": 49005
    },
    {
      "epoch": 2.487943550434032,
      "grad_norm": 0.34745267033576965,
      "learning_rate": 3.413709663773119e-06,
      "loss": 0.0268,
      "step": 49010
    },
    {
      "epoch": 2.4881973704248947,
      "grad_norm": 0.489314466714859,
      "learning_rate": 3.4120175305007023e-06,
      "loss": 0.0364,
      "step": 49015
    },
    {
      "epoch": 2.4884511904157574,
      "grad_norm": 0.8166823983192444,
      "learning_rate": 3.410325397228286e-06,
      "loss": 0.035,
      "step": 49020
    },
    {
      "epoch": 2.4887050104066195,
      "grad_norm": 0.2987882196903229,
      "learning_rate": 3.4086332639558694e-06,
      "loss": 0.0366,
      "step": 49025
    },
    {
      "epoch": 2.488958830397482,
      "grad_norm": 0.25694194436073303,
      "learning_rate": 3.4069411306834525e-06,
      "loss": 0.0292,
      "step": 49030
    },
    {
      "epoch": 2.4892126503883447,
      "grad_norm": 0.9844125509262085,
      "learning_rate": 3.4052489974110365e-06,
      "loss": 0.0358,
      "step": 49035
    },
    {
      "epoch": 2.489466470379207,
      "grad_norm": 0.287350594997406,
      "learning_rate": 3.4035568641386196e-06,
      "loss": 0.0298,
      "step": 49040
    },
    {
      "epoch": 2.4897202903700695,
      "grad_norm": 0.5002569556236267,
      "learning_rate": 3.4018647308662036e-06,
      "loss": 0.0301,
      "step": 49045
    },
    {
      "epoch": 2.489974110360932,
      "grad_norm": 0.36009514331817627,
      "learning_rate": 3.4001725975937867e-06,
      "loss": 0.0319,
      "step": 49050
    },
    {
      "epoch": 2.4902279303517947,
      "grad_norm": 0.27335289120674133,
      "learning_rate": 3.3984804643213702e-06,
      "loss": 0.0241,
      "step": 49055
    },
    {
      "epoch": 2.490481750342657,
      "grad_norm": 0.43379515409469604,
      "learning_rate": 3.3967883310489534e-06,
      "loss": 0.0336,
      "step": 49060
    },
    {
      "epoch": 2.4907355703335194,
      "grad_norm": 0.5943365693092346,
      "learning_rate": 3.3950961977765373e-06,
      "loss": 0.0305,
      "step": 49065
    },
    {
      "epoch": 2.490989390324382,
      "grad_norm": 0.3196374475955963,
      "learning_rate": 3.3934040645041205e-06,
      "loss": 0.0322,
      "step": 49070
    },
    {
      "epoch": 2.491243210315244,
      "grad_norm": 0.34475764632225037,
      "learning_rate": 3.3917119312317044e-06,
      "loss": 0.032,
      "step": 49075
    },
    {
      "epoch": 2.491497030306107,
      "grad_norm": 0.27804192900657654,
      "learning_rate": 3.3900197979592875e-06,
      "loss": 0.0314,
      "step": 49080
    },
    {
      "epoch": 2.4917508502969694,
      "grad_norm": 0.3123543858528137,
      "learning_rate": 3.388327664686871e-06,
      "loss": 0.0288,
      "step": 49085
    },
    {
      "epoch": 2.492004670287832,
      "grad_norm": 0.2633233368396759,
      "learning_rate": 3.3866355314144546e-06,
      "loss": 0.0332,
      "step": 49090
    },
    {
      "epoch": 2.492258490278694,
      "grad_norm": 0.3150838613510132,
      "learning_rate": 3.384943398142038e-06,
      "loss": 0.0298,
      "step": 49095
    },
    {
      "epoch": 2.4925123102695568,
      "grad_norm": 0.2555038332939148,
      "learning_rate": 3.3832512648696213e-06,
      "loss": 0.0304,
      "step": 49100
    },
    {
      "epoch": 2.4927661302604194,
      "grad_norm": 0.2612498700618744,
      "learning_rate": 3.3815591315972044e-06,
      "loss": 0.0283,
      "step": 49105
    },
    {
      "epoch": 2.493019950251282,
      "grad_norm": 0.26925814151763916,
      "learning_rate": 3.3798669983247884e-06,
      "loss": 0.0275,
      "step": 49110
    },
    {
      "epoch": 2.493273770242144,
      "grad_norm": 0.2616172134876251,
      "learning_rate": 3.3781748650523715e-06,
      "loss": 0.0345,
      "step": 49115
    },
    {
      "epoch": 2.4935275902330067,
      "grad_norm": 0.24253271520137787,
      "learning_rate": 3.3764827317799555e-06,
      "loss": 0.0232,
      "step": 49120
    },
    {
      "epoch": 2.4937814102238693,
      "grad_norm": 0.2660481929779053,
      "learning_rate": 3.3747905985075386e-06,
      "loss": 0.0302,
      "step": 49125
    },
    {
      "epoch": 2.494035230214732,
      "grad_norm": 0.39158353209495544,
      "learning_rate": 3.373098465235122e-06,
      "loss": 0.0331,
      "step": 49130
    },
    {
      "epoch": 2.494289050205594,
      "grad_norm": 0.2616908550262451,
      "learning_rate": 3.3714063319627057e-06,
      "loss": 0.0333,
      "step": 49135
    },
    {
      "epoch": 2.4945428701964567,
      "grad_norm": 0.34958600997924805,
      "learning_rate": 3.3697141986902892e-06,
      "loss": 0.0341,
      "step": 49140
    },
    {
      "epoch": 2.4947966901873193,
      "grad_norm": 0.23418071866035461,
      "learning_rate": 3.3680220654178724e-06,
      "loss": 0.0362,
      "step": 49145
    },
    {
      "epoch": 2.4950505101781815,
      "grad_norm": 0.34714365005493164,
      "learning_rate": 3.3663299321454563e-06,
      "loss": 0.0299,
      "step": 49150
    },
    {
      "epoch": 2.495304330169044,
      "grad_norm": 0.47345075011253357,
      "learning_rate": 3.3646377988730394e-06,
      "loss": 0.0311,
      "step": 49155
    },
    {
      "epoch": 2.4955581501599067,
      "grad_norm": 0.26879772543907166,
      "learning_rate": 3.362945665600623e-06,
      "loss": 0.0292,
      "step": 49160
    },
    {
      "epoch": 2.4958119701507693,
      "grad_norm": 0.4332658350467682,
      "learning_rate": 3.3612535323282065e-06,
      "loss": 0.0307,
      "step": 49165
    },
    {
      "epoch": 2.4960657901416314,
      "grad_norm": 0.23341147601604462,
      "learning_rate": 3.35956139905579e-06,
      "loss": 0.0325,
      "step": 49170
    },
    {
      "epoch": 2.496319610132494,
      "grad_norm": 0.30529898405075073,
      "learning_rate": 3.357869265783373e-06,
      "loss": 0.0297,
      "step": 49175
    },
    {
      "epoch": 2.4965734301233566,
      "grad_norm": 0.2946845293045044,
      "learning_rate": 3.356177132510957e-06,
      "loss": 0.0252,
      "step": 49180
    },
    {
      "epoch": 2.496827250114219,
      "grad_norm": 0.32320156693458557,
      "learning_rate": 3.3544849992385403e-06,
      "loss": 0.03,
      "step": 49185
    },
    {
      "epoch": 2.4970810701050814,
      "grad_norm": 0.6085475087165833,
      "learning_rate": 3.3527928659661234e-06,
      "loss": 0.025,
      "step": 49190
    },
    {
      "epoch": 2.497334890095944,
      "grad_norm": 1.414477825164795,
      "learning_rate": 3.3511007326937074e-06,
      "loss": 0.0313,
      "step": 49195
    },
    {
      "epoch": 2.4975887100868066,
      "grad_norm": 0.36679041385650635,
      "learning_rate": 3.3494085994212905e-06,
      "loss": 0.0366,
      "step": 49200
    },
    {
      "epoch": 2.4978425300776688,
      "grad_norm": 0.4237407147884369,
      "learning_rate": 3.347716466148874e-06,
      "loss": 0.0366,
      "step": 49205
    },
    {
      "epoch": 2.4980963500685314,
      "grad_norm": 0.2822991907596588,
      "learning_rate": 3.3460243328764576e-06,
      "loss": 0.0307,
      "step": 49210
    },
    {
      "epoch": 2.498350170059394,
      "grad_norm": 0.21562591195106506,
      "learning_rate": 3.344332199604041e-06,
      "loss": 0.0298,
      "step": 49215
    },
    {
      "epoch": 2.498603990050256,
      "grad_norm": 0.3568347096443176,
      "learning_rate": 3.3426400663316243e-06,
      "loss": 0.0308,
      "step": 49220
    },
    {
      "epoch": 2.4988578100411187,
      "grad_norm": 0.42248597741127014,
      "learning_rate": 3.3409479330592082e-06,
      "loss": 0.0291,
      "step": 49225
    },
    {
      "epoch": 2.4991116300319813,
      "grad_norm": 0.3704330623149872,
      "learning_rate": 3.3392557997867913e-06,
      "loss": 0.0299,
      "step": 49230
    },
    {
      "epoch": 2.499365450022844,
      "grad_norm": 0.2916848659515381,
      "learning_rate": 3.3375636665143753e-06,
      "loss": 0.0315,
      "step": 49235
    },
    {
      "epoch": 2.499619270013706,
      "grad_norm": 0.25191444158554077,
      "learning_rate": 3.3358715332419584e-06,
      "loss": 0.0293,
      "step": 49240
    },
    {
      "epoch": 2.4998730900045687,
      "grad_norm": 0.3909609019756317,
      "learning_rate": 3.334179399969542e-06,
      "loss": 0.0355,
      "step": 49245
    },
    {
      "epoch": 2.5001269099954313,
      "grad_norm": 0.3124801218509674,
      "learning_rate": 3.332487266697125e-06,
      "loss": 0.0252,
      "step": 49250
    },
    {
      "epoch": 2.5003807299862935,
      "grad_norm": 1.1593356132507324,
      "learning_rate": 3.330795133424709e-06,
      "loss": 0.0293,
      "step": 49255
    },
    {
      "epoch": 2.500634549977156,
      "grad_norm": 0.318891704082489,
      "learning_rate": 3.329103000152292e-06,
      "loss": 0.0277,
      "step": 49260
    },
    {
      "epoch": 2.5008883699680187,
      "grad_norm": 0.25046369433403015,
      "learning_rate": 3.3274108668798753e-06,
      "loss": 0.0236,
      "step": 49265
    },
    {
      "epoch": 2.5011421899588813,
      "grad_norm": 0.8154666423797607,
      "learning_rate": 3.3257187336074593e-06,
      "loss": 0.04,
      "step": 49270
    },
    {
      "epoch": 2.501396009949744,
      "grad_norm": 0.5566936731338501,
      "learning_rate": 3.3240266003350424e-06,
      "loss": 0.0282,
      "step": 49275
    },
    {
      "epoch": 2.501649829940606,
      "grad_norm": 0.3415115177631378,
      "learning_rate": 3.3223344670626264e-06,
      "loss": 0.0314,
      "step": 49280
    },
    {
      "epoch": 2.5019036499314686,
      "grad_norm": 0.2734259366989136,
      "learning_rate": 3.3206423337902095e-06,
      "loss": 0.031,
      "step": 49285
    },
    {
      "epoch": 2.5021574699223312,
      "grad_norm": 0.179351806640625,
      "learning_rate": 3.318950200517793e-06,
      "loss": 0.025,
      "step": 49290
    },
    {
      "epoch": 2.5024112899131934,
      "grad_norm": 0.3378109335899353,
      "learning_rate": 3.317258067245376e-06,
      "loss": 0.0262,
      "step": 49295
    },
    {
      "epoch": 2.502665109904056,
      "grad_norm": 0.38636019825935364,
      "learning_rate": 3.31556593397296e-06,
      "loss": 0.0248,
      "step": 49300
    },
    {
      "epoch": 2.5029189298949186,
      "grad_norm": 0.21963395178318024,
      "learning_rate": 3.3138738007005432e-06,
      "loss": 0.027,
      "step": 49305
    },
    {
      "epoch": 2.503172749885781,
      "grad_norm": 0.2768746614456177,
      "learning_rate": 3.312181667428127e-06,
      "loss": 0.0253,
      "step": 49310
    },
    {
      "epoch": 2.5034265698766434,
      "grad_norm": 0.41382497549057007,
      "learning_rate": 3.3104895341557103e-06,
      "loss": 0.0315,
      "step": 49315
    },
    {
      "epoch": 2.503680389867506,
      "grad_norm": 0.5263127088546753,
      "learning_rate": 3.308797400883294e-06,
      "loss": 0.0287,
      "step": 49320
    },
    {
      "epoch": 2.5039342098583686,
      "grad_norm": 0.33877232670783997,
      "learning_rate": 3.3071052676108774e-06,
      "loss": 0.0286,
      "step": 49325
    },
    {
      "epoch": 2.5041880298492307,
      "grad_norm": 0.27942949533462524,
      "learning_rate": 3.305413134338461e-06,
      "loss": 0.0354,
      "step": 49330
    },
    {
      "epoch": 2.5044418498400933,
      "grad_norm": 0.35031718015670776,
      "learning_rate": 3.303721001066044e-06,
      "loss": 0.0369,
      "step": 49335
    },
    {
      "epoch": 2.504695669830956,
      "grad_norm": 0.32853952050209045,
      "learning_rate": 3.302028867793628e-06,
      "loss": 0.0297,
      "step": 49340
    },
    {
      "epoch": 2.5049494898218185,
      "grad_norm": 0.2991621792316437,
      "learning_rate": 3.300336734521211e-06,
      "loss": 0.0328,
      "step": 49345
    },
    {
      "epoch": 2.5052033098126807,
      "grad_norm": 0.28316929936408997,
      "learning_rate": 3.2986446012487943e-06,
      "loss": 0.0245,
      "step": 49350
    },
    {
      "epoch": 2.5054571298035433,
      "grad_norm": 0.20764075219631195,
      "learning_rate": 3.2969524679763783e-06,
      "loss": 0.0282,
      "step": 49355
    },
    {
      "epoch": 2.505710949794406,
      "grad_norm": 0.19954226911067963,
      "learning_rate": 3.2952603347039614e-06,
      "loss": 0.0246,
      "step": 49360
    },
    {
      "epoch": 2.505964769785268,
      "grad_norm": 0.3868962526321411,
      "learning_rate": 3.293568201431545e-06,
      "loss": 0.0236,
      "step": 49365
    },
    {
      "epoch": 2.5062185897761307,
      "grad_norm": 0.23645120859146118,
      "learning_rate": 3.2918760681591285e-06,
      "loss": 0.0395,
      "step": 49370
    },
    {
      "epoch": 2.5064724097669933,
      "grad_norm": 0.504305362701416,
      "learning_rate": 3.290183934886712e-06,
      "loss": 0.031,
      "step": 49375
    },
    {
      "epoch": 2.506726229757856,
      "grad_norm": 0.3593463599681854,
      "learning_rate": 3.288491801614295e-06,
      "loss": 0.0329,
      "step": 49380
    },
    {
      "epoch": 2.5069800497487185,
      "grad_norm": 0.3594639301300049,
      "learning_rate": 3.286799668341879e-06,
      "loss": 0.0345,
      "step": 49385
    },
    {
      "epoch": 2.5072338697395806,
      "grad_norm": 0.40261438488960266,
      "learning_rate": 3.2851075350694622e-06,
      "loss": 0.0306,
      "step": 49390
    },
    {
      "epoch": 2.5074876897304432,
      "grad_norm": 0.1871395707130432,
      "learning_rate": 3.283415401797046e-06,
      "loss": 0.0269,
      "step": 49395
    },
    {
      "epoch": 2.5077415097213054,
      "grad_norm": 0.26229771971702576,
      "learning_rate": 3.2817232685246293e-06,
      "loss": 0.0298,
      "step": 49400
    },
    {
      "epoch": 2.507995329712168,
      "grad_norm": 0.32470229268074036,
      "learning_rate": 3.280031135252213e-06,
      "loss": 0.0231,
      "step": 49405
    },
    {
      "epoch": 2.5082491497030306,
      "grad_norm": 0.24094648659229279,
      "learning_rate": 3.278339001979796e-06,
      "loss": 0.0268,
      "step": 49410
    },
    {
      "epoch": 2.508502969693893,
      "grad_norm": 0.30596256256103516,
      "learning_rate": 3.27664686870738e-06,
      "loss": 0.0267,
      "step": 49415
    },
    {
      "epoch": 2.508756789684756,
      "grad_norm": 0.6077698469161987,
      "learning_rate": 3.274954735434963e-06,
      "loss": 0.0372,
      "step": 49420
    },
    {
      "epoch": 2.509010609675618,
      "grad_norm": 0.3845747411251068,
      "learning_rate": 3.273262602162546e-06,
      "loss": 0.04,
      "step": 49425
    },
    {
      "epoch": 2.5092644296664806,
      "grad_norm": 0.3875996172428131,
      "learning_rate": 3.27157046889013e-06,
      "loss": 0.0307,
      "step": 49430
    },
    {
      "epoch": 2.509518249657343,
      "grad_norm": 0.33381351828575134,
      "learning_rate": 3.2698783356177133e-06,
      "loss": 0.034,
      "step": 49435
    },
    {
      "epoch": 2.5097720696482053,
      "grad_norm": 0.6025996208190918,
      "learning_rate": 3.2681862023452972e-06,
      "loss": 0.0387,
      "step": 49440
    },
    {
      "epoch": 2.510025889639068,
      "grad_norm": 0.2946641147136688,
      "learning_rate": 3.2664940690728804e-06,
      "loss": 0.0315,
      "step": 49445
    },
    {
      "epoch": 2.5102797096299305,
      "grad_norm": 0.6482958793640137,
      "learning_rate": 3.264801935800464e-06,
      "loss": 0.0308,
      "step": 49450
    },
    {
      "epoch": 2.510533529620793,
      "grad_norm": 0.26497596502304077,
      "learning_rate": 3.263109802528047e-06,
      "loss": 0.0322,
      "step": 49455
    },
    {
      "epoch": 2.5107873496116553,
      "grad_norm": 0.5210638046264648,
      "learning_rate": 3.261417669255631e-06,
      "loss": 0.0343,
      "step": 49460
    },
    {
      "epoch": 2.511041169602518,
      "grad_norm": 0.23202571272850037,
      "learning_rate": 3.259725535983214e-06,
      "loss": 0.0296,
      "step": 49465
    },
    {
      "epoch": 2.5112949895933805,
      "grad_norm": 0.3447119891643524,
      "learning_rate": 3.258033402710798e-06,
      "loss": 0.0361,
      "step": 49470
    },
    {
      "epoch": 2.5115488095842426,
      "grad_norm": 0.22966940701007843,
      "learning_rate": 3.2563412694383812e-06,
      "loss": 0.0253,
      "step": 49475
    },
    {
      "epoch": 2.5118026295751053,
      "grad_norm": 0.43663614988327026,
      "learning_rate": 3.2546491361659648e-06,
      "loss": 0.0384,
      "step": 49480
    },
    {
      "epoch": 2.512056449565968,
      "grad_norm": 0.3742697238922119,
      "learning_rate": 3.2529570028935483e-06,
      "loss": 0.0419,
      "step": 49485
    },
    {
      "epoch": 2.5123102695568305,
      "grad_norm": 0.3168654441833496,
      "learning_rate": 3.251264869621132e-06,
      "loss": 0.0234,
      "step": 49490
    },
    {
      "epoch": 2.5125640895476926,
      "grad_norm": 0.35261592268943787,
      "learning_rate": 3.249572736348715e-06,
      "loss": 0.0348,
      "step": 49495
    },
    {
      "epoch": 2.512817909538555,
      "grad_norm": 0.3630485534667969,
      "learning_rate": 3.247880603076299e-06,
      "loss": 0.0372,
      "step": 49500
    },
    {
      "epoch": 2.513071729529418,
      "grad_norm": 0.32719823718070984,
      "learning_rate": 3.246188469803882e-06,
      "loss": 0.0292,
      "step": 49505
    },
    {
      "epoch": 2.51332554952028,
      "grad_norm": 0.3303397297859192,
      "learning_rate": 3.244496336531465e-06,
      "loss": 0.0361,
      "step": 49510
    },
    {
      "epoch": 2.5135793695111426,
      "grad_norm": 0.3644379675388336,
      "learning_rate": 3.242804203259049e-06,
      "loss": 0.0334,
      "step": 49515
    },
    {
      "epoch": 2.513833189502005,
      "grad_norm": 0.2639160752296448,
      "learning_rate": 3.2411120699866323e-06,
      "loss": 0.0218,
      "step": 49520
    },
    {
      "epoch": 2.514087009492868,
      "grad_norm": 0.408854216337204,
      "learning_rate": 3.239419936714216e-06,
      "loss": 0.03,
      "step": 49525
    },
    {
      "epoch": 2.5143408294837304,
      "grad_norm": 0.29264143109321594,
      "learning_rate": 3.2377278034417994e-06,
      "loss": 0.0264,
      "step": 49530
    },
    {
      "epoch": 2.5145946494745925,
      "grad_norm": 0.25693920254707336,
      "learning_rate": 3.236035670169383e-06,
      "loss": 0.0387,
      "step": 49535
    },
    {
      "epoch": 2.514848469465455,
      "grad_norm": 0.36717018485069275,
      "learning_rate": 3.234343536896966e-06,
      "loss": 0.0352,
      "step": 49540
    },
    {
      "epoch": 2.5151022894563173,
      "grad_norm": 0.2756657004356384,
      "learning_rate": 3.23265140362455e-06,
      "loss": 0.0352,
      "step": 49545
    },
    {
      "epoch": 2.51535610944718,
      "grad_norm": 0.2907589077949524,
      "learning_rate": 3.230959270352133e-06,
      "loss": 0.0312,
      "step": 49550
    },
    {
      "epoch": 2.5156099294380425,
      "grad_norm": 0.3871802091598511,
      "learning_rate": 3.2292671370797167e-06,
      "loss": 0.041,
      "step": 49555
    },
    {
      "epoch": 2.515863749428905,
      "grad_norm": 0.3290282189846039,
      "learning_rate": 3.2275750038073e-06,
      "loss": 0.0229,
      "step": 49560
    },
    {
      "epoch": 2.5161175694197677,
      "grad_norm": 0.8194361925125122,
      "learning_rate": 3.2258828705348837e-06,
      "loss": 0.0366,
      "step": 49565
    },
    {
      "epoch": 2.51637138941063,
      "grad_norm": 0.3643014430999756,
      "learning_rate": 3.224190737262467e-06,
      "loss": 0.0345,
      "step": 49570
    },
    {
      "epoch": 2.5166252094014925,
      "grad_norm": 0.5090135931968689,
      "learning_rate": 3.222498603990051e-06,
      "loss": 0.0307,
      "step": 49575
    },
    {
      "epoch": 2.516879029392355,
      "grad_norm": 0.31708189845085144,
      "learning_rate": 3.220806470717634e-06,
      "loss": 0.0289,
      "step": 49580
    },
    {
      "epoch": 2.5171328493832172,
      "grad_norm": 0.4696919023990631,
      "learning_rate": 3.219114337445217e-06,
      "loss": 0.0393,
      "step": 49585
    },
    {
      "epoch": 2.51738666937408,
      "grad_norm": 0.4421710669994354,
      "learning_rate": 3.217422204172801e-06,
      "loss": 0.0371,
      "step": 49590
    },
    {
      "epoch": 2.5176404893649424,
      "grad_norm": 0.38796231150627136,
      "learning_rate": 3.215730070900384e-06,
      "loss": 0.0321,
      "step": 49595
    },
    {
      "epoch": 2.517894309355805,
      "grad_norm": 0.27723702788352966,
      "learning_rate": 3.2140379376279677e-06,
      "loss": 0.0235,
      "step": 49600
    },
    {
      "epoch": 2.518148129346667,
      "grad_norm": 0.27128395438194275,
      "learning_rate": 3.2123458043555513e-06,
      "loss": 0.0289,
      "step": 49605
    },
    {
      "epoch": 2.51840194933753,
      "grad_norm": 0.31028029322624207,
      "learning_rate": 3.210653671083135e-06,
      "loss": 0.0296,
      "step": 49610
    },
    {
      "epoch": 2.5186557693283924,
      "grad_norm": 0.2169944792985916,
      "learning_rate": 3.208961537810718e-06,
      "loss": 0.0299,
      "step": 49615
    },
    {
      "epoch": 2.5189095893192546,
      "grad_norm": 0.4345290958881378,
      "learning_rate": 3.207269404538302e-06,
      "loss": 0.0319,
      "step": 49620
    },
    {
      "epoch": 2.519163409310117,
      "grad_norm": 0.32916420698165894,
      "learning_rate": 3.205577271265885e-06,
      "loss": 0.0314,
      "step": 49625
    },
    {
      "epoch": 2.5194172293009798,
      "grad_norm": 0.2635617256164551,
      "learning_rate": 3.203885137993469e-06,
      "loss": 0.026,
      "step": 49630
    },
    {
      "epoch": 2.5196710492918424,
      "grad_norm": 0.5266566276550293,
      "learning_rate": 3.202193004721052e-06,
      "loss": 0.0374,
      "step": 49635
    },
    {
      "epoch": 2.519924869282705,
      "grad_norm": 0.35786283016204834,
      "learning_rate": 3.2005008714486356e-06,
      "loss": 0.0336,
      "step": 49640
    },
    {
      "epoch": 2.520178689273567,
      "grad_norm": 0.34204965829849243,
      "learning_rate": 3.1988087381762188e-06,
      "loss": 0.0341,
      "step": 49645
    },
    {
      "epoch": 2.5204325092644297,
      "grad_norm": 2.4020159244537354,
      "learning_rate": 3.1971166049038027e-06,
      "loss": 0.0332,
      "step": 49650
    },
    {
      "epoch": 2.520686329255292,
      "grad_norm": 0.30302706360816956,
      "learning_rate": 3.195424471631386e-06,
      "loss": 0.0282,
      "step": 49655
    },
    {
      "epoch": 2.5209401492461545,
      "grad_norm": 0.3472190201282501,
      "learning_rate": 3.19373233835897e-06,
      "loss": 0.0292,
      "step": 49660
    },
    {
      "epoch": 2.521193969237017,
      "grad_norm": 0.31304502487182617,
      "learning_rate": 3.192040205086553e-06,
      "loss": 0.0257,
      "step": 49665
    },
    {
      "epoch": 2.5214477892278797,
      "grad_norm": 0.3431473970413208,
      "learning_rate": 3.190348071814136e-06,
      "loss": 0.0325,
      "step": 49670
    },
    {
      "epoch": 2.5217016092187423,
      "grad_norm": 0.22988146543502808,
      "learning_rate": 3.18865593854172e-06,
      "loss": 0.0266,
      "step": 49675
    },
    {
      "epoch": 2.5219554292096045,
      "grad_norm": 0.32324403524398804,
      "learning_rate": 3.186963805269303e-06,
      "loss": 0.0264,
      "step": 49680
    },
    {
      "epoch": 2.522209249200467,
      "grad_norm": 0.374230295419693,
      "learning_rate": 3.1852716719968867e-06,
      "loss": 0.0336,
      "step": 49685
    },
    {
      "epoch": 2.5224630691913292,
      "grad_norm": 0.37457334995269775,
      "learning_rate": 3.18357953872447e-06,
      "loss": 0.0384,
      "step": 49690
    },
    {
      "epoch": 2.522716889182192,
      "grad_norm": 0.35331031680107117,
      "learning_rate": 3.1818874054520538e-06,
      "loss": 0.0324,
      "step": 49695
    },
    {
      "epoch": 2.5229707091730544,
      "grad_norm": 0.24105656147003174,
      "learning_rate": 3.180195272179637e-06,
      "loss": 0.0325,
      "step": 49700
    },
    {
      "epoch": 2.523224529163917,
      "grad_norm": 0.6230748891830444,
      "learning_rate": 3.178503138907221e-06,
      "loss": 0.0341,
      "step": 49705
    },
    {
      "epoch": 2.5234783491547796,
      "grad_norm": 0.27622297406196594,
      "learning_rate": 3.176811005634804e-06,
      "loss": 0.0314,
      "step": 49710
    },
    {
      "epoch": 2.523732169145642,
      "grad_norm": 0.4867340922355652,
      "learning_rate": 3.1751188723623875e-06,
      "loss": 0.0327,
      "step": 49715
    },
    {
      "epoch": 2.5239859891365044,
      "grad_norm": 0.19795769453048706,
      "learning_rate": 3.173426739089971e-06,
      "loss": 0.0297,
      "step": 49720
    },
    {
      "epoch": 2.524239809127367,
      "grad_norm": 0.32494819164276123,
      "learning_rate": 3.1717346058175546e-06,
      "loss": 0.0239,
      "step": 49725
    },
    {
      "epoch": 2.524493629118229,
      "grad_norm": 0.3516275882720947,
      "learning_rate": 3.1700424725451377e-06,
      "loss": 0.034,
      "step": 49730
    },
    {
      "epoch": 2.5247474491090918,
      "grad_norm": 0.3350643217563629,
      "learning_rate": 3.1683503392727217e-06,
      "loss": 0.0334,
      "step": 49735
    },
    {
      "epoch": 2.5250012690999544,
      "grad_norm": 0.24879878759384155,
      "learning_rate": 3.166658206000305e-06,
      "loss": 0.0302,
      "step": 49740
    },
    {
      "epoch": 2.525255089090817,
      "grad_norm": 0.28340357542037964,
      "learning_rate": 3.1649660727278884e-06,
      "loss": 0.0348,
      "step": 49745
    },
    {
      "epoch": 2.525508909081679,
      "grad_norm": 0.2749658524990082,
      "learning_rate": 3.163273939455472e-06,
      "loss": 0.0321,
      "step": 49750
    },
    {
      "epoch": 2.5257627290725417,
      "grad_norm": 0.45516085624694824,
      "learning_rate": 3.161581806183055e-06,
      "loss": 0.0397,
      "step": 49755
    },
    {
      "epoch": 2.5260165490634043,
      "grad_norm": 0.514147937297821,
      "learning_rate": 3.1598896729106386e-06,
      "loss": 0.0381,
      "step": 49760
    },
    {
      "epoch": 2.5262703690542665,
      "grad_norm": 0.2310241311788559,
      "learning_rate": 3.158197539638222e-06,
      "loss": 0.0309,
      "step": 49765
    },
    {
      "epoch": 2.526524189045129,
      "grad_norm": 0.33417701721191406,
      "learning_rate": 3.1565054063658057e-06,
      "loss": 0.0352,
      "step": 49770
    },
    {
      "epoch": 2.5267780090359917,
      "grad_norm": 0.19402511417865753,
      "learning_rate": 3.154813273093389e-06,
      "loss": 0.0324,
      "step": 49775
    },
    {
      "epoch": 2.5270318290268543,
      "grad_norm": 0.4040732681751251,
      "learning_rate": 3.1531211398209728e-06,
      "loss": 0.0405,
      "step": 49780
    },
    {
      "epoch": 2.527285649017717,
      "grad_norm": 0.45084550976753235,
      "learning_rate": 3.151429006548556e-06,
      "loss": 0.0321,
      "step": 49785
    },
    {
      "epoch": 2.527539469008579,
      "grad_norm": 0.3171873390674591,
      "learning_rate": 3.1497368732761394e-06,
      "loss": 0.0253,
      "step": 49790
    },
    {
      "epoch": 2.5277932889994417,
      "grad_norm": 0.33110058307647705,
      "learning_rate": 3.148044740003723e-06,
      "loss": 0.0292,
      "step": 49795
    },
    {
      "epoch": 2.528047108990304,
      "grad_norm": 0.3826856315135956,
      "learning_rate": 3.1463526067313065e-06,
      "loss": 0.031,
      "step": 49800
    },
    {
      "epoch": 2.5283009289811664,
      "grad_norm": 0.2924165725708008,
      "learning_rate": 3.1446604734588896e-06,
      "loss": 0.0296,
      "step": 49805
    },
    {
      "epoch": 2.528554748972029,
      "grad_norm": 0.22940395772457123,
      "learning_rate": 3.1429683401864736e-06,
      "loss": 0.0255,
      "step": 49810
    },
    {
      "epoch": 2.5288085689628916,
      "grad_norm": 0.20305858552455902,
      "learning_rate": 3.1412762069140567e-06,
      "loss": 0.0251,
      "step": 49815
    },
    {
      "epoch": 2.5290623889537542,
      "grad_norm": 0.36855000257492065,
      "learning_rate": 3.1395840736416407e-06,
      "loss": 0.03,
      "step": 49820
    },
    {
      "epoch": 2.5293162089446164,
      "grad_norm": 0.3701513409614563,
      "learning_rate": 3.137891940369224e-06,
      "loss": 0.0385,
      "step": 49825
    },
    {
      "epoch": 2.529570028935479,
      "grad_norm": 0.368150919675827,
      "learning_rate": 3.136199807096807e-06,
      "loss": 0.0305,
      "step": 49830
    },
    {
      "epoch": 2.5298238489263416,
      "grad_norm": 0.407752126455307,
      "learning_rate": 3.1345076738243905e-06,
      "loss": 0.038,
      "step": 49835
    },
    {
      "epoch": 2.5300776689172038,
      "grad_norm": 0.28789639472961426,
      "learning_rate": 3.132815540551974e-06,
      "loss": 0.0259,
      "step": 49840
    },
    {
      "epoch": 2.5303314889080664,
      "grad_norm": 0.30561700463294983,
      "learning_rate": 3.1311234072795576e-06,
      "loss": 0.0359,
      "step": 49845
    },
    {
      "epoch": 2.530585308898929,
      "grad_norm": 0.4357641339302063,
      "learning_rate": 3.1294312740071407e-06,
      "loss": 0.0244,
      "step": 49850
    },
    {
      "epoch": 2.5308391288897916,
      "grad_norm": 0.3821885287761688,
      "learning_rate": 3.1277391407347247e-06,
      "loss": 0.033,
      "step": 49855
    },
    {
      "epoch": 2.5310929488806537,
      "grad_norm": 0.28230446577072144,
      "learning_rate": 3.126047007462308e-06,
      "loss": 0.0385,
      "step": 49860
    },
    {
      "epoch": 2.5313467688715163,
      "grad_norm": 0.39753055572509766,
      "learning_rate": 3.1243548741898918e-06,
      "loss": 0.0361,
      "step": 49865
    },
    {
      "epoch": 2.531600588862379,
      "grad_norm": 0.2797062397003174,
      "learning_rate": 3.122662740917475e-06,
      "loss": 0.0287,
      "step": 49870
    },
    {
      "epoch": 2.531854408853241,
      "grad_norm": 0.27603286504745483,
      "learning_rate": 3.1209706076450584e-06,
      "loss": 0.0242,
      "step": 49875
    },
    {
      "epoch": 2.5321082288441037,
      "grad_norm": 0.3599488139152527,
      "learning_rate": 3.1192784743726415e-06,
      "loss": 0.0369,
      "step": 49880
    },
    {
      "epoch": 2.5323620488349663,
      "grad_norm": 0.2599182426929474,
      "learning_rate": 3.1175863411002255e-06,
      "loss": 0.0299,
      "step": 49885
    },
    {
      "epoch": 2.532615868825829,
      "grad_norm": 0.3144007921218872,
      "learning_rate": 3.1158942078278086e-06,
      "loss": 0.0284,
      "step": 49890
    },
    {
      "epoch": 2.532869688816691,
      "grad_norm": 0.24892964959144592,
      "learning_rate": 3.1142020745553926e-06,
      "loss": 0.0319,
      "step": 49895
    },
    {
      "epoch": 2.5331235088075537,
      "grad_norm": 0.3454008102416992,
      "learning_rate": 3.1125099412829757e-06,
      "loss": 0.0287,
      "step": 49900
    },
    {
      "epoch": 2.5333773287984163,
      "grad_norm": 0.6400479674339294,
      "learning_rate": 3.1108178080105593e-06,
      "loss": 0.0318,
      "step": 49905
    },
    {
      "epoch": 2.5336311487892784,
      "grad_norm": 0.2760767936706543,
      "learning_rate": 3.109125674738143e-06,
      "loss": 0.0311,
      "step": 49910
    },
    {
      "epoch": 2.533884968780141,
      "grad_norm": 0.40467485785484314,
      "learning_rate": 3.107433541465726e-06,
      "loss": 0.0323,
      "step": 49915
    },
    {
      "epoch": 2.5341387887710036,
      "grad_norm": 0.46811002492904663,
      "learning_rate": 3.1057414081933095e-06,
      "loss": 0.0348,
      "step": 49920
    },
    {
      "epoch": 2.5343926087618662,
      "grad_norm": 0.43257901072502136,
      "learning_rate": 3.1040492749208926e-06,
      "loss": 0.0345,
      "step": 49925
    },
    {
      "epoch": 2.534646428752729,
      "grad_norm": 0.2802751660346985,
      "learning_rate": 3.1023571416484766e-06,
      "loss": 0.0319,
      "step": 49930
    },
    {
      "epoch": 2.534900248743591,
      "grad_norm": 0.2783406376838684,
      "learning_rate": 3.1006650083760597e-06,
      "loss": 0.0304,
      "step": 49935
    },
    {
      "epoch": 2.5351540687344536,
      "grad_norm": 0.3503018021583557,
      "learning_rate": 3.0989728751036437e-06,
      "loss": 0.029,
      "step": 49940
    },
    {
      "epoch": 2.5354078887253158,
      "grad_norm": 0.42926421761512756,
      "learning_rate": 3.0972807418312268e-06,
      "loss": 0.0306,
      "step": 49945
    },
    {
      "epoch": 2.5356617087161784,
      "grad_norm": 0.2726653516292572,
      "learning_rate": 3.0955886085588103e-06,
      "loss": 0.0237,
      "step": 49950
    },
    {
      "epoch": 2.535915528707041,
      "grad_norm": 0.3227391242980957,
      "learning_rate": 3.093896475286394e-06,
      "loss": 0.0301,
      "step": 49955
    },
    {
      "epoch": 2.5361693486979036,
      "grad_norm": 0.28392598032951355,
      "learning_rate": 3.0922043420139774e-06,
      "loss": 0.0299,
      "step": 49960
    },
    {
      "epoch": 2.536423168688766,
      "grad_norm": 0.3558497726917267,
      "learning_rate": 3.0905122087415605e-06,
      "loss": 0.0283,
      "step": 49965
    },
    {
      "epoch": 2.5366769886796283,
      "grad_norm": 0.16093190014362335,
      "learning_rate": 3.0888200754691445e-06,
      "loss": 0.0243,
      "step": 49970
    },
    {
      "epoch": 2.536930808670491,
      "grad_norm": 0.3414202034473419,
      "learning_rate": 3.0871279421967276e-06,
      "loss": 0.035,
      "step": 49975
    },
    {
      "epoch": 2.5371846286613535,
      "grad_norm": 0.2506318688392639,
      "learning_rate": 3.0854358089243116e-06,
      "loss": 0.0319,
      "step": 49980
    },
    {
      "epoch": 2.5374384486522157,
      "grad_norm": 0.350211501121521,
      "learning_rate": 3.0837436756518947e-06,
      "loss": 0.0339,
      "step": 49985
    },
    {
      "epoch": 2.5376922686430783,
      "grad_norm": 0.4482111632823944,
      "learning_rate": 3.082051542379478e-06,
      "loss": 0.0364,
      "step": 49990
    },
    {
      "epoch": 2.537946088633941,
      "grad_norm": 0.25266894698143005,
      "learning_rate": 3.0803594091070614e-06,
      "loss": 0.0314,
      "step": 49995
    },
    {
      "epoch": 2.5381999086248035,
      "grad_norm": 0.2782626748085022,
      "learning_rate": 3.078667275834645e-06,
      "loss": 0.0269,
      "step": 50000
    },
    {
      "epoch": 2.5384537286156656,
      "grad_norm": 0.3150825798511505,
      "learning_rate": 3.0769751425622285e-06,
      "loss": 0.0337,
      "step": 50005
    },
    {
      "epoch": 2.5387075486065283,
      "grad_norm": 0.33296340703964233,
      "learning_rate": 3.0752830092898116e-06,
      "loss": 0.0312,
      "step": 50010
    },
    {
      "epoch": 2.538961368597391,
      "grad_norm": 0.4527866840362549,
      "learning_rate": 3.0735908760173956e-06,
      "loss": 0.0298,
      "step": 50015
    },
    {
      "epoch": 2.539215188588253,
      "grad_norm": 0.33419156074523926,
      "learning_rate": 3.0718987427449787e-06,
      "loss": 0.0305,
      "step": 50020
    },
    {
      "epoch": 2.5394690085791156,
      "grad_norm": 0.29681098461151123,
      "learning_rate": 3.0702066094725626e-06,
      "loss": 0.0314,
      "step": 50025
    },
    {
      "epoch": 2.539722828569978,
      "grad_norm": 0.3465264141559601,
      "learning_rate": 3.0685144762001458e-06,
      "loss": 0.027,
      "step": 50030
    },
    {
      "epoch": 2.539976648560841,
      "grad_norm": 0.3378564417362213,
      "learning_rate": 3.0668223429277293e-06,
      "loss": 0.0308,
      "step": 50035
    },
    {
      "epoch": 2.540230468551703,
      "grad_norm": 0.2836815118789673,
      "learning_rate": 3.0651302096553124e-06,
      "loss": 0.0269,
      "step": 50040
    },
    {
      "epoch": 2.5404842885425656,
      "grad_norm": 0.32169854640960693,
      "learning_rate": 3.0634380763828964e-06,
      "loss": 0.0253,
      "step": 50045
    },
    {
      "epoch": 2.540738108533428,
      "grad_norm": 0.29076623916625977,
      "learning_rate": 3.0617459431104795e-06,
      "loss": 0.0341,
      "step": 50050
    },
    {
      "epoch": 2.5409919285242903,
      "grad_norm": 0.36522313952445984,
      "learning_rate": 3.0600538098380635e-06,
      "loss": 0.0313,
      "step": 50055
    },
    {
      "epoch": 2.541245748515153,
      "grad_norm": 0.4595192074775696,
      "learning_rate": 3.0583616765656466e-06,
      "loss": 0.0383,
      "step": 50060
    },
    {
      "epoch": 2.5414995685060155,
      "grad_norm": 0.18228591978549957,
      "learning_rate": 3.05666954329323e-06,
      "loss": 0.0287,
      "step": 50065
    },
    {
      "epoch": 2.541753388496878,
      "grad_norm": 0.3524336814880371,
      "learning_rate": 3.0549774100208137e-06,
      "loss": 0.0272,
      "step": 50070
    },
    {
      "epoch": 2.5420072084877408,
      "grad_norm": 0.2865443825721741,
      "learning_rate": 3.053285276748397e-06,
      "loss": 0.031,
      "step": 50075
    },
    {
      "epoch": 2.542261028478603,
      "grad_norm": 0.3282105028629303,
      "learning_rate": 3.0515931434759804e-06,
      "loss": 0.0278,
      "step": 50080
    },
    {
      "epoch": 2.5425148484694655,
      "grad_norm": 0.27646273374557495,
      "learning_rate": 3.0499010102035635e-06,
      "loss": 0.0293,
      "step": 50085
    },
    {
      "epoch": 2.5427686684603277,
      "grad_norm": 0.3110489547252655,
      "learning_rate": 3.0482088769311475e-06,
      "loss": 0.0237,
      "step": 50090
    },
    {
      "epoch": 2.5430224884511903,
      "grad_norm": 0.4063291549682617,
      "learning_rate": 3.0465167436587306e-06,
      "loss": 0.0306,
      "step": 50095
    },
    {
      "epoch": 2.543276308442053,
      "grad_norm": 0.3129997253417969,
      "learning_rate": 3.0448246103863145e-06,
      "loss": 0.0305,
      "step": 50100
    },
    {
      "epoch": 2.5435301284329155,
      "grad_norm": 0.39521104097366333,
      "learning_rate": 3.0431324771138977e-06,
      "loss": 0.0406,
      "step": 50105
    },
    {
      "epoch": 2.543783948423778,
      "grad_norm": 0.29474636912345886,
      "learning_rate": 3.041440343841481e-06,
      "loss": 0.0318,
      "step": 50110
    },
    {
      "epoch": 2.5440377684146402,
      "grad_norm": 0.2978830933570862,
      "learning_rate": 3.0397482105690647e-06,
      "loss": 0.0348,
      "step": 50115
    },
    {
      "epoch": 2.544291588405503,
      "grad_norm": 0.21669217944145203,
      "learning_rate": 3.0380560772966483e-06,
      "loss": 0.0252,
      "step": 50120
    },
    {
      "epoch": 2.5445454083963654,
      "grad_norm": 0.429474413394928,
      "learning_rate": 3.0363639440242314e-06,
      "loss": 0.0327,
      "step": 50125
    },
    {
      "epoch": 2.5447992283872276,
      "grad_norm": 0.27436283230781555,
      "learning_rate": 3.0346718107518154e-06,
      "loss": 0.0335,
      "step": 50130
    },
    {
      "epoch": 2.54505304837809,
      "grad_norm": 0.22250846028327942,
      "learning_rate": 3.0329796774793985e-06,
      "loss": 0.0285,
      "step": 50135
    },
    {
      "epoch": 2.545306868368953,
      "grad_norm": 0.33518174290657043,
      "learning_rate": 3.031287544206982e-06,
      "loss": 0.0315,
      "step": 50140
    },
    {
      "epoch": 2.5455606883598154,
      "grad_norm": 0.24940243363380432,
      "learning_rate": 3.0295954109345656e-06,
      "loss": 0.0306,
      "step": 50145
    },
    {
      "epoch": 2.5458145083506776,
      "grad_norm": 0.44675156474113464,
      "learning_rate": 3.0279032776621487e-06,
      "loss": 0.0356,
      "step": 50150
    },
    {
      "epoch": 2.54606832834154,
      "grad_norm": 0.28247472643852234,
      "learning_rate": 3.0262111443897323e-06,
      "loss": 0.0397,
      "step": 50155
    },
    {
      "epoch": 2.546322148332403,
      "grad_norm": 0.3223744034767151,
      "learning_rate": 3.024519011117316e-06,
      "loss": 0.0289,
      "step": 50160
    },
    {
      "epoch": 2.546575968323265,
      "grad_norm": 0.3707750141620636,
      "learning_rate": 3.0228268778448993e-06,
      "loss": 0.0246,
      "step": 50165
    },
    {
      "epoch": 2.5468297883141275,
      "grad_norm": 0.27658745646476746,
      "learning_rate": 3.0211347445724825e-06,
      "loss": 0.033,
      "step": 50170
    },
    {
      "epoch": 2.54708360830499,
      "grad_norm": 0.42075684666633606,
      "learning_rate": 3.0194426113000664e-06,
      "loss": 0.0345,
      "step": 50175
    },
    {
      "epoch": 2.5473374282958527,
      "grad_norm": 0.2999601364135742,
      "learning_rate": 3.0177504780276496e-06,
      "loss": 0.0333,
      "step": 50180
    },
    {
      "epoch": 2.547591248286715,
      "grad_norm": 0.3456297218799591,
      "learning_rate": 3.016058344755233e-06,
      "loss": 0.0355,
      "step": 50185
    },
    {
      "epoch": 2.5478450682775775,
      "grad_norm": 0.24620822072029114,
      "learning_rate": 3.0143662114828166e-06,
      "loss": 0.027,
      "step": 50190
    },
    {
      "epoch": 2.54809888826844,
      "grad_norm": 0.4098372757434845,
      "learning_rate": 3.0126740782104e-06,
      "loss": 0.029,
      "step": 50195
    },
    {
      "epoch": 2.5483527082593023,
      "grad_norm": 0.2857387065887451,
      "learning_rate": 3.0109819449379833e-06,
      "loss": 0.0321,
      "step": 50200
    },
    {
      "epoch": 2.548606528250165,
      "grad_norm": 0.345076322555542,
      "learning_rate": 3.0092898116655673e-06,
      "loss": 0.0363,
      "step": 50205
    },
    {
      "epoch": 2.5488603482410275,
      "grad_norm": 0.3622076213359833,
      "learning_rate": 3.0075976783931504e-06,
      "loss": 0.0325,
      "step": 50210
    },
    {
      "epoch": 2.54911416823189,
      "grad_norm": 0.24855151772499084,
      "learning_rate": 3.0059055451207344e-06,
      "loss": 0.0286,
      "step": 50215
    },
    {
      "epoch": 2.5493679882227527,
      "grad_norm": 0.4033049941062927,
      "learning_rate": 3.0042134118483175e-06,
      "loss": 0.0244,
      "step": 50220
    },
    {
      "epoch": 2.549621808213615,
      "grad_norm": 0.16056959331035614,
      "learning_rate": 3.002521278575901e-06,
      "loss": 0.0292,
      "step": 50225
    },
    {
      "epoch": 2.5498756282044774,
      "grad_norm": 0.3058248460292816,
      "learning_rate": 3.000829145303484e-06,
      "loss": 0.0342,
      "step": 50230
    },
    {
      "epoch": 2.5501294481953396,
      "grad_norm": 0.5386940836906433,
      "learning_rate": 2.9991370120310677e-06,
      "loss": 0.0327,
      "step": 50235
    },
    {
      "epoch": 2.550383268186202,
      "grad_norm": 0.2744958698749542,
      "learning_rate": 2.9974448787586512e-06,
      "loss": 0.0252,
      "step": 50240
    },
    {
      "epoch": 2.550637088177065,
      "grad_norm": 0.325539767742157,
      "learning_rate": 2.9957527454862344e-06,
      "loss": 0.0305,
      "step": 50245
    },
    {
      "epoch": 2.5508909081679274,
      "grad_norm": 0.3471980392932892,
      "learning_rate": 2.9940606122138183e-06,
      "loss": 0.0335,
      "step": 50250
    },
    {
      "epoch": 2.55114472815879,
      "grad_norm": 0.5069236755371094,
      "learning_rate": 2.9923684789414015e-06,
      "loss": 0.0347,
      "step": 50255
    },
    {
      "epoch": 2.551398548149652,
      "grad_norm": 0.34677132964134216,
      "learning_rate": 2.9906763456689854e-06,
      "loss": 0.0312,
      "step": 50260
    },
    {
      "epoch": 2.5516523681405148,
      "grad_norm": 0.6088144183158875,
      "learning_rate": 2.9889842123965685e-06,
      "loss": 0.0342,
      "step": 50265
    },
    {
      "epoch": 2.5519061881313774,
      "grad_norm": 0.27897128462791443,
      "learning_rate": 2.987292079124152e-06,
      "loss": 0.0325,
      "step": 50270
    },
    {
      "epoch": 2.5521600081222395,
      "grad_norm": 0.2830289304256439,
      "learning_rate": 2.985599945851735e-06,
      "loss": 0.0338,
      "step": 50275
    },
    {
      "epoch": 2.552413828113102,
      "grad_norm": 0.3052821457386017,
      "learning_rate": 2.983907812579319e-06,
      "loss": 0.0302,
      "step": 50280
    },
    {
      "epoch": 2.5526676481039647,
      "grad_norm": 0.3779296576976776,
      "learning_rate": 2.9822156793069023e-06,
      "loss": 0.0309,
      "step": 50285
    },
    {
      "epoch": 2.5529214680948273,
      "grad_norm": 0.343472957611084,
      "learning_rate": 2.9805235460344863e-06,
      "loss": 0.0321,
      "step": 50290
    },
    {
      "epoch": 2.5531752880856895,
      "grad_norm": 0.299538254737854,
      "learning_rate": 2.9788314127620694e-06,
      "loss": 0.0349,
      "step": 50295
    },
    {
      "epoch": 2.553429108076552,
      "grad_norm": 0.24920311570167542,
      "learning_rate": 2.977139279489653e-06,
      "loss": 0.0271,
      "step": 50300
    },
    {
      "epoch": 2.5536829280674147,
      "grad_norm": 0.4240674376487732,
      "learning_rate": 2.9754471462172365e-06,
      "loss": 0.0382,
      "step": 50305
    },
    {
      "epoch": 2.553936748058277,
      "grad_norm": 0.17833265662193298,
      "learning_rate": 2.97375501294482e-06,
      "loss": 0.0297,
      "step": 50310
    },
    {
      "epoch": 2.5541905680491395,
      "grad_norm": 0.3323668837547302,
      "learning_rate": 2.972062879672403e-06,
      "loss": 0.029,
      "step": 50315
    },
    {
      "epoch": 2.554444388040002,
      "grad_norm": 0.23749475181102753,
      "learning_rate": 2.9703707463999863e-06,
      "loss": 0.0328,
      "step": 50320
    },
    {
      "epoch": 2.5546982080308647,
      "grad_norm": 0.25050121545791626,
      "learning_rate": 2.9686786131275702e-06,
      "loss": 0.0266,
      "step": 50325
    },
    {
      "epoch": 2.554952028021727,
      "grad_norm": 0.2653757333755493,
      "learning_rate": 2.9669864798551534e-06,
      "loss": 0.0302,
      "step": 50330
    },
    {
      "epoch": 2.5552058480125894,
      "grad_norm": 0.4240018427371979,
      "learning_rate": 2.9652943465827373e-06,
      "loss": 0.0373,
      "step": 50335
    },
    {
      "epoch": 2.555459668003452,
      "grad_norm": 0.2510708272457123,
      "learning_rate": 2.9636022133103204e-06,
      "loss": 0.028,
      "step": 50340
    },
    {
      "epoch": 2.555713487994314,
      "grad_norm": 0.4198581874370575,
      "learning_rate": 2.961910080037904e-06,
      "loss": 0.0411,
      "step": 50345
    },
    {
      "epoch": 2.555967307985177,
      "grad_norm": 0.40445321798324585,
      "learning_rate": 2.9602179467654875e-06,
      "loss": 0.0378,
      "step": 50350
    },
    {
      "epoch": 2.5562211279760394,
      "grad_norm": 0.3391188681125641,
      "learning_rate": 2.958525813493071e-06,
      "loss": 0.041,
      "step": 50355
    },
    {
      "epoch": 2.556474947966902,
      "grad_norm": 0.4442570209503174,
      "learning_rate": 2.956833680220654e-06,
      "loss": 0.0288,
      "step": 50360
    },
    {
      "epoch": 2.5567287679577646,
      "grad_norm": 0.3415834605693817,
      "learning_rate": 2.955141546948238e-06,
      "loss": 0.0284,
      "step": 50365
    },
    {
      "epoch": 2.5569825879486268,
      "grad_norm": 0.35875076055526733,
      "learning_rate": 2.9534494136758213e-06,
      "loss": 0.0305,
      "step": 50370
    },
    {
      "epoch": 2.5572364079394894,
      "grad_norm": 0.26490122079849243,
      "learning_rate": 2.951757280403405e-06,
      "loss": 0.0333,
      "step": 50375
    },
    {
      "epoch": 2.5574902279303515,
      "grad_norm": 0.31595557928085327,
      "learning_rate": 2.9500651471309884e-06,
      "loss": 0.0283,
      "step": 50380
    },
    {
      "epoch": 2.557744047921214,
      "grad_norm": 0.36508527398109436,
      "learning_rate": 2.948373013858572e-06,
      "loss": 0.0388,
      "step": 50385
    },
    {
      "epoch": 2.5579978679120767,
      "grad_norm": 0.2791566848754883,
      "learning_rate": 2.946680880586155e-06,
      "loss": 0.0299,
      "step": 50390
    },
    {
      "epoch": 2.5582516879029393,
      "grad_norm": 0.2070358693599701,
      "learning_rate": 2.9449887473137386e-06,
      "loss": 0.0326,
      "step": 50395
    },
    {
      "epoch": 2.558505507893802,
      "grad_norm": 0.2233690321445465,
      "learning_rate": 2.943296614041322e-06,
      "loss": 0.0283,
      "step": 50400
    },
    {
      "epoch": 2.558759327884664,
      "grad_norm": 0.3481098711490631,
      "learning_rate": 2.9416044807689053e-06,
      "loss": 0.0272,
      "step": 50405
    },
    {
      "epoch": 2.5590131478755267,
      "grad_norm": 0.41446927189826965,
      "learning_rate": 2.9399123474964892e-06,
      "loss": 0.0264,
      "step": 50410
    },
    {
      "epoch": 2.5592669678663893,
      "grad_norm": 0.20012246072292328,
      "learning_rate": 2.9382202142240723e-06,
      "loss": 0.0297,
      "step": 50415
    },
    {
      "epoch": 2.5595207878572515,
      "grad_norm": 0.6422848701477051,
      "learning_rate": 2.936528080951656e-06,
      "loss": 0.0281,
      "step": 50420
    },
    {
      "epoch": 2.559774607848114,
      "grad_norm": 0.3305734694004059,
      "learning_rate": 2.9348359476792394e-06,
      "loss": 0.0264,
      "step": 50425
    },
    {
      "epoch": 2.5600284278389767,
      "grad_norm": 0.25522372126579285,
      "learning_rate": 2.933143814406823e-06,
      "loss": 0.0337,
      "step": 50430
    },
    {
      "epoch": 2.5602822478298393,
      "grad_norm": 0.19020916521549225,
      "learning_rate": 2.931451681134406e-06,
      "loss": 0.0351,
      "step": 50435
    },
    {
      "epoch": 2.5605360678207014,
      "grad_norm": 0.3231838643550873,
      "learning_rate": 2.92975954786199e-06,
      "loss": 0.028,
      "step": 50440
    },
    {
      "epoch": 2.560789887811564,
      "grad_norm": 0.3561446964740753,
      "learning_rate": 2.928067414589573e-06,
      "loss": 0.0377,
      "step": 50445
    },
    {
      "epoch": 2.5610437078024266,
      "grad_norm": 0.25820067524909973,
      "learning_rate": 2.926375281317157e-06,
      "loss": 0.029,
      "step": 50450
    },
    {
      "epoch": 2.561297527793289,
      "grad_norm": 0.2987731397151947,
      "learning_rate": 2.9246831480447403e-06,
      "loss": 0.0316,
      "step": 50455
    },
    {
      "epoch": 2.5615513477841514,
      "grad_norm": 0.39317697286605835,
      "learning_rate": 2.922991014772324e-06,
      "loss": 0.0377,
      "step": 50460
    },
    {
      "epoch": 2.561805167775014,
      "grad_norm": 0.26665055751800537,
      "learning_rate": 2.921298881499907e-06,
      "loss": 0.0391,
      "step": 50465
    },
    {
      "epoch": 2.5620589877658766,
      "grad_norm": 0.4422428011894226,
      "learning_rate": 2.919606748227491e-06,
      "loss": 0.0285,
      "step": 50470
    },
    {
      "epoch": 2.562312807756739,
      "grad_norm": 1.0131126642227173,
      "learning_rate": 2.917914614955074e-06,
      "loss": 0.0384,
      "step": 50475
    },
    {
      "epoch": 2.5625666277476014,
      "grad_norm": 0.20424379408359528,
      "learning_rate": 2.916222481682657e-06,
      "loss": 0.0305,
      "step": 50480
    },
    {
      "epoch": 2.562820447738464,
      "grad_norm": 0.8846186399459839,
      "learning_rate": 2.914530348410241e-06,
      "loss": 0.0286,
      "step": 50485
    },
    {
      "epoch": 2.563074267729326,
      "grad_norm": 0.5644127726554871,
      "learning_rate": 2.9128382151378242e-06,
      "loss": 0.0321,
      "step": 50490
    },
    {
      "epoch": 2.5633280877201887,
      "grad_norm": 0.24908645451068878,
      "learning_rate": 2.911146081865408e-06,
      "loss": 0.0316,
      "step": 50495
    },
    {
      "epoch": 2.5635819077110513,
      "grad_norm": 0.2304484248161316,
      "learning_rate": 2.9094539485929913e-06,
      "loss": 0.0278,
      "step": 50500
    },
    {
      "epoch": 2.563835727701914,
      "grad_norm": 0.3092939257621765,
      "learning_rate": 2.907761815320575e-06,
      "loss": 0.0268,
      "step": 50505
    },
    {
      "epoch": 2.5640895476927765,
      "grad_norm": 0.18987365067005157,
      "learning_rate": 2.906069682048158e-06,
      "loss": 0.0242,
      "step": 50510
    },
    {
      "epoch": 2.5643433676836387,
      "grad_norm": 0.3280569016933441,
      "learning_rate": 2.904377548775742e-06,
      "loss": 0.0314,
      "step": 50515
    },
    {
      "epoch": 2.5645971876745013,
      "grad_norm": 0.26012980937957764,
      "learning_rate": 2.902685415503325e-06,
      "loss": 0.0333,
      "step": 50520
    },
    {
      "epoch": 2.5648510076653634,
      "grad_norm": 0.34960871934890747,
      "learning_rate": 2.900993282230909e-06,
      "loss": 0.0284,
      "step": 50525
    },
    {
      "epoch": 2.565104827656226,
      "grad_norm": 0.3368133306503296,
      "learning_rate": 2.899301148958492e-06,
      "loss": 0.0401,
      "step": 50530
    },
    {
      "epoch": 2.5653586476470887,
      "grad_norm": 0.24301111698150635,
      "learning_rate": 2.8976090156860757e-06,
      "loss": 0.0288,
      "step": 50535
    },
    {
      "epoch": 2.5656124676379513,
      "grad_norm": 0.33640941977500916,
      "learning_rate": 2.8959168824136593e-06,
      "loss": 0.029,
      "step": 50540
    },
    {
      "epoch": 2.565866287628814,
      "grad_norm": 0.2389553189277649,
      "learning_rate": 2.894224749141243e-06,
      "loss": 0.0349,
      "step": 50545
    },
    {
      "epoch": 2.566120107619676,
      "grad_norm": 0.34752050042152405,
      "learning_rate": 2.892532615868826e-06,
      "loss": 0.0256,
      "step": 50550
    },
    {
      "epoch": 2.5663739276105386,
      "grad_norm": 0.48742684721946716,
      "learning_rate": 2.8908404825964095e-06,
      "loss": 0.0262,
      "step": 50555
    },
    {
      "epoch": 2.566627747601401,
      "grad_norm": 0.2560529410839081,
      "learning_rate": 2.889148349323993e-06,
      "loss": 0.0298,
      "step": 50560
    },
    {
      "epoch": 2.5668815675922634,
      "grad_norm": 0.3415350317955017,
      "learning_rate": 2.887456216051576e-06,
      "loss": 0.029,
      "step": 50565
    },
    {
      "epoch": 2.567135387583126,
      "grad_norm": 0.26284313201904297,
      "learning_rate": 2.88576408277916e-06,
      "loss": 0.0316,
      "step": 50570
    },
    {
      "epoch": 2.5673892075739886,
      "grad_norm": 0.29155880212783813,
      "learning_rate": 2.8840719495067432e-06,
      "loss": 0.0358,
      "step": 50575
    },
    {
      "epoch": 2.567643027564851,
      "grad_norm": 0.3447701334953308,
      "learning_rate": 2.8823798162343268e-06,
      "loss": 0.0312,
      "step": 50580
    },
    {
      "epoch": 2.5678968475557133,
      "grad_norm": 0.3020358085632324,
      "learning_rate": 2.8806876829619103e-06,
      "loss": 0.0259,
      "step": 50585
    },
    {
      "epoch": 2.568150667546576,
      "grad_norm": 0.6446263790130615,
      "learning_rate": 2.878995549689494e-06,
      "loss": 0.0411,
      "step": 50590
    },
    {
      "epoch": 2.5684044875374386,
      "grad_norm": 0.6810723543167114,
      "learning_rate": 2.877303416417077e-06,
      "loss": 0.0298,
      "step": 50595
    },
    {
      "epoch": 2.5686583075283007,
      "grad_norm": 0.2936282157897949,
      "learning_rate": 2.875611283144661e-06,
      "loss": 0.0311,
      "step": 50600
    },
    {
      "epoch": 2.5689121275191633,
      "grad_norm": 0.23811903595924377,
      "learning_rate": 2.873919149872244e-06,
      "loss": 0.0336,
      "step": 50605
    },
    {
      "epoch": 2.569165947510026,
      "grad_norm": 0.29496702551841736,
      "learning_rate": 2.872227016599828e-06,
      "loss": 0.0324,
      "step": 50610
    },
    {
      "epoch": 2.5694197675008885,
      "grad_norm": 0.293457567691803,
      "learning_rate": 2.870534883327411e-06,
      "loss": 0.0335,
      "step": 50615
    },
    {
      "epoch": 2.569673587491751,
      "grad_norm": 0.36301565170288086,
      "learning_rate": 2.8688427500549947e-06,
      "loss": 0.0312,
      "step": 50620
    },
    {
      "epoch": 2.5699274074826133,
      "grad_norm": 0.32104015350341797,
      "learning_rate": 2.867150616782578e-06,
      "loss": 0.0289,
      "step": 50625
    },
    {
      "epoch": 2.570181227473476,
      "grad_norm": 0.3693762421607971,
      "learning_rate": 2.865458483510162e-06,
      "loss": 0.0374,
      "step": 50630
    },
    {
      "epoch": 2.570435047464338,
      "grad_norm": 0.4718092978000641,
      "learning_rate": 2.863766350237745e-06,
      "loss": 0.0373,
      "step": 50635
    },
    {
      "epoch": 2.5706888674552006,
      "grad_norm": 0.3577287197113037,
      "learning_rate": 2.862074216965328e-06,
      "loss": 0.0311,
      "step": 50640
    },
    {
      "epoch": 2.5709426874460632,
      "grad_norm": 0.3808220624923706,
      "learning_rate": 2.860382083692912e-06,
      "loss": 0.0357,
      "step": 50645
    },
    {
      "epoch": 2.571196507436926,
      "grad_norm": 0.41456541419029236,
      "learning_rate": 2.858689950420495e-06,
      "loss": 0.0367,
      "step": 50650
    },
    {
      "epoch": 2.5714503274277885,
      "grad_norm": 0.33753418922424316,
      "learning_rate": 2.856997817148079e-06,
      "loss": 0.0277,
      "step": 50655
    },
    {
      "epoch": 2.5717041474186506,
      "grad_norm": 0.30118709802627563,
      "learning_rate": 2.855305683875662e-06,
      "loss": 0.0342,
      "step": 50660
    },
    {
      "epoch": 2.571957967409513,
      "grad_norm": 0.3879167437553406,
      "learning_rate": 2.8536135506032458e-06,
      "loss": 0.0329,
      "step": 50665
    },
    {
      "epoch": 2.572211787400376,
      "grad_norm": 0.3201780617237091,
      "learning_rate": 2.851921417330829e-06,
      "loss": 0.0292,
      "step": 50670
    },
    {
      "epoch": 2.572465607391238,
      "grad_norm": 0.28086307644844055,
      "learning_rate": 2.850229284058413e-06,
      "loss": 0.0301,
      "step": 50675
    },
    {
      "epoch": 2.5727194273821006,
      "grad_norm": 0.8992939591407776,
      "learning_rate": 2.848537150785996e-06,
      "loss": 0.0234,
      "step": 50680
    },
    {
      "epoch": 2.572973247372963,
      "grad_norm": 0.3172586262226105,
      "learning_rate": 2.84684501751358e-06,
      "loss": 0.0268,
      "step": 50685
    },
    {
      "epoch": 2.573227067363826,
      "grad_norm": 0.25958219170570374,
      "learning_rate": 2.845152884241163e-06,
      "loss": 0.026,
      "step": 50690
    },
    {
      "epoch": 2.573480887354688,
      "grad_norm": 0.22440466284751892,
      "learning_rate": 2.8434607509687466e-06,
      "loss": 0.0292,
      "step": 50695
    },
    {
      "epoch": 2.5737347073455505,
      "grad_norm": 0.34167757630348206,
      "learning_rate": 2.84176861769633e-06,
      "loss": 0.031,
      "step": 50700
    },
    {
      "epoch": 2.573988527336413,
      "grad_norm": 0.3714410960674286,
      "learning_rate": 2.8400764844239137e-06,
      "loss": 0.0321,
      "step": 50705
    },
    {
      "epoch": 2.5742423473272753,
      "grad_norm": 0.38244038820266724,
      "learning_rate": 2.838384351151497e-06,
      "loss": 0.0268,
      "step": 50710
    },
    {
      "epoch": 2.574496167318138,
      "grad_norm": 0.42816251516342163,
      "learning_rate": 2.83669221787908e-06,
      "loss": 0.0337,
      "step": 50715
    },
    {
      "epoch": 2.5747499873090005,
      "grad_norm": 0.27166709303855896,
      "learning_rate": 2.835000084606664e-06,
      "loss": 0.0317,
      "step": 50720
    },
    {
      "epoch": 2.575003807299863,
      "grad_norm": 0.2813893258571625,
      "learning_rate": 2.833307951334247e-06,
      "loss": 0.0301,
      "step": 50725
    },
    {
      "epoch": 2.5752576272907253,
      "grad_norm": 0.6570308208465576,
      "learning_rate": 2.831615818061831e-06,
      "loss": 0.038,
      "step": 50730
    },
    {
      "epoch": 2.575511447281588,
      "grad_norm": 0.31474921107292175,
      "learning_rate": 2.829923684789414e-06,
      "loss": 0.0252,
      "step": 50735
    },
    {
      "epoch": 2.5757652672724505,
      "grad_norm": 0.21304818987846375,
      "learning_rate": 2.8282315515169977e-06,
      "loss": 0.0347,
      "step": 50740
    },
    {
      "epoch": 2.5760190872633126,
      "grad_norm": 0.5252552032470703,
      "learning_rate": 2.826539418244581e-06,
      "loss": 0.0312,
      "step": 50745
    },
    {
      "epoch": 2.5762729072541752,
      "grad_norm": 0.355497807264328,
      "learning_rate": 2.8248472849721647e-06,
      "loss": 0.026,
      "step": 50750
    },
    {
      "epoch": 2.576526727245038,
      "grad_norm": 0.2816254496574402,
      "learning_rate": 2.823155151699748e-06,
      "loss": 0.0286,
      "step": 50755
    },
    {
      "epoch": 2.5767805472359004,
      "grad_norm": 0.35320496559143066,
      "learning_rate": 2.821463018427332e-06,
      "loss": 0.0251,
      "step": 50760
    },
    {
      "epoch": 2.577034367226763,
      "grad_norm": 0.36286720633506775,
      "learning_rate": 2.819770885154915e-06,
      "loss": 0.0306,
      "step": 50765
    },
    {
      "epoch": 2.577288187217625,
      "grad_norm": 0.25672924518585205,
      "learning_rate": 2.8180787518824985e-06,
      "loss": 0.0244,
      "step": 50770
    },
    {
      "epoch": 2.577542007208488,
      "grad_norm": 0.2837968170642853,
      "learning_rate": 2.816386618610082e-06,
      "loss": 0.0316,
      "step": 50775
    },
    {
      "epoch": 2.57779582719935,
      "grad_norm": 0.24833479523658752,
      "learning_rate": 2.8146944853376656e-06,
      "loss": 0.0274,
      "step": 50780
    },
    {
      "epoch": 2.5780496471902126,
      "grad_norm": 0.3615385591983795,
      "learning_rate": 2.8130023520652487e-06,
      "loss": 0.0261,
      "step": 50785
    },
    {
      "epoch": 2.578303467181075,
      "grad_norm": 0.24618951976299286,
      "learning_rate": 2.8113102187928327e-06,
      "loss": 0.0318,
      "step": 50790
    },
    {
      "epoch": 2.5785572871719378,
      "grad_norm": 0.3765259385108948,
      "learning_rate": 2.809618085520416e-06,
      "loss": 0.0413,
      "step": 50795
    },
    {
      "epoch": 2.5788111071628004,
      "grad_norm": 0.27768805623054504,
      "learning_rate": 2.807925952247999e-06,
      "loss": 0.0261,
      "step": 50800
    },
    {
      "epoch": 2.5790649271536625,
      "grad_norm": 0.3444536030292511,
      "learning_rate": 2.806233818975583e-06,
      "loss": 0.0369,
      "step": 50805
    },
    {
      "epoch": 2.579318747144525,
      "grad_norm": 0.24263224005699158,
      "learning_rate": 2.804541685703166e-06,
      "loss": 0.0336,
      "step": 50810
    },
    {
      "epoch": 2.5795725671353877,
      "grad_norm": 0.33877697587013245,
      "learning_rate": 2.8028495524307496e-06,
      "loss": 0.032,
      "step": 50815
    },
    {
      "epoch": 2.57982638712625,
      "grad_norm": 0.2877173125743866,
      "learning_rate": 2.801157419158333e-06,
      "loss": 0.0345,
      "step": 50820
    },
    {
      "epoch": 2.5800802071171125,
      "grad_norm": 0.3200043737888336,
      "learning_rate": 2.7994652858859166e-06,
      "loss": 0.0354,
      "step": 50825
    },
    {
      "epoch": 2.580334027107975,
      "grad_norm": 0.29038625955581665,
      "learning_rate": 2.7977731526134998e-06,
      "loss": 0.0324,
      "step": 50830
    },
    {
      "epoch": 2.5805878470988377,
      "grad_norm": 0.24091961979866028,
      "learning_rate": 2.7960810193410837e-06,
      "loss": 0.0297,
      "step": 50835
    },
    {
      "epoch": 2.5808416670897,
      "grad_norm": 0.3212089538574219,
      "learning_rate": 2.794388886068667e-06,
      "loss": 0.0308,
      "step": 50840
    },
    {
      "epoch": 2.5810954870805625,
      "grad_norm": 0.33707278966903687,
      "learning_rate": 2.792696752796251e-06,
      "loss": 0.0366,
      "step": 50845
    },
    {
      "epoch": 2.581349307071425,
      "grad_norm": 0.6692639589309692,
      "learning_rate": 2.791004619523834e-06,
      "loss": 0.0304,
      "step": 50850
    },
    {
      "epoch": 2.5816031270622872,
      "grad_norm": 0.46995946764945984,
      "learning_rate": 2.7893124862514175e-06,
      "loss": 0.0363,
      "step": 50855
    },
    {
      "epoch": 2.58185694705315,
      "grad_norm": 0.2652070224285126,
      "learning_rate": 2.7876203529790006e-06,
      "loss": 0.0245,
      "step": 50860
    },
    {
      "epoch": 2.5821107670440124,
      "grad_norm": 0.24837477505207062,
      "learning_rate": 2.7859282197065846e-06,
      "loss": 0.0295,
      "step": 50865
    },
    {
      "epoch": 2.582364587034875,
      "grad_norm": 0.3100050389766693,
      "learning_rate": 2.7842360864341677e-06,
      "loss": 0.0354,
      "step": 50870
    },
    {
      "epoch": 2.582618407025737,
      "grad_norm": 0.22369235754013062,
      "learning_rate": 2.7825439531617517e-06,
      "loss": 0.0359,
      "step": 50875
    },
    {
      "epoch": 2.5828722270166,
      "grad_norm": 0.32840296626091003,
      "learning_rate": 2.7808518198893348e-06,
      "loss": 0.0281,
      "step": 50880
    },
    {
      "epoch": 2.5831260470074624,
      "grad_norm": 0.2909887135028839,
      "learning_rate": 2.779159686616918e-06,
      "loss": 0.0323,
      "step": 50885
    },
    {
      "epoch": 2.5833798669983246,
      "grad_norm": 0.4277086555957794,
      "learning_rate": 2.777467553344502e-06,
      "loss": 0.0356,
      "step": 50890
    },
    {
      "epoch": 2.583633686989187,
      "grad_norm": 0.33212006092071533,
      "learning_rate": 2.775775420072085e-06,
      "loss": 0.034,
      "step": 50895
    },
    {
      "epoch": 2.5838875069800498,
      "grad_norm": 0.30871519446372986,
      "learning_rate": 2.7740832867996685e-06,
      "loss": 0.0257,
      "step": 50900
    },
    {
      "epoch": 2.5841413269709124,
      "grad_norm": 0.26937437057495117,
      "learning_rate": 2.7723911535272517e-06,
      "loss": 0.0301,
      "step": 50905
    },
    {
      "epoch": 2.584395146961775,
      "grad_norm": 0.6625112891197205,
      "learning_rate": 2.7706990202548356e-06,
      "loss": 0.0302,
      "step": 50910
    },
    {
      "epoch": 2.584648966952637,
      "grad_norm": 0.5626937747001648,
      "learning_rate": 2.7690068869824187e-06,
      "loss": 0.0266,
      "step": 50915
    },
    {
      "epoch": 2.5849027869434997,
      "grad_norm": 0.29541710019111633,
      "learning_rate": 2.7673147537100027e-06,
      "loss": 0.0308,
      "step": 50920
    },
    {
      "epoch": 2.585156606934362,
      "grad_norm": 0.31842437386512756,
      "learning_rate": 2.765622620437586e-06,
      "loss": 0.0303,
      "step": 50925
    },
    {
      "epoch": 2.5854104269252245,
      "grad_norm": 0.35333383083343506,
      "learning_rate": 2.7639304871651694e-06,
      "loss": 0.0245,
      "step": 50930
    },
    {
      "epoch": 2.585664246916087,
      "grad_norm": 0.2786847949028015,
      "learning_rate": 2.762238353892753e-06,
      "loss": 0.0235,
      "step": 50935
    },
    {
      "epoch": 2.5859180669069497,
      "grad_norm": 0.36661455035209656,
      "learning_rate": 2.7605462206203365e-06,
      "loss": 0.0327,
      "step": 50940
    },
    {
      "epoch": 2.5861718868978123,
      "grad_norm": 0.42655545473098755,
      "learning_rate": 2.7588540873479196e-06,
      "loss": 0.0285,
      "step": 50945
    },
    {
      "epoch": 2.5864257068886745,
      "grad_norm": 0.3568045198917389,
      "learning_rate": 2.7571619540755036e-06,
      "loss": 0.0232,
      "step": 50950
    },
    {
      "epoch": 2.586679526879537,
      "grad_norm": 0.34968164563179016,
      "learning_rate": 2.7554698208030867e-06,
      "loss": 0.0268,
      "step": 50955
    },
    {
      "epoch": 2.5869333468703997,
      "grad_norm": 0.26196905970573425,
      "learning_rate": 2.75377768753067e-06,
      "loss": 0.0259,
      "step": 50960
    },
    {
      "epoch": 2.587187166861262,
      "grad_norm": 0.3904362618923187,
      "learning_rate": 2.7520855542582538e-06,
      "loss": 0.0343,
      "step": 50965
    },
    {
      "epoch": 2.5874409868521244,
      "grad_norm": 0.3048829138278961,
      "learning_rate": 2.750393420985837e-06,
      "loss": 0.028,
      "step": 50970
    },
    {
      "epoch": 2.587694806842987,
      "grad_norm": 0.365691214799881,
      "learning_rate": 2.7487012877134204e-06,
      "loss": 0.0261,
      "step": 50975
    },
    {
      "epoch": 2.5879486268338496,
      "grad_norm": 0.37310320138931274,
      "learning_rate": 2.747009154441004e-06,
      "loss": 0.0307,
      "step": 50980
    },
    {
      "epoch": 2.588202446824712,
      "grad_norm": 0.2797809839248657,
      "learning_rate": 2.7453170211685875e-06,
      "loss": 0.0265,
      "step": 50985
    },
    {
      "epoch": 2.5884562668155744,
      "grad_norm": 1.3563828468322754,
      "learning_rate": 2.7436248878961706e-06,
      "loss": 0.0269,
      "step": 50990
    },
    {
      "epoch": 2.588710086806437,
      "grad_norm": 0.2800024747848511,
      "learning_rate": 2.7419327546237546e-06,
      "loss": 0.0282,
      "step": 50995
    },
    {
      "epoch": 2.588963906797299,
      "grad_norm": 0.4817359745502472,
      "learning_rate": 2.7402406213513377e-06,
      "loss": 0.0353,
      "step": 51000
    },
    {
      "epoch": 2.5892177267881618,
      "grad_norm": 0.28154054284095764,
      "learning_rate": 2.7385484880789213e-06,
      "loss": 0.0306,
      "step": 51005
    },
    {
      "epoch": 2.5894715467790244,
      "grad_norm": 1.3750730752944946,
      "learning_rate": 2.736856354806505e-06,
      "loss": 0.0247,
      "step": 51010
    },
    {
      "epoch": 2.589725366769887,
      "grad_norm": 0.4572613537311554,
      "learning_rate": 2.7351642215340884e-06,
      "loss": 0.0349,
      "step": 51015
    },
    {
      "epoch": 2.589979186760749,
      "grad_norm": 0.28127723932266235,
      "learning_rate": 2.7334720882616715e-06,
      "loss": 0.0331,
      "step": 51020
    },
    {
      "epoch": 2.5902330067516117,
      "grad_norm": 0.3305010497570038,
      "learning_rate": 2.7317799549892555e-06,
      "loss": 0.0323,
      "step": 51025
    },
    {
      "epoch": 2.5904868267424743,
      "grad_norm": 0.3264330327510834,
      "learning_rate": 2.7300878217168386e-06,
      "loss": 0.0311,
      "step": 51030
    },
    {
      "epoch": 2.5907406467333365,
      "grad_norm": 0.351087749004364,
      "learning_rate": 2.7283956884444225e-06,
      "loss": 0.0298,
      "step": 51035
    },
    {
      "epoch": 2.590994466724199,
      "grad_norm": 0.3994324505329132,
      "learning_rate": 2.7267035551720057e-06,
      "loss": 0.0315,
      "step": 51040
    },
    {
      "epoch": 2.5912482867150617,
      "grad_norm": 0.31754663586616516,
      "learning_rate": 2.7250114218995888e-06,
      "loss": 0.0363,
      "step": 51045
    },
    {
      "epoch": 2.5915021067059243,
      "grad_norm": 0.2738948464393616,
      "learning_rate": 2.7233192886271723e-06,
      "loss": 0.0315,
      "step": 51050
    },
    {
      "epoch": 2.591755926696787,
      "grad_norm": 0.2503436803817749,
      "learning_rate": 2.721627155354756e-06,
      "loss": 0.03,
      "step": 51055
    },
    {
      "epoch": 2.592009746687649,
      "grad_norm": 0.32593145966529846,
      "learning_rate": 2.7199350220823394e-06,
      "loss": 0.0317,
      "step": 51060
    },
    {
      "epoch": 2.5922635666785117,
      "grad_norm": 0.23604324460029602,
      "learning_rate": 2.7182428888099225e-06,
      "loss": 0.0271,
      "step": 51065
    },
    {
      "epoch": 2.592517386669374,
      "grad_norm": 0.2927345335483551,
      "learning_rate": 2.7165507555375065e-06,
      "loss": 0.0335,
      "step": 51070
    },
    {
      "epoch": 2.5927712066602364,
      "grad_norm": 0.27732741832733154,
      "learning_rate": 2.7148586222650896e-06,
      "loss": 0.0304,
      "step": 51075
    },
    {
      "epoch": 2.593025026651099,
      "grad_norm": 0.28966188430786133,
      "learning_rate": 2.7131664889926736e-06,
      "loss": 0.0279,
      "step": 51080
    },
    {
      "epoch": 2.5932788466419616,
      "grad_norm": 0.4134257435798645,
      "learning_rate": 2.7114743557202567e-06,
      "loss": 0.0316,
      "step": 51085
    },
    {
      "epoch": 2.5935326666328242,
      "grad_norm": 0.8288419246673584,
      "learning_rate": 2.7097822224478403e-06,
      "loss": 0.0348,
      "step": 51090
    },
    {
      "epoch": 2.5937864866236864,
      "grad_norm": 0.45848527550697327,
      "learning_rate": 2.708090089175424e-06,
      "loss": 0.0364,
      "step": 51095
    },
    {
      "epoch": 2.594040306614549,
      "grad_norm": 0.3152906000614166,
      "learning_rate": 2.7063979559030074e-06,
      "loss": 0.029,
      "step": 51100
    },
    {
      "epoch": 2.5942941266054116,
      "grad_norm": 0.31780606508255005,
      "learning_rate": 2.7047058226305905e-06,
      "loss": 0.0339,
      "step": 51105
    },
    {
      "epoch": 2.5945479465962737,
      "grad_norm": 0.3507641851902008,
      "learning_rate": 2.7030136893581744e-06,
      "loss": 0.0339,
      "step": 51110
    },
    {
      "epoch": 2.5948017665871363,
      "grad_norm": 0.2539495527744293,
      "learning_rate": 2.7013215560857576e-06,
      "loss": 0.0257,
      "step": 51115
    },
    {
      "epoch": 2.595055586577999,
      "grad_norm": 0.3520859479904175,
      "learning_rate": 2.6996294228133407e-06,
      "loss": 0.032,
      "step": 51120
    },
    {
      "epoch": 2.5953094065688616,
      "grad_norm": 0.27110981941223145,
      "learning_rate": 2.6979372895409247e-06,
      "loss": 0.0285,
      "step": 51125
    },
    {
      "epoch": 2.5955632265597237,
      "grad_norm": 0.3485463857650757,
      "learning_rate": 2.6962451562685078e-06,
      "loss": 0.0292,
      "step": 51130
    },
    {
      "epoch": 2.5958170465505863,
      "grad_norm": 0.5310997366905212,
      "learning_rate": 2.6945530229960913e-06,
      "loss": 0.039,
      "step": 51135
    },
    {
      "epoch": 2.596070866541449,
      "grad_norm": 0.2819546163082123,
      "learning_rate": 2.692860889723675e-06,
      "loss": 0.0298,
      "step": 51140
    },
    {
      "epoch": 2.596324686532311,
      "grad_norm": 0.5256403088569641,
      "learning_rate": 2.6911687564512584e-06,
      "loss": 0.0371,
      "step": 51145
    },
    {
      "epoch": 2.5965785065231737,
      "grad_norm": 0.21236854791641235,
      "learning_rate": 2.6894766231788415e-06,
      "loss": 0.0273,
      "step": 51150
    },
    {
      "epoch": 2.5968323265140363,
      "grad_norm": 0.32847344875335693,
      "learning_rate": 2.6877844899064255e-06,
      "loss": 0.0307,
      "step": 51155
    },
    {
      "epoch": 2.597086146504899,
      "grad_norm": 0.3638588488101959,
      "learning_rate": 2.6860923566340086e-06,
      "loss": 0.035,
      "step": 51160
    },
    {
      "epoch": 2.597339966495761,
      "grad_norm": 0.584497332572937,
      "learning_rate": 2.684400223361592e-06,
      "loss": 0.0305,
      "step": 51165
    },
    {
      "epoch": 2.5975937864866236,
      "grad_norm": 0.2996261715888977,
      "learning_rate": 2.6827080900891757e-06,
      "loss": 0.0295,
      "step": 51170
    },
    {
      "epoch": 2.5978476064774862,
      "grad_norm": 0.35180866718292236,
      "learning_rate": 2.6810159568167593e-06,
      "loss": 0.0264,
      "step": 51175
    },
    {
      "epoch": 2.5981014264683484,
      "grad_norm": 0.29140573740005493,
      "learning_rate": 2.6793238235443424e-06,
      "loss": 0.0358,
      "step": 51180
    },
    {
      "epoch": 2.598355246459211,
      "grad_norm": 0.3289799690246582,
      "learning_rate": 2.6776316902719263e-06,
      "loss": 0.0268,
      "step": 51185
    },
    {
      "epoch": 2.5986090664500736,
      "grad_norm": 0.3575092554092407,
      "learning_rate": 2.6759395569995095e-06,
      "loss": 0.0291,
      "step": 51190
    },
    {
      "epoch": 2.598862886440936,
      "grad_norm": 0.3847915232181549,
      "learning_rate": 2.6742474237270934e-06,
      "loss": 0.0379,
      "step": 51195
    },
    {
      "epoch": 2.599116706431799,
      "grad_norm": 0.44137245416641235,
      "learning_rate": 2.6725552904546766e-06,
      "loss": 0.0276,
      "step": 51200
    },
    {
      "epoch": 2.599370526422661,
      "grad_norm": 0.3935795724391937,
      "learning_rate": 2.6708631571822597e-06,
      "loss": 0.0337,
      "step": 51205
    },
    {
      "epoch": 2.5996243464135236,
      "grad_norm": 0.3812408149242401,
      "learning_rate": 2.6691710239098432e-06,
      "loss": 0.0342,
      "step": 51210
    },
    {
      "epoch": 2.5998781664043857,
      "grad_norm": 0.3082849383354187,
      "learning_rate": 2.6674788906374268e-06,
      "loss": 0.0262,
      "step": 51215
    },
    {
      "epoch": 2.6001319863952483,
      "grad_norm": 0.4165632128715515,
      "learning_rate": 2.6657867573650103e-06,
      "loss": 0.0226,
      "step": 51220
    },
    {
      "epoch": 2.600385806386111,
      "grad_norm": 0.9450509548187256,
      "learning_rate": 2.6640946240925934e-06,
      "loss": 0.0395,
      "step": 51225
    },
    {
      "epoch": 2.6006396263769735,
      "grad_norm": 0.6938175559043884,
      "learning_rate": 2.6624024908201774e-06,
      "loss": 0.0318,
      "step": 51230
    },
    {
      "epoch": 2.600893446367836,
      "grad_norm": 0.23680560290813446,
      "learning_rate": 2.6607103575477605e-06,
      "loss": 0.0267,
      "step": 51235
    },
    {
      "epoch": 2.6011472663586983,
      "grad_norm": 0.25722989439964294,
      "learning_rate": 2.6590182242753445e-06,
      "loss": 0.0328,
      "step": 51240
    },
    {
      "epoch": 2.601401086349561,
      "grad_norm": 0.22755849361419678,
      "learning_rate": 2.6573260910029276e-06,
      "loss": 0.033,
      "step": 51245
    },
    {
      "epoch": 2.6016549063404235,
      "grad_norm": 0.3570180833339691,
      "learning_rate": 2.655633957730511e-06,
      "loss": 0.0372,
      "step": 51250
    },
    {
      "epoch": 2.6019087263312857,
      "grad_norm": 0.3486759662628174,
      "learning_rate": 2.6539418244580943e-06,
      "loss": 0.0295,
      "step": 51255
    },
    {
      "epoch": 2.6021625463221483,
      "grad_norm": 0.27461516857147217,
      "learning_rate": 2.6522496911856782e-06,
      "loss": 0.0279,
      "step": 51260
    },
    {
      "epoch": 2.602416366313011,
      "grad_norm": 0.24911971390247345,
      "learning_rate": 2.6505575579132614e-06,
      "loss": 0.0291,
      "step": 51265
    },
    {
      "epoch": 2.6026701863038735,
      "grad_norm": 0.23654115200042725,
      "learning_rate": 2.6488654246408453e-06,
      "loss": 0.0233,
      "step": 51270
    },
    {
      "epoch": 2.6029240062947356,
      "grad_norm": 0.4130326807498932,
      "learning_rate": 2.6471732913684284e-06,
      "loss": 0.0386,
      "step": 51275
    },
    {
      "epoch": 2.6031778262855982,
      "grad_norm": 0.19910484552383423,
      "learning_rate": 2.6454811580960116e-06,
      "loss": 0.0304,
      "step": 51280
    },
    {
      "epoch": 2.603431646276461,
      "grad_norm": 0.3008933961391449,
      "learning_rate": 2.6437890248235955e-06,
      "loss": 0.0343,
      "step": 51285
    },
    {
      "epoch": 2.603685466267323,
      "grad_norm": 0.38071808218955994,
      "learning_rate": 2.6420968915511787e-06,
      "loss": 0.0278,
      "step": 51290
    },
    {
      "epoch": 2.6039392862581856,
      "grad_norm": 0.22071151435375214,
      "learning_rate": 2.640404758278762e-06,
      "loss": 0.0262,
      "step": 51295
    },
    {
      "epoch": 2.604193106249048,
      "grad_norm": 0.4619128108024597,
      "learning_rate": 2.6387126250063453e-06,
      "loss": 0.0307,
      "step": 51300
    },
    {
      "epoch": 2.604446926239911,
      "grad_norm": 0.6991639733314514,
      "learning_rate": 2.6370204917339293e-06,
      "loss": 0.0329,
      "step": 51305
    },
    {
      "epoch": 2.6047007462307734,
      "grad_norm": 0.3269241154193878,
      "learning_rate": 2.6353283584615124e-06,
      "loss": 0.0281,
      "step": 51310
    },
    {
      "epoch": 2.6049545662216356,
      "grad_norm": 0.4293957054615021,
      "learning_rate": 2.6336362251890964e-06,
      "loss": 0.037,
      "step": 51315
    },
    {
      "epoch": 2.605208386212498,
      "grad_norm": 0.211612269282341,
      "learning_rate": 2.6319440919166795e-06,
      "loss": 0.0286,
      "step": 51320
    },
    {
      "epoch": 2.6054622062033603,
      "grad_norm": 0.2835038900375366,
      "learning_rate": 2.630251958644263e-06,
      "loss": 0.032,
      "step": 51325
    },
    {
      "epoch": 2.605716026194223,
      "grad_norm": 0.507549524307251,
      "learning_rate": 2.6285598253718466e-06,
      "loss": 0.0345,
      "step": 51330
    },
    {
      "epoch": 2.6059698461850855,
      "grad_norm": 0.4087376296520233,
      "learning_rate": 2.62686769209943e-06,
      "loss": 0.0322,
      "step": 51335
    },
    {
      "epoch": 2.606223666175948,
      "grad_norm": 0.3005754053592682,
      "learning_rate": 2.6251755588270133e-06,
      "loss": 0.0349,
      "step": 51340
    },
    {
      "epoch": 2.6064774861668107,
      "grad_norm": 0.4824678897857666,
      "learning_rate": 2.6234834255545972e-06,
      "loss": 0.0309,
      "step": 51345
    },
    {
      "epoch": 2.606731306157673,
      "grad_norm": 0.3900478780269623,
      "learning_rate": 2.6217912922821803e-06,
      "loss": 0.0324,
      "step": 51350
    },
    {
      "epoch": 2.6069851261485355,
      "grad_norm": 0.326042115688324,
      "learning_rate": 2.620099159009764e-06,
      "loss": 0.0299,
      "step": 51355
    },
    {
      "epoch": 2.6072389461393977,
      "grad_norm": 0.29876914620399475,
      "learning_rate": 2.6184070257373474e-06,
      "loss": 0.0265,
      "step": 51360
    },
    {
      "epoch": 2.6074927661302603,
      "grad_norm": 0.26470544934272766,
      "learning_rate": 2.6167148924649306e-06,
      "loss": 0.0289,
      "step": 51365
    },
    {
      "epoch": 2.607746586121123,
      "grad_norm": 0.31948786973953247,
      "learning_rate": 2.615022759192514e-06,
      "loss": 0.0273,
      "step": 51370
    },
    {
      "epoch": 2.6080004061119855,
      "grad_norm": 0.2920685112476349,
      "learning_rate": 2.6133306259200976e-06,
      "loss": 0.0326,
      "step": 51375
    },
    {
      "epoch": 2.608254226102848,
      "grad_norm": 0.4184175133705139,
      "learning_rate": 2.611638492647681e-06,
      "loss": 0.0311,
      "step": 51380
    },
    {
      "epoch": 2.6085080460937102,
      "grad_norm": 0.30750107765197754,
      "learning_rate": 2.6099463593752643e-06,
      "loss": 0.0313,
      "step": 51385
    },
    {
      "epoch": 2.608761866084573,
      "grad_norm": 0.37827202677726746,
      "learning_rate": 2.6082542261028483e-06,
      "loss": 0.0315,
      "step": 51390
    },
    {
      "epoch": 2.6090156860754354,
      "grad_norm": 0.2605781555175781,
      "learning_rate": 2.6065620928304314e-06,
      "loss": 0.0263,
      "step": 51395
    },
    {
      "epoch": 2.6092695060662976,
      "grad_norm": 0.40529435873031616,
      "learning_rate": 2.604869959558015e-06,
      "loss": 0.0262,
      "step": 51400
    },
    {
      "epoch": 2.60952332605716,
      "grad_norm": 0.4452066123485565,
      "learning_rate": 2.6031778262855985e-06,
      "loss": 0.0391,
      "step": 51405
    },
    {
      "epoch": 2.609777146048023,
      "grad_norm": 0.36278364062309265,
      "learning_rate": 2.601485693013182e-06,
      "loss": 0.0311,
      "step": 51410
    },
    {
      "epoch": 2.6100309660388854,
      "grad_norm": 0.25109902024269104,
      "learning_rate": 2.599793559740765e-06,
      "loss": 0.0296,
      "step": 51415
    },
    {
      "epoch": 2.6102847860297476,
      "grad_norm": 0.32898685336112976,
      "learning_rate": 2.598101426468349e-06,
      "loss": 0.0298,
      "step": 51420
    },
    {
      "epoch": 2.61053860602061,
      "grad_norm": 0.3027734160423279,
      "learning_rate": 2.5964092931959322e-06,
      "loss": 0.0232,
      "step": 51425
    },
    {
      "epoch": 2.6107924260114728,
      "grad_norm": 0.24649985134601593,
      "learning_rate": 2.594717159923516e-06,
      "loss": 0.0285,
      "step": 51430
    },
    {
      "epoch": 2.611046246002335,
      "grad_norm": 0.3023594319820404,
      "learning_rate": 2.5930250266510993e-06,
      "loss": 0.0275,
      "step": 51435
    },
    {
      "epoch": 2.6113000659931975,
      "grad_norm": 0.31649115681648254,
      "learning_rate": 2.591332893378683e-06,
      "loss": 0.0319,
      "step": 51440
    },
    {
      "epoch": 2.61155388598406,
      "grad_norm": 0.6469734311103821,
      "learning_rate": 2.589640760106266e-06,
      "loss": 0.0289,
      "step": 51445
    },
    {
      "epoch": 2.6118077059749227,
      "grad_norm": 0.415568470954895,
      "learning_rate": 2.5879486268338495e-06,
      "loss": 0.0256,
      "step": 51450
    },
    {
      "epoch": 2.6120615259657853,
      "grad_norm": 0.3895602822303772,
      "learning_rate": 2.586256493561433e-06,
      "loss": 0.0302,
      "step": 51455
    },
    {
      "epoch": 2.6123153459566475,
      "grad_norm": 0.33492904901504517,
      "learning_rate": 2.584564360289016e-06,
      "loss": 0.0388,
      "step": 51460
    },
    {
      "epoch": 2.61256916594751,
      "grad_norm": 0.5842387080192566,
      "learning_rate": 2.5828722270166e-06,
      "loss": 0.0331,
      "step": 51465
    },
    {
      "epoch": 2.6128229859383723,
      "grad_norm": 0.2856934368610382,
      "learning_rate": 2.5811800937441833e-06,
      "loss": 0.0305,
      "step": 51470
    },
    {
      "epoch": 2.613076805929235,
      "grad_norm": 0.35018250346183777,
      "learning_rate": 2.5794879604717673e-06,
      "loss": 0.0348,
      "step": 51475
    },
    {
      "epoch": 2.6133306259200975,
      "grad_norm": 0.2299485057592392,
      "learning_rate": 2.5777958271993504e-06,
      "loss": 0.0262,
      "step": 51480
    },
    {
      "epoch": 2.61358444591096,
      "grad_norm": 0.7264879941940308,
      "learning_rate": 2.576103693926934e-06,
      "loss": 0.0276,
      "step": 51485
    },
    {
      "epoch": 2.6138382659018227,
      "grad_norm": 0.30724015831947327,
      "learning_rate": 2.574411560654517e-06,
      "loss": 0.0316,
      "step": 51490
    },
    {
      "epoch": 2.614092085892685,
      "grad_norm": 0.309587299823761,
      "learning_rate": 2.572719427382101e-06,
      "loss": 0.0355,
      "step": 51495
    },
    {
      "epoch": 2.6143459058835474,
      "grad_norm": 1.0329458713531494,
      "learning_rate": 2.571027294109684e-06,
      "loss": 0.0293,
      "step": 51500
    },
    {
      "epoch": 2.61459972587441,
      "grad_norm": 0.30419212579727173,
      "learning_rate": 2.569335160837268e-06,
      "loss": 0.0299,
      "step": 51505
    },
    {
      "epoch": 2.614853545865272,
      "grad_norm": 0.336566686630249,
      "learning_rate": 2.5676430275648512e-06,
      "loss": 0.0285,
      "step": 51510
    },
    {
      "epoch": 2.615107365856135,
      "grad_norm": 0.3229808807373047,
      "learning_rate": 2.5659508942924348e-06,
      "loss": 0.0327,
      "step": 51515
    },
    {
      "epoch": 2.6153611858469974,
      "grad_norm": 0.25162097811698914,
      "learning_rate": 2.5642587610200183e-06,
      "loss": 0.031,
      "step": 51520
    },
    {
      "epoch": 2.61561500583786,
      "grad_norm": 0.5137783288955688,
      "learning_rate": 2.5625666277476014e-06,
      "loss": 0.0284,
      "step": 51525
    },
    {
      "epoch": 2.615868825828722,
      "grad_norm": 0.2492351233959198,
      "learning_rate": 2.560874494475185e-06,
      "loss": 0.0274,
      "step": 51530
    },
    {
      "epoch": 2.6161226458195848,
      "grad_norm": 0.5326012372970581,
      "learning_rate": 2.559182361202768e-06,
      "loss": 0.0369,
      "step": 51535
    },
    {
      "epoch": 2.6163764658104474,
      "grad_norm": 0.37052249908447266,
      "learning_rate": 2.557490227930352e-06,
      "loss": 0.0359,
      "step": 51540
    },
    {
      "epoch": 2.6166302858013095,
      "grad_norm": 0.20791205763816833,
      "learning_rate": 2.555798094657935e-06,
      "loss": 0.021,
      "step": 51545
    },
    {
      "epoch": 2.616884105792172,
      "grad_norm": 0.2117859572172165,
      "learning_rate": 2.554105961385519e-06,
      "loss": 0.0484,
      "step": 51550
    },
    {
      "epoch": 2.6171379257830347,
      "grad_norm": 0.452799528837204,
      "learning_rate": 2.5524138281131023e-06,
      "loss": 0.0291,
      "step": 51555
    },
    {
      "epoch": 2.6173917457738973,
      "grad_norm": 0.2676321864128113,
      "learning_rate": 2.550721694840686e-06,
      "loss": 0.0321,
      "step": 51560
    },
    {
      "epoch": 2.6176455657647595,
      "grad_norm": 0.39478787779808044,
      "learning_rate": 2.5490295615682694e-06,
      "loss": 0.034,
      "step": 51565
    },
    {
      "epoch": 2.617899385755622,
      "grad_norm": 0.3360627293586731,
      "learning_rate": 2.547337428295853e-06,
      "loss": 0.0293,
      "step": 51570
    },
    {
      "epoch": 2.6181532057464847,
      "grad_norm": 0.3433999717235565,
      "learning_rate": 2.545645295023436e-06,
      "loss": 0.0314,
      "step": 51575
    },
    {
      "epoch": 2.618407025737347,
      "grad_norm": 0.37200894951820374,
      "learning_rate": 2.54395316175102e-06,
      "loss": 0.0324,
      "step": 51580
    },
    {
      "epoch": 2.6186608457282095,
      "grad_norm": 0.4213031232357025,
      "learning_rate": 2.542261028478603e-06,
      "loss": 0.0339,
      "step": 51585
    },
    {
      "epoch": 2.618914665719072,
      "grad_norm": 0.4242783784866333,
      "learning_rate": 2.5405688952061867e-06,
      "loss": 0.0406,
      "step": 51590
    },
    {
      "epoch": 2.6191684857099347,
      "grad_norm": 0.3434397280216217,
      "learning_rate": 2.5388767619337702e-06,
      "loss": 0.03,
      "step": 51595
    },
    {
      "epoch": 2.6194223057007973,
      "grad_norm": 0.3476824462413788,
      "learning_rate": 2.5371846286613538e-06,
      "loss": 0.0326,
      "step": 51600
    },
    {
      "epoch": 2.6196761256916594,
      "grad_norm": 0.3337551951408386,
      "learning_rate": 2.535492495388937e-06,
      "loss": 0.0322,
      "step": 51605
    },
    {
      "epoch": 2.619929945682522,
      "grad_norm": 0.2564079761505127,
      "learning_rate": 2.5338003621165204e-06,
      "loss": 0.0347,
      "step": 51610
    },
    {
      "epoch": 2.620183765673384,
      "grad_norm": 0.4180367887020111,
      "learning_rate": 2.532108228844104e-06,
      "loss": 0.0297,
      "step": 51615
    },
    {
      "epoch": 2.620437585664247,
      "grad_norm": 0.3430015444755554,
      "learning_rate": 2.530416095571687e-06,
      "loss": 0.0303,
      "step": 51620
    },
    {
      "epoch": 2.6206914056551094,
      "grad_norm": 0.3892328143119812,
      "learning_rate": 2.528723962299271e-06,
      "loss": 0.0341,
      "step": 51625
    },
    {
      "epoch": 2.620945225645972,
      "grad_norm": 0.32110515236854553,
      "learning_rate": 2.527031829026854e-06,
      "loss": 0.0293,
      "step": 51630
    },
    {
      "epoch": 2.6211990456368346,
      "grad_norm": 0.3606153428554535,
      "learning_rate": 2.5253396957544377e-06,
      "loss": 0.033,
      "step": 51635
    },
    {
      "epoch": 2.6214528656276967,
      "grad_norm": 0.49945294857025146,
      "learning_rate": 2.5236475624820213e-06,
      "loss": 0.0346,
      "step": 51640
    },
    {
      "epoch": 2.6217066856185594,
      "grad_norm": 0.9243090748786926,
      "learning_rate": 2.521955429209605e-06,
      "loss": 0.029,
      "step": 51645
    },
    {
      "epoch": 2.621960505609422,
      "grad_norm": 0.20461325347423553,
      "learning_rate": 2.520263295937188e-06,
      "loss": 0.0294,
      "step": 51650
    },
    {
      "epoch": 2.622214325600284,
      "grad_norm": 0.19676488637924194,
      "learning_rate": 2.518571162664772e-06,
      "loss": 0.0318,
      "step": 51655
    },
    {
      "epoch": 2.6224681455911467,
      "grad_norm": 0.5037747621536255,
      "learning_rate": 2.516879029392355e-06,
      "loss": 0.0273,
      "step": 51660
    },
    {
      "epoch": 2.6227219655820093,
      "grad_norm": 0.37965553998947144,
      "learning_rate": 2.515186896119939e-06,
      "loss": 0.0333,
      "step": 51665
    },
    {
      "epoch": 2.622975785572872,
      "grad_norm": 0.6465805768966675,
      "learning_rate": 2.513494762847522e-06,
      "loss": 0.0321,
      "step": 51670
    },
    {
      "epoch": 2.623229605563734,
      "grad_norm": 0.2858799993991852,
      "learning_rate": 2.5118026295751057e-06,
      "loss": 0.0298,
      "step": 51675
    },
    {
      "epoch": 2.6234834255545967,
      "grad_norm": 0.4108503460884094,
      "learning_rate": 2.510110496302689e-06,
      "loss": 0.0394,
      "step": 51680
    },
    {
      "epoch": 2.6237372455454593,
      "grad_norm": 0.3374457359313965,
      "learning_rate": 2.5084183630302723e-06,
      "loss": 0.0277,
      "step": 51685
    },
    {
      "epoch": 2.6239910655363214,
      "grad_norm": 0.3668394386768341,
      "learning_rate": 2.506726229757856e-06,
      "loss": 0.0354,
      "step": 51690
    },
    {
      "epoch": 2.624244885527184,
      "grad_norm": 0.4046284854412079,
      "learning_rate": 2.505034096485439e-06,
      "loss": 0.0336,
      "step": 51695
    },
    {
      "epoch": 2.6244987055180466,
      "grad_norm": 0.32230740785598755,
      "learning_rate": 2.503341963213023e-06,
      "loss": 0.0322,
      "step": 51700
    },
    {
      "epoch": 2.6247525255089093,
      "grad_norm": 0.4608549475669861,
      "learning_rate": 2.501649829940606e-06,
      "loss": 0.0316,
      "step": 51705
    },
    {
      "epoch": 2.6250063454997714,
      "grad_norm": 0.43577080965042114,
      "learning_rate": 2.49995769666819e-06,
      "loss": 0.0345,
      "step": 51710
    },
    {
      "epoch": 2.625260165490634,
      "grad_norm": 0.21909955143928528,
      "learning_rate": 2.498265563395773e-06,
      "loss": 0.034,
      "step": 51715
    },
    {
      "epoch": 2.6255139854814966,
      "grad_norm": 0.31630709767341614,
      "learning_rate": 2.4965734301233567e-06,
      "loss": 0.0248,
      "step": 51720
    },
    {
      "epoch": 2.6257678054723588,
      "grad_norm": 0.23227165639400482,
      "learning_rate": 2.4948812968509403e-06,
      "loss": 0.0258,
      "step": 51725
    },
    {
      "epoch": 2.6260216254632214,
      "grad_norm": 0.3392637372016907,
      "learning_rate": 2.4931891635785234e-06,
      "loss": 0.0344,
      "step": 51730
    },
    {
      "epoch": 2.626275445454084,
      "grad_norm": 0.4881313145160675,
      "learning_rate": 2.491497030306107e-06,
      "loss": 0.033,
      "step": 51735
    },
    {
      "epoch": 2.6265292654449466,
      "grad_norm": 0.27817487716674805,
      "learning_rate": 2.4898048970336905e-06,
      "loss": 0.0287,
      "step": 51740
    },
    {
      "epoch": 2.626783085435809,
      "grad_norm": 0.31453368067741394,
      "learning_rate": 2.488112763761274e-06,
      "loss": 0.0288,
      "step": 51745
    },
    {
      "epoch": 2.6270369054266713,
      "grad_norm": 0.2960861921310425,
      "learning_rate": 2.4864206304888576e-06,
      "loss": 0.0319,
      "step": 51750
    },
    {
      "epoch": 2.627290725417534,
      "grad_norm": 0.34762755036354065,
      "learning_rate": 2.484728497216441e-06,
      "loss": 0.0285,
      "step": 51755
    },
    {
      "epoch": 2.627544545408396,
      "grad_norm": 0.4663408398628235,
      "learning_rate": 2.4830363639440242e-06,
      "loss": 0.0329,
      "step": 51760
    },
    {
      "epoch": 2.6277983653992587,
      "grad_norm": 0.3391950726509094,
      "learning_rate": 2.4813442306716078e-06,
      "loss": 0.0277,
      "step": 51765
    },
    {
      "epoch": 2.6280521853901213,
      "grad_norm": 0.27818772196769714,
      "learning_rate": 2.4796520973991913e-06,
      "loss": 0.0241,
      "step": 51770
    },
    {
      "epoch": 2.628306005380984,
      "grad_norm": 0.5784787535667419,
      "learning_rate": 2.477959964126775e-06,
      "loss": 0.0338,
      "step": 51775
    },
    {
      "epoch": 2.6285598253718465,
      "grad_norm": 0.25116094946861267,
      "learning_rate": 2.4762678308543584e-06,
      "loss": 0.0267,
      "step": 51780
    },
    {
      "epoch": 2.6288136453627087,
      "grad_norm": 0.4190313518047333,
      "learning_rate": 2.474575697581942e-06,
      "loss": 0.0366,
      "step": 51785
    },
    {
      "epoch": 2.6290674653535713,
      "grad_norm": 0.3404175341129303,
      "learning_rate": 2.4728835643095255e-06,
      "loss": 0.03,
      "step": 51790
    },
    {
      "epoch": 2.629321285344434,
      "grad_norm": 0.3389876186847687,
      "learning_rate": 2.4711914310371086e-06,
      "loss": 0.0292,
      "step": 51795
    },
    {
      "epoch": 2.629575105335296,
      "grad_norm": 0.8526211977005005,
      "learning_rate": 2.469499297764692e-06,
      "loss": 0.0283,
      "step": 51800
    },
    {
      "epoch": 2.6298289253261586,
      "grad_norm": 0.6599206924438477,
      "learning_rate": 2.4678071644922753e-06,
      "loss": 0.0448,
      "step": 51805
    },
    {
      "epoch": 2.6300827453170212,
      "grad_norm": 0.29481494426727295,
      "learning_rate": 2.466115031219859e-06,
      "loss": 0.0267,
      "step": 51810
    },
    {
      "epoch": 2.630336565307884,
      "grad_norm": 0.2825844883918762,
      "learning_rate": 2.4644228979474424e-06,
      "loss": 0.0232,
      "step": 51815
    },
    {
      "epoch": 2.630590385298746,
      "grad_norm": 0.3380671739578247,
      "learning_rate": 2.462730764675026e-06,
      "loss": 0.0391,
      "step": 51820
    },
    {
      "epoch": 2.6308442052896086,
      "grad_norm": 0.21934860944747925,
      "learning_rate": 2.4610386314026095e-06,
      "loss": 0.028,
      "step": 51825
    },
    {
      "epoch": 2.631098025280471,
      "grad_norm": 1.0363450050354004,
      "learning_rate": 2.459346498130193e-06,
      "loss": 0.0367,
      "step": 51830
    },
    {
      "epoch": 2.6313518452713334,
      "grad_norm": 0.4414408504962921,
      "learning_rate": 2.4576543648577765e-06,
      "loss": 0.0292,
      "step": 51835
    },
    {
      "epoch": 2.631605665262196,
      "grad_norm": 0.29103752970695496,
      "learning_rate": 2.4559622315853597e-06,
      "loss": 0.0279,
      "step": 51840
    },
    {
      "epoch": 2.6318594852530586,
      "grad_norm": 0.247410848736763,
      "learning_rate": 2.454270098312943e-06,
      "loss": 0.0352,
      "step": 51845
    },
    {
      "epoch": 2.632113305243921,
      "grad_norm": 0.36777010560035706,
      "learning_rate": 2.4525779650405268e-06,
      "loss": 0.0389,
      "step": 51850
    },
    {
      "epoch": 2.6323671252347833,
      "grad_norm": 0.3306533396244049,
      "learning_rate": 2.4508858317681103e-06,
      "loss": 0.0315,
      "step": 51855
    },
    {
      "epoch": 2.632620945225646,
      "grad_norm": 0.336128294467926,
      "learning_rate": 2.449193698495694e-06,
      "loss": 0.0365,
      "step": 51860
    },
    {
      "epoch": 2.6328747652165085,
      "grad_norm": 0.7833350896835327,
      "learning_rate": 2.4475015652232774e-06,
      "loss": 0.0352,
      "step": 51865
    },
    {
      "epoch": 2.6331285852073707,
      "grad_norm": 0.28254038095474243,
      "learning_rate": 2.445809431950861e-06,
      "loss": 0.0274,
      "step": 51870
    },
    {
      "epoch": 2.6333824051982333,
      "grad_norm": 0.21460244059562683,
      "learning_rate": 2.444117298678444e-06,
      "loss": 0.0301,
      "step": 51875
    },
    {
      "epoch": 2.633636225189096,
      "grad_norm": 0.2699926495552063,
      "learning_rate": 2.4424251654060276e-06,
      "loss": 0.0322,
      "step": 51880
    },
    {
      "epoch": 2.6338900451799585,
      "grad_norm": 0.20997563004493713,
      "learning_rate": 2.440733032133611e-06,
      "loss": 0.0222,
      "step": 51885
    },
    {
      "epoch": 2.634143865170821,
      "grad_norm": 0.5877103805541992,
      "learning_rate": 2.4390408988611943e-06,
      "loss": 0.037,
      "step": 51890
    },
    {
      "epoch": 2.6343976851616833,
      "grad_norm": 0.3468745946884155,
      "learning_rate": 2.437348765588778e-06,
      "loss": 0.0283,
      "step": 51895
    },
    {
      "epoch": 2.634651505152546,
      "grad_norm": 0.32325586676597595,
      "learning_rate": 2.4356566323163614e-06,
      "loss": 0.0345,
      "step": 51900
    },
    {
      "epoch": 2.634905325143408,
      "grad_norm": 0.3084450662136078,
      "learning_rate": 2.433964499043945e-06,
      "loss": 0.0383,
      "step": 51905
    },
    {
      "epoch": 2.6351591451342706,
      "grad_norm": 0.2576277554035187,
      "learning_rate": 2.4322723657715284e-06,
      "loss": 0.0308,
      "step": 51910
    },
    {
      "epoch": 2.6354129651251332,
      "grad_norm": 0.4685831367969513,
      "learning_rate": 2.430580232499112e-06,
      "loss": 0.0384,
      "step": 51915
    },
    {
      "epoch": 2.635666785115996,
      "grad_norm": 0.8822712302207947,
      "learning_rate": 2.428888099226695e-06,
      "loss": 0.0363,
      "step": 51920
    },
    {
      "epoch": 2.6359206051068584,
      "grad_norm": 0.2899301052093506,
      "learning_rate": 2.4271959659542787e-06,
      "loss": 0.0258,
      "step": 51925
    },
    {
      "epoch": 2.6361744250977206,
      "grad_norm": 0.40878719091415405,
      "learning_rate": 2.425503832681862e-06,
      "loss": 0.0341,
      "step": 51930
    },
    {
      "epoch": 2.636428245088583,
      "grad_norm": 0.37059518694877625,
      "learning_rate": 2.4238116994094457e-06,
      "loss": 0.0324,
      "step": 51935
    },
    {
      "epoch": 2.636682065079446,
      "grad_norm": 0.2958579957485199,
      "learning_rate": 2.4221195661370293e-06,
      "loss": 0.029,
      "step": 51940
    },
    {
      "epoch": 2.636935885070308,
      "grad_norm": 0.31784969568252563,
      "learning_rate": 2.420427432864613e-06,
      "loss": 0.024,
      "step": 51945
    },
    {
      "epoch": 2.6371897050611706,
      "grad_norm": 0.2704399526119232,
      "learning_rate": 2.4187352995921964e-06,
      "loss": 0.032,
      "step": 51950
    },
    {
      "epoch": 2.637443525052033,
      "grad_norm": 0.28617197275161743,
      "learning_rate": 2.4170431663197795e-06,
      "loss": 0.0315,
      "step": 51955
    },
    {
      "epoch": 2.6376973450428958,
      "grad_norm": 0.257956862449646,
      "learning_rate": 2.415351033047363e-06,
      "loss": 0.0271,
      "step": 51960
    },
    {
      "epoch": 2.637951165033758,
      "grad_norm": 0.25092050433158875,
      "learning_rate": 2.4136588997749466e-06,
      "loss": 0.0242,
      "step": 51965
    },
    {
      "epoch": 2.6382049850246205,
      "grad_norm": 0.36452189087867737,
      "learning_rate": 2.4119667665025297e-06,
      "loss": 0.0324,
      "step": 51970
    },
    {
      "epoch": 2.638458805015483,
      "grad_norm": 0.44480210542678833,
      "learning_rate": 2.4102746332301132e-06,
      "loss": 0.034,
      "step": 51975
    },
    {
      "epoch": 2.6387126250063453,
      "grad_norm": 0.1927006095647812,
      "learning_rate": 2.408582499957697e-06,
      "loss": 0.0254,
      "step": 51980
    },
    {
      "epoch": 2.638966444997208,
      "grad_norm": 0.2507188022136688,
      "learning_rate": 2.4068903666852803e-06,
      "loss": 0.0291,
      "step": 51985
    },
    {
      "epoch": 2.6392202649880705,
      "grad_norm": 0.3480997681617737,
      "learning_rate": 2.405198233412864e-06,
      "loss": 0.0336,
      "step": 51990
    },
    {
      "epoch": 2.639474084978933,
      "grad_norm": 0.28692418336868286,
      "learning_rate": 2.4035061001404474e-06,
      "loss": 0.0247,
      "step": 51995
    },
    {
      "epoch": 2.6397279049697953,
      "grad_norm": 0.3567029535770416,
      "learning_rate": 2.4018139668680305e-06,
      "loss": 0.0361,
      "step": 52000
    },
    {
      "epoch": 2.639981724960658,
      "grad_norm": 0.24052301049232483,
      "learning_rate": 2.400121833595614e-06,
      "loss": 0.0249,
      "step": 52005
    },
    {
      "epoch": 2.6402355449515205,
      "grad_norm": 0.34244805574417114,
      "learning_rate": 2.3984297003231976e-06,
      "loss": 0.0312,
      "step": 52010
    },
    {
      "epoch": 2.6404893649423826,
      "grad_norm": 1.291332483291626,
      "learning_rate": 2.396737567050781e-06,
      "loss": 0.032,
      "step": 52015
    },
    {
      "epoch": 2.6407431849332452,
      "grad_norm": 0.32589441537857056,
      "learning_rate": 2.3950454337783647e-06,
      "loss": 0.0273,
      "step": 52020
    },
    {
      "epoch": 2.640997004924108,
      "grad_norm": 0.2426830381155014,
      "learning_rate": 2.3933533005059483e-06,
      "loss": 0.0274,
      "step": 52025
    },
    {
      "epoch": 2.6412508249149704,
      "grad_norm": 0.2730412483215332,
      "learning_rate": 2.3916611672335314e-06,
      "loss": 0.0346,
      "step": 52030
    },
    {
      "epoch": 2.641504644905833,
      "grad_norm": 0.23146089911460876,
      "learning_rate": 2.389969033961115e-06,
      "loss": 0.0236,
      "step": 52035
    },
    {
      "epoch": 2.641758464896695,
      "grad_norm": 0.20168402791023254,
      "learning_rate": 2.3882769006886985e-06,
      "loss": 0.0281,
      "step": 52040
    },
    {
      "epoch": 2.642012284887558,
      "grad_norm": 0.3644895851612091,
      "learning_rate": 2.386584767416282e-06,
      "loss": 0.0337,
      "step": 52045
    },
    {
      "epoch": 2.64226610487842,
      "grad_norm": 0.47276461124420166,
      "learning_rate": 2.384892634143865e-06,
      "loss": 0.0304,
      "step": 52050
    },
    {
      "epoch": 2.6425199248692826,
      "grad_norm": 0.3690647780895233,
      "learning_rate": 2.3832005008714487e-06,
      "loss": 0.032,
      "step": 52055
    },
    {
      "epoch": 2.642773744860145,
      "grad_norm": 0.33413854241371155,
      "learning_rate": 2.3815083675990322e-06,
      "loss": 0.0267,
      "step": 52060
    },
    {
      "epoch": 2.6430275648510078,
      "grad_norm": 1.1163268089294434,
      "learning_rate": 2.3798162343266158e-06,
      "loss": 0.0275,
      "step": 52065
    },
    {
      "epoch": 2.6432813848418704,
      "grad_norm": 0.34472498297691345,
      "learning_rate": 2.3781241010541993e-06,
      "loss": 0.0305,
      "step": 52070
    },
    {
      "epoch": 2.6435352048327325,
      "grad_norm": 0.2940658926963806,
      "learning_rate": 2.3764319677817824e-06,
      "loss": 0.0293,
      "step": 52075
    },
    {
      "epoch": 2.643789024823595,
      "grad_norm": 0.24414999783039093,
      "learning_rate": 2.374739834509366e-06,
      "loss": 0.025,
      "step": 52080
    },
    {
      "epoch": 2.6440428448144577,
      "grad_norm": 0.2933432161808014,
      "learning_rate": 2.3730477012369495e-06,
      "loss": 0.0302,
      "step": 52085
    },
    {
      "epoch": 2.64429666480532,
      "grad_norm": 0.5342914462089539,
      "learning_rate": 2.371355567964533e-06,
      "loss": 0.0273,
      "step": 52090
    },
    {
      "epoch": 2.6445504847961825,
      "grad_norm": 0.35974082350730896,
      "learning_rate": 2.3696634346921166e-06,
      "loss": 0.0255,
      "step": 52095
    },
    {
      "epoch": 2.644804304787045,
      "grad_norm": 0.27689170837402344,
      "learning_rate": 2.3679713014197e-06,
      "loss": 0.0277,
      "step": 52100
    },
    {
      "epoch": 2.6450581247779077,
      "grad_norm": 0.5819672346115112,
      "learning_rate": 2.3662791681472837e-06,
      "loss": 0.0255,
      "step": 52105
    },
    {
      "epoch": 2.64531194476877,
      "grad_norm": 0.3903255760669708,
      "learning_rate": 2.364587034874867e-06,
      "loss": 0.0361,
      "step": 52110
    },
    {
      "epoch": 2.6455657647596325,
      "grad_norm": 0.2258889079093933,
      "learning_rate": 2.3628949016024504e-06,
      "loss": 0.0289,
      "step": 52115
    },
    {
      "epoch": 2.645819584750495,
      "grad_norm": 0.32111355662345886,
      "learning_rate": 2.361202768330034e-06,
      "loss": 0.023,
      "step": 52120
    },
    {
      "epoch": 2.646073404741357,
      "grad_norm": 0.2620217800140381,
      "learning_rate": 2.3595106350576175e-06,
      "loss": 0.0259,
      "step": 52125
    },
    {
      "epoch": 2.64632722473222,
      "grad_norm": 0.3304223120212555,
      "learning_rate": 2.3578185017852006e-06,
      "loss": 0.0291,
      "step": 52130
    },
    {
      "epoch": 2.6465810447230824,
      "grad_norm": 0.2463671863079071,
      "learning_rate": 2.356126368512784e-06,
      "loss": 0.0274,
      "step": 52135
    },
    {
      "epoch": 2.646834864713945,
      "grad_norm": 0.4745607376098633,
      "learning_rate": 2.3544342352403677e-06,
      "loss": 0.0259,
      "step": 52140
    },
    {
      "epoch": 2.6470886847048076,
      "grad_norm": 0.18332301080226898,
      "learning_rate": 2.3527421019679512e-06,
      "loss": 0.0306,
      "step": 52145
    },
    {
      "epoch": 2.64734250469567,
      "grad_norm": 0.37587037682533264,
      "learning_rate": 2.3510499686955348e-06,
      "loss": 0.0299,
      "step": 52150
    },
    {
      "epoch": 2.6475963246865324,
      "grad_norm": 0.4496591091156006,
      "learning_rate": 2.349357835423118e-06,
      "loss": 0.0279,
      "step": 52155
    },
    {
      "epoch": 2.6478501446773945,
      "grad_norm": 0.6839533448219299,
      "learning_rate": 2.3476657021507014e-06,
      "loss": 0.0284,
      "step": 52160
    },
    {
      "epoch": 2.648103964668257,
      "grad_norm": 0.32634589076042175,
      "learning_rate": 2.345973568878285e-06,
      "loss": 0.0281,
      "step": 52165
    },
    {
      "epoch": 2.6483577846591198,
      "grad_norm": 0.2792976200580597,
      "learning_rate": 2.3442814356058685e-06,
      "loss": 0.0332,
      "step": 52170
    },
    {
      "epoch": 2.6486116046499824,
      "grad_norm": 0.4249747693538666,
      "learning_rate": 2.342589302333452e-06,
      "loss": 0.0297,
      "step": 52175
    },
    {
      "epoch": 2.648865424640845,
      "grad_norm": 0.28616654872894287,
      "learning_rate": 2.3408971690610356e-06,
      "loss": 0.0327,
      "step": 52180
    },
    {
      "epoch": 2.649119244631707,
      "grad_norm": 0.2563963830471039,
      "learning_rate": 2.339205035788619e-06,
      "loss": 0.0272,
      "step": 52185
    },
    {
      "epoch": 2.6493730646225697,
      "grad_norm": 0.4164884388446808,
      "learning_rate": 2.3375129025162023e-06,
      "loss": 0.0285,
      "step": 52190
    },
    {
      "epoch": 2.649626884613432,
      "grad_norm": 0.4970816969871521,
      "learning_rate": 2.335820769243786e-06,
      "loss": 0.0301,
      "step": 52195
    },
    {
      "epoch": 2.6498807046042945,
      "grad_norm": 0.2564595341682434,
      "learning_rate": 2.3341286359713694e-06,
      "loss": 0.0295,
      "step": 52200
    },
    {
      "epoch": 2.650134524595157,
      "grad_norm": 0.343791127204895,
      "learning_rate": 2.332436502698953e-06,
      "loss": 0.0295,
      "step": 52205
    },
    {
      "epoch": 2.6503883445860197,
      "grad_norm": 0.2889670133590698,
      "learning_rate": 2.330744369426536e-06,
      "loss": 0.0255,
      "step": 52210
    },
    {
      "epoch": 2.6506421645768823,
      "grad_norm": 0.5983048677444458,
      "learning_rate": 2.3290522361541196e-06,
      "loss": 0.0273,
      "step": 52215
    },
    {
      "epoch": 2.6508959845677444,
      "grad_norm": 0.5745328664779663,
      "learning_rate": 2.327360102881703e-06,
      "loss": 0.0264,
      "step": 52220
    },
    {
      "epoch": 2.651149804558607,
      "grad_norm": 0.28300371766090393,
      "learning_rate": 2.3256679696092867e-06,
      "loss": 0.0311,
      "step": 52225
    },
    {
      "epoch": 2.6514036245494697,
      "grad_norm": 0.2970012128353119,
      "learning_rate": 2.32397583633687e-06,
      "loss": 0.0338,
      "step": 52230
    },
    {
      "epoch": 2.651657444540332,
      "grad_norm": 0.4181409776210785,
      "learning_rate": 2.3222837030644533e-06,
      "loss": 0.026,
      "step": 52235
    },
    {
      "epoch": 2.6519112645311944,
      "grad_norm": 0.31793472170829773,
      "learning_rate": 2.320591569792037e-06,
      "loss": 0.0355,
      "step": 52240
    },
    {
      "epoch": 2.652165084522057,
      "grad_norm": 0.3255579173564911,
      "learning_rate": 2.3188994365196204e-06,
      "loss": 0.0306,
      "step": 52245
    },
    {
      "epoch": 2.6524189045129196,
      "grad_norm": 0.35661157965660095,
      "learning_rate": 2.317207303247204e-06,
      "loss": 0.0335,
      "step": 52250
    },
    {
      "epoch": 2.6526727245037818,
      "grad_norm": 0.2453710436820984,
      "learning_rate": 2.3155151699747875e-06,
      "loss": 0.0285,
      "step": 52255
    },
    {
      "epoch": 2.6529265444946444,
      "grad_norm": 0.36720964312553406,
      "learning_rate": 2.313823036702371e-06,
      "loss": 0.034,
      "step": 52260
    },
    {
      "epoch": 2.653180364485507,
      "grad_norm": 0.32312971353530884,
      "learning_rate": 2.3121309034299546e-06,
      "loss": 0.0299,
      "step": 52265
    },
    {
      "epoch": 2.653434184476369,
      "grad_norm": 0.39574339985847473,
      "learning_rate": 2.3104387701575377e-06,
      "loss": 0.0318,
      "step": 52270
    },
    {
      "epoch": 2.6536880044672317,
      "grad_norm": 0.21212691068649292,
      "learning_rate": 2.3087466368851213e-06,
      "loss": 0.0292,
      "step": 52275
    },
    {
      "epoch": 2.6539418244580943,
      "grad_norm": 0.2754327356815338,
      "learning_rate": 2.307054503612705e-06,
      "loss": 0.0278,
      "step": 52280
    },
    {
      "epoch": 2.654195644448957,
      "grad_norm": 0.3506637513637543,
      "learning_rate": 2.3053623703402884e-06,
      "loss": 0.0373,
      "step": 52285
    },
    {
      "epoch": 2.6544494644398196,
      "grad_norm": 0.3828669786453247,
      "learning_rate": 2.3036702370678715e-06,
      "loss": 0.0338,
      "step": 52290
    },
    {
      "epoch": 2.6547032844306817,
      "grad_norm": 0.31429722905158997,
      "learning_rate": 2.301978103795455e-06,
      "loss": 0.0328,
      "step": 52295
    },
    {
      "epoch": 2.6549571044215443,
      "grad_norm": 0.3292172849178314,
      "learning_rate": 2.3002859705230386e-06,
      "loss": 0.0309,
      "step": 52300
    },
    {
      "epoch": 2.6552109244124065,
      "grad_norm": 0.3343614339828491,
      "learning_rate": 2.298593837250622e-06,
      "loss": 0.0296,
      "step": 52305
    },
    {
      "epoch": 2.655464744403269,
      "grad_norm": 0.245424285531044,
      "learning_rate": 2.2969017039782057e-06,
      "loss": 0.0309,
      "step": 52310
    },
    {
      "epoch": 2.6557185643941317,
      "grad_norm": 0.3650681674480438,
      "learning_rate": 2.2952095707057888e-06,
      "loss": 0.0319,
      "step": 52315
    },
    {
      "epoch": 2.6559723843849943,
      "grad_norm": 0.5489715337753296,
      "learning_rate": 2.2935174374333723e-06,
      "loss": 0.0289,
      "step": 52320
    },
    {
      "epoch": 2.656226204375857,
      "grad_norm": 0.6062617301940918,
      "learning_rate": 2.291825304160956e-06,
      "loss": 0.0304,
      "step": 52325
    },
    {
      "epoch": 2.656480024366719,
      "grad_norm": 0.27395087480545044,
      "learning_rate": 2.2901331708885394e-06,
      "loss": 0.0307,
      "step": 52330
    },
    {
      "epoch": 2.6567338443575816,
      "grad_norm": 0.2297271490097046,
      "learning_rate": 2.288441037616123e-06,
      "loss": 0.0285,
      "step": 52335
    },
    {
      "epoch": 2.6569876643484442,
      "grad_norm": 0.32028794288635254,
      "learning_rate": 2.2867489043437065e-06,
      "loss": 0.0345,
      "step": 52340
    },
    {
      "epoch": 2.6572414843393064,
      "grad_norm": 0.20256635546684265,
      "learning_rate": 2.2850567710712896e-06,
      "loss": 0.027,
      "step": 52345
    },
    {
      "epoch": 2.657495304330169,
      "grad_norm": 0.7101668119430542,
      "learning_rate": 2.283364637798873e-06,
      "loss": 0.0304,
      "step": 52350
    },
    {
      "epoch": 2.6577491243210316,
      "grad_norm": 0.2871030271053314,
      "learning_rate": 2.2816725045264567e-06,
      "loss": 0.0293,
      "step": 52355
    },
    {
      "epoch": 2.658002944311894,
      "grad_norm": 0.39420485496520996,
      "learning_rate": 2.2799803712540402e-06,
      "loss": 0.0349,
      "step": 52360
    },
    {
      "epoch": 2.6582567643027564,
      "grad_norm": 0.2553163468837738,
      "learning_rate": 2.278288237981624e-06,
      "loss": 0.0279,
      "step": 52365
    },
    {
      "epoch": 2.658510584293619,
      "grad_norm": 0.24502582848072052,
      "learning_rate": 2.2765961047092073e-06,
      "loss": 0.028,
      "step": 52370
    },
    {
      "epoch": 2.6587644042844816,
      "grad_norm": 0.41072553396224976,
      "learning_rate": 2.2749039714367905e-06,
      "loss": 0.0246,
      "step": 52375
    },
    {
      "epoch": 2.6590182242753437,
      "grad_norm": 0.7050197720527649,
      "learning_rate": 2.273211838164374e-06,
      "loss": 0.0271,
      "step": 52380
    },
    {
      "epoch": 2.6592720442662063,
      "grad_norm": 0.6191830039024353,
      "learning_rate": 2.2715197048919575e-06,
      "loss": 0.0309,
      "step": 52385
    },
    {
      "epoch": 2.659525864257069,
      "grad_norm": 0.22344526648521423,
      "learning_rate": 2.2698275716195407e-06,
      "loss": 0.0266,
      "step": 52390
    },
    {
      "epoch": 2.6597796842479315,
      "grad_norm": 0.3696863353252411,
      "learning_rate": 2.2681354383471242e-06,
      "loss": 0.0291,
      "step": 52395
    },
    {
      "epoch": 2.6600335042387937,
      "grad_norm": 0.4067589044570923,
      "learning_rate": 2.2664433050747078e-06,
      "loss": 0.0308,
      "step": 52400
    },
    {
      "epoch": 2.6602873242296563,
      "grad_norm": 0.2778363525867462,
      "learning_rate": 2.2647511718022913e-06,
      "loss": 0.0253,
      "step": 52405
    },
    {
      "epoch": 2.660541144220519,
      "grad_norm": 0.40471798181533813,
      "learning_rate": 2.263059038529875e-06,
      "loss": 0.029,
      "step": 52410
    },
    {
      "epoch": 2.660794964211381,
      "grad_norm": 0.31421124935150146,
      "learning_rate": 2.2613669052574584e-06,
      "loss": 0.0298,
      "step": 52415
    },
    {
      "epoch": 2.6610487842022437,
      "grad_norm": 0.39784589409828186,
      "learning_rate": 2.259674771985042e-06,
      "loss": 0.0369,
      "step": 52420
    },
    {
      "epoch": 2.6613026041931063,
      "grad_norm": 0.3476801812648773,
      "learning_rate": 2.257982638712625e-06,
      "loss": 0.0311,
      "step": 52425
    },
    {
      "epoch": 2.661556424183969,
      "grad_norm": 0.3795233368873596,
      "learning_rate": 2.2562905054402086e-06,
      "loss": 0.039,
      "step": 52430
    },
    {
      "epoch": 2.6618102441748315,
      "grad_norm": 0.6271621584892273,
      "learning_rate": 2.254598372167792e-06,
      "loss": 0.0313,
      "step": 52435
    },
    {
      "epoch": 2.6620640641656936,
      "grad_norm": 0.3377338647842407,
      "learning_rate": 2.2529062388953757e-06,
      "loss": 0.0317,
      "step": 52440
    },
    {
      "epoch": 2.6623178841565562,
      "grad_norm": 0.3012075126171112,
      "learning_rate": 2.2512141056229592e-06,
      "loss": 0.0316,
      "step": 52445
    },
    {
      "epoch": 2.6625717041474184,
      "grad_norm": 0.3194831311702728,
      "learning_rate": 2.2495219723505428e-06,
      "loss": 0.0339,
      "step": 52450
    },
    {
      "epoch": 2.662825524138281,
      "grad_norm": 0.3819662630558014,
      "learning_rate": 2.247829839078126e-06,
      "loss": 0.0286,
      "step": 52455
    },
    {
      "epoch": 2.6630793441291436,
      "grad_norm": 0.24752876162528992,
      "learning_rate": 2.2461377058057094e-06,
      "loss": 0.0328,
      "step": 52460
    },
    {
      "epoch": 2.663333164120006,
      "grad_norm": 0.3253592550754547,
      "learning_rate": 2.244445572533293e-06,
      "loss": 0.0269,
      "step": 52465
    },
    {
      "epoch": 2.663586984110869,
      "grad_norm": 0.3416394591331482,
      "learning_rate": 2.242753439260876e-06,
      "loss": 0.028,
      "step": 52470
    },
    {
      "epoch": 2.663840804101731,
      "grad_norm": 0.3783033490180969,
      "learning_rate": 2.2410613059884597e-06,
      "loss": 0.0319,
      "step": 52475
    },
    {
      "epoch": 2.6640946240925936,
      "grad_norm": 0.43123677372932434,
      "learning_rate": 2.239369172716043e-06,
      "loss": 0.0392,
      "step": 52480
    },
    {
      "epoch": 2.664348444083456,
      "grad_norm": 0.28624406456947327,
      "learning_rate": 2.2376770394436267e-06,
      "loss": 0.026,
      "step": 52485
    },
    {
      "epoch": 2.6646022640743183,
      "grad_norm": 0.3098219335079193,
      "learning_rate": 2.2359849061712103e-06,
      "loss": 0.0311,
      "step": 52490
    },
    {
      "epoch": 2.664856084065181,
      "grad_norm": 0.3867039978504181,
      "learning_rate": 2.234292772898794e-06,
      "loss": 0.0354,
      "step": 52495
    },
    {
      "epoch": 2.6651099040560435,
      "grad_norm": 0.7705091834068298,
      "learning_rate": 2.2326006396263774e-06,
      "loss": 0.0286,
      "step": 52500
    },
    {
      "epoch": 2.665363724046906,
      "grad_norm": 0.3141399025917053,
      "learning_rate": 2.2309085063539605e-06,
      "loss": 0.0261,
      "step": 52505
    },
    {
      "epoch": 2.6656175440377683,
      "grad_norm": 0.9174738526344299,
      "learning_rate": 2.229216373081544e-06,
      "loss": 0.0257,
      "step": 52510
    },
    {
      "epoch": 2.665871364028631,
      "grad_norm": 0.25226977467536926,
      "learning_rate": 2.2275242398091276e-06,
      "loss": 0.0275,
      "step": 52515
    },
    {
      "epoch": 2.6661251840194935,
      "grad_norm": 0.7154881358146667,
      "learning_rate": 2.225832106536711e-06,
      "loss": 0.0298,
      "step": 52520
    },
    {
      "epoch": 2.6663790040103557,
      "grad_norm": 0.5491287708282471,
      "learning_rate": 2.2241399732642947e-06,
      "loss": 0.0268,
      "step": 52525
    },
    {
      "epoch": 2.6666328240012183,
      "grad_norm": 0.24701310694217682,
      "learning_rate": 2.2224478399918782e-06,
      "loss": 0.0329,
      "step": 52530
    },
    {
      "epoch": 2.666886643992081,
      "grad_norm": 1.1633200645446777,
      "learning_rate": 2.2207557067194613e-06,
      "loss": 0.0353,
      "step": 52535
    },
    {
      "epoch": 2.6671404639829435,
      "grad_norm": 0.30238714814186096,
      "learning_rate": 2.219063573447045e-06,
      "loss": 0.0331,
      "step": 52540
    },
    {
      "epoch": 2.6673942839738056,
      "grad_norm": 0.33450546860694885,
      "learning_rate": 2.2173714401746284e-06,
      "loss": 0.0367,
      "step": 52545
    },
    {
      "epoch": 2.6676481039646682,
      "grad_norm": 0.5556016564369202,
      "learning_rate": 2.2156793069022116e-06,
      "loss": 0.0366,
      "step": 52550
    },
    {
      "epoch": 2.667901923955531,
      "grad_norm": 0.30191656947135925,
      "learning_rate": 2.213987173629795e-06,
      "loss": 0.0265,
      "step": 52555
    },
    {
      "epoch": 2.668155743946393,
      "grad_norm": 0.40027421712875366,
      "learning_rate": 2.2122950403573786e-06,
      "loss": 0.0304,
      "step": 52560
    },
    {
      "epoch": 2.6684095639372556,
      "grad_norm": 0.2920042872428894,
      "learning_rate": 2.210602907084962e-06,
      "loss": 0.0322,
      "step": 52565
    },
    {
      "epoch": 2.668663383928118,
      "grad_norm": 0.29836907982826233,
      "learning_rate": 2.2089107738125457e-06,
      "loss": 0.036,
      "step": 52570
    },
    {
      "epoch": 2.668917203918981,
      "grad_norm": 0.2802891135215759,
      "learning_rate": 2.2072186405401293e-06,
      "loss": 0.0324,
      "step": 52575
    },
    {
      "epoch": 2.6691710239098434,
      "grad_norm": 0.282135933637619,
      "learning_rate": 2.205526507267713e-06,
      "loss": 0.0262,
      "step": 52580
    },
    {
      "epoch": 2.6694248439007056,
      "grad_norm": 0.35920336842536926,
      "learning_rate": 2.203834373995296e-06,
      "loss": 0.0236,
      "step": 52585
    },
    {
      "epoch": 2.669678663891568,
      "grad_norm": 0.44473469257354736,
      "learning_rate": 2.2021422407228795e-06,
      "loss": 0.03,
      "step": 52590
    },
    {
      "epoch": 2.6699324838824303,
      "grad_norm": 0.37438032031059265,
      "learning_rate": 2.200450107450463e-06,
      "loss": 0.0303,
      "step": 52595
    },
    {
      "epoch": 2.670186303873293,
      "grad_norm": 0.7606229782104492,
      "learning_rate": 2.1987579741780466e-06,
      "loss": 0.0332,
      "step": 52600
    },
    {
      "epoch": 2.6704401238641555,
      "grad_norm": 0.2716493606567383,
      "learning_rate": 2.19706584090563e-06,
      "loss": 0.0284,
      "step": 52605
    },
    {
      "epoch": 2.670693943855018,
      "grad_norm": 0.38287749886512756,
      "learning_rate": 2.1953737076332137e-06,
      "loss": 0.0348,
      "step": 52610
    },
    {
      "epoch": 2.6709477638458807,
      "grad_norm": 0.31366923451423645,
      "learning_rate": 2.1936815743607968e-06,
      "loss": 0.0283,
      "step": 52615
    },
    {
      "epoch": 2.671201583836743,
      "grad_norm": 0.3830898106098175,
      "learning_rate": 2.1919894410883803e-06,
      "loss": 0.0251,
      "step": 52620
    },
    {
      "epoch": 2.6714554038276055,
      "grad_norm": 0.2692367434501648,
      "learning_rate": 2.190297307815964e-06,
      "loss": 0.0276,
      "step": 52625
    },
    {
      "epoch": 2.671709223818468,
      "grad_norm": 0.27730458974838257,
      "learning_rate": 2.188605174543547e-06,
      "loss": 0.026,
      "step": 52630
    },
    {
      "epoch": 2.6719630438093303,
      "grad_norm": 0.36511486768722534,
      "learning_rate": 2.1869130412711305e-06,
      "loss": 0.0258,
      "step": 52635
    },
    {
      "epoch": 2.672216863800193,
      "grad_norm": 0.3859044909477234,
      "learning_rate": 2.185220907998714e-06,
      "loss": 0.0237,
      "step": 52640
    },
    {
      "epoch": 2.6724706837910555,
      "grad_norm": 0.35351133346557617,
      "learning_rate": 2.1835287747262976e-06,
      "loss": 0.0282,
      "step": 52645
    },
    {
      "epoch": 2.672724503781918,
      "grad_norm": 0.301961749792099,
      "learning_rate": 2.181836641453881e-06,
      "loss": 0.0298,
      "step": 52650
    },
    {
      "epoch": 2.67297832377278,
      "grad_norm": 0.3102557063102722,
      "learning_rate": 2.1801445081814647e-06,
      "loss": 0.0286,
      "step": 52655
    },
    {
      "epoch": 2.673232143763643,
      "grad_norm": 0.3127071261405945,
      "learning_rate": 2.178452374909048e-06,
      "loss": 0.033,
      "step": 52660
    },
    {
      "epoch": 2.6734859637545054,
      "grad_norm": 0.24851199984550476,
      "learning_rate": 2.1767602416366314e-06,
      "loss": 0.0271,
      "step": 52665
    },
    {
      "epoch": 2.6737397837453676,
      "grad_norm": 0.2732860743999481,
      "learning_rate": 2.175068108364215e-06,
      "loss": 0.0266,
      "step": 52670
    },
    {
      "epoch": 2.67399360373623,
      "grad_norm": 0.40874701738357544,
      "learning_rate": 2.1733759750917985e-06,
      "loss": 0.0247,
      "step": 52675
    },
    {
      "epoch": 2.674247423727093,
      "grad_norm": 0.18156781792640686,
      "learning_rate": 2.171683841819382e-06,
      "loss": 0.0251,
      "step": 52680
    },
    {
      "epoch": 2.6745012437179554,
      "grad_norm": 0.4294818043708801,
      "learning_rate": 2.1699917085469656e-06,
      "loss": 0.0282,
      "step": 52685
    },
    {
      "epoch": 2.6747550637088175,
      "grad_norm": 0.4318823218345642,
      "learning_rate": 2.168299575274549e-06,
      "loss": 0.0325,
      "step": 52690
    },
    {
      "epoch": 2.67500888369968,
      "grad_norm": 0.2763345241546631,
      "learning_rate": 2.1666074420021322e-06,
      "loss": 0.0276,
      "step": 52695
    },
    {
      "epoch": 2.6752627036905428,
      "grad_norm": 0.3835084140300751,
      "learning_rate": 2.1649153087297158e-06,
      "loss": 0.035,
      "step": 52700
    },
    {
      "epoch": 2.675516523681405,
      "grad_norm": 0.3311332166194916,
      "learning_rate": 2.163223175457299e-06,
      "loss": 0.0235,
      "step": 52705
    },
    {
      "epoch": 2.6757703436722675,
      "grad_norm": 0.2773440182209015,
      "learning_rate": 2.1615310421848824e-06,
      "loss": 0.0273,
      "step": 52710
    },
    {
      "epoch": 2.67602416366313,
      "grad_norm": 0.508554995059967,
      "learning_rate": 2.159838908912466e-06,
      "loss": 0.0305,
      "step": 52715
    },
    {
      "epoch": 2.6762779836539927,
      "grad_norm": 0.24255989491939545,
      "learning_rate": 2.1581467756400495e-06,
      "loss": 0.0314,
      "step": 52720
    },
    {
      "epoch": 2.6765318036448553,
      "grad_norm": 0.37368252873420715,
      "learning_rate": 2.156454642367633e-06,
      "loss": 0.0298,
      "step": 52725
    },
    {
      "epoch": 2.6767856236357175,
      "grad_norm": 0.41288262605667114,
      "learning_rate": 2.1547625090952166e-06,
      "loss": 0.0274,
      "step": 52730
    },
    {
      "epoch": 2.67703944362658,
      "grad_norm": 0.3736118674278259,
      "learning_rate": 2.1530703758228e-06,
      "loss": 0.0297,
      "step": 52735
    },
    {
      "epoch": 2.6772932636174422,
      "grad_norm": 0.3442663252353668,
      "learning_rate": 2.1513782425503833e-06,
      "loss": 0.0319,
      "step": 52740
    },
    {
      "epoch": 2.677547083608305,
      "grad_norm": 0.2854013741016388,
      "learning_rate": 2.149686109277967e-06,
      "loss": 0.0279,
      "step": 52745
    },
    {
      "epoch": 2.6778009035991674,
      "grad_norm": 0.2770628333091736,
      "learning_rate": 2.1479939760055504e-06,
      "loss": 0.031,
      "step": 52750
    },
    {
      "epoch": 2.67805472359003,
      "grad_norm": 0.2934173345565796,
      "learning_rate": 2.146301842733134e-06,
      "loss": 0.0259,
      "step": 52755
    },
    {
      "epoch": 2.6783085435808927,
      "grad_norm": 0.24669590592384338,
      "learning_rate": 2.1446097094607175e-06,
      "loss": 0.0308,
      "step": 52760
    },
    {
      "epoch": 2.678562363571755,
      "grad_norm": 0.5056079030036926,
      "learning_rate": 2.142917576188301e-06,
      "loss": 0.028,
      "step": 52765
    },
    {
      "epoch": 2.6788161835626174,
      "grad_norm": 0.4430501163005829,
      "learning_rate": 2.1412254429158845e-06,
      "loss": 0.0335,
      "step": 52770
    },
    {
      "epoch": 2.67907000355348,
      "grad_norm": 0.41625797748565674,
      "learning_rate": 2.1395333096434677e-06,
      "loss": 0.0381,
      "step": 52775
    },
    {
      "epoch": 2.679323823544342,
      "grad_norm": 0.2931457757949829,
      "learning_rate": 2.1378411763710512e-06,
      "loss": 0.0279,
      "step": 52780
    },
    {
      "epoch": 2.679577643535205,
      "grad_norm": 0.28468772768974304,
      "learning_rate": 2.1361490430986343e-06,
      "loss": 0.0296,
      "step": 52785
    },
    {
      "epoch": 2.6798314635260674,
      "grad_norm": 0.8410632610321045,
      "learning_rate": 2.134456909826218e-06,
      "loss": 0.0332,
      "step": 52790
    },
    {
      "epoch": 2.68008528351693,
      "grad_norm": 0.31851619482040405,
      "learning_rate": 2.1327647765538014e-06,
      "loss": 0.0326,
      "step": 52795
    },
    {
      "epoch": 2.680339103507792,
      "grad_norm": 0.3129703104496002,
      "learning_rate": 2.131072643281385e-06,
      "loss": 0.0289,
      "step": 52800
    },
    {
      "epoch": 2.6805929234986547,
      "grad_norm": 0.3168695271015167,
      "learning_rate": 2.1293805100089685e-06,
      "loss": 0.0252,
      "step": 52805
    },
    {
      "epoch": 2.6808467434895173,
      "grad_norm": 0.4538348615169525,
      "learning_rate": 2.127688376736552e-06,
      "loss": 0.0294,
      "step": 52810
    },
    {
      "epoch": 2.6811005634803795,
      "grad_norm": 0.9396774768829346,
      "learning_rate": 2.1259962434641356e-06,
      "loss": 0.0286,
      "step": 52815
    },
    {
      "epoch": 2.681354383471242,
      "grad_norm": 0.3485766351222992,
      "learning_rate": 2.1243041101917187e-06,
      "loss": 0.0326,
      "step": 52820
    },
    {
      "epoch": 2.6816082034621047,
      "grad_norm": 0.27177703380584717,
      "learning_rate": 2.1226119769193023e-06,
      "loss": 0.0305,
      "step": 52825
    },
    {
      "epoch": 2.6818620234529673,
      "grad_norm": 0.3578769862651825,
      "learning_rate": 2.120919843646886e-06,
      "loss": 0.0273,
      "step": 52830
    },
    {
      "epoch": 2.6821158434438295,
      "grad_norm": 0.2714287042617798,
      "learning_rate": 2.1192277103744694e-06,
      "loss": 0.0323,
      "step": 52835
    },
    {
      "epoch": 2.682369663434692,
      "grad_norm": 0.22217968106269836,
      "learning_rate": 2.117535577102053e-06,
      "loss": 0.0334,
      "step": 52840
    },
    {
      "epoch": 2.6826234834255547,
      "grad_norm": 0.2560397684574127,
      "learning_rate": 2.1158434438296364e-06,
      "loss": 0.032,
      "step": 52845
    },
    {
      "epoch": 2.682877303416417,
      "grad_norm": 0.4822799861431122,
      "learning_rate": 2.11415131055722e-06,
      "loss": 0.0314,
      "step": 52850
    },
    {
      "epoch": 2.6831311234072794,
      "grad_norm": 0.37949657440185547,
      "learning_rate": 2.112459177284803e-06,
      "loss": 0.0335,
      "step": 52855
    },
    {
      "epoch": 2.683384943398142,
      "grad_norm": 0.4751673638820648,
      "learning_rate": 2.1107670440123867e-06,
      "loss": 0.0316,
      "step": 52860
    },
    {
      "epoch": 2.6836387633890046,
      "grad_norm": 0.17257443070411682,
      "learning_rate": 2.1090749107399698e-06,
      "loss": 0.0286,
      "step": 52865
    },
    {
      "epoch": 2.6838925833798672,
      "grad_norm": 0.28788110613822937,
      "learning_rate": 2.1073827774675533e-06,
      "loss": 0.0273,
      "step": 52870
    },
    {
      "epoch": 2.6841464033707294,
      "grad_norm": 0.37915948033332825,
      "learning_rate": 2.105690644195137e-06,
      "loss": 0.0249,
      "step": 52875
    },
    {
      "epoch": 2.684400223361592,
      "grad_norm": 0.29168131947517395,
      "learning_rate": 2.1039985109227204e-06,
      "loss": 0.0308,
      "step": 52880
    },
    {
      "epoch": 2.684654043352454,
      "grad_norm": 0.43253839015960693,
      "learning_rate": 2.102306377650304e-06,
      "loss": 0.0289,
      "step": 52885
    },
    {
      "epoch": 2.6849078633433168,
      "grad_norm": 0.3503051698207855,
      "learning_rate": 2.1006142443778875e-06,
      "loss": 0.0344,
      "step": 52890
    },
    {
      "epoch": 2.6851616833341794,
      "grad_norm": 0.30547627806663513,
      "learning_rate": 2.098922111105471e-06,
      "loss": 0.0327,
      "step": 52895
    },
    {
      "epoch": 2.685415503325042,
      "grad_norm": 0.5834780931472778,
      "learning_rate": 2.097229977833054e-06,
      "loss": 0.0353,
      "step": 52900
    },
    {
      "epoch": 2.6856693233159046,
      "grad_norm": 0.22508762776851654,
      "learning_rate": 2.0955378445606377e-06,
      "loss": 0.0277,
      "step": 52905
    },
    {
      "epoch": 2.6859231433067667,
      "grad_norm": 0.36343008279800415,
      "learning_rate": 2.0938457112882213e-06,
      "loss": 0.0336,
      "step": 52910
    },
    {
      "epoch": 2.6861769632976293,
      "grad_norm": 0.3569226861000061,
      "learning_rate": 2.092153578015805e-06,
      "loss": 0.0343,
      "step": 52915
    },
    {
      "epoch": 2.686430783288492,
      "grad_norm": 0.5335173010826111,
      "learning_rate": 2.0904614447433883e-06,
      "loss": 0.0273,
      "step": 52920
    },
    {
      "epoch": 2.686684603279354,
      "grad_norm": 2.068007707595825,
      "learning_rate": 2.088769311470972e-06,
      "loss": 0.0362,
      "step": 52925
    },
    {
      "epoch": 2.6869384232702167,
      "grad_norm": 0.3901418447494507,
      "learning_rate": 2.087077178198555e-06,
      "loss": 0.0302,
      "step": 52930
    },
    {
      "epoch": 2.6871922432610793,
      "grad_norm": 0.4862616956233978,
      "learning_rate": 2.0853850449261386e-06,
      "loss": 0.0289,
      "step": 52935
    },
    {
      "epoch": 2.687446063251942,
      "grad_norm": 0.38844648003578186,
      "learning_rate": 2.083692911653722e-06,
      "loss": 0.0397,
      "step": 52940
    },
    {
      "epoch": 2.687699883242804,
      "grad_norm": 0.3375749886035919,
      "learning_rate": 2.0820007783813052e-06,
      "loss": 0.0247,
      "step": 52945
    },
    {
      "epoch": 2.6879537032336667,
      "grad_norm": 0.6121558547019958,
      "learning_rate": 2.0803086451088888e-06,
      "loss": 0.0341,
      "step": 52950
    },
    {
      "epoch": 2.6882075232245293,
      "grad_norm": 0.4033074676990509,
      "learning_rate": 2.0786165118364723e-06,
      "loss": 0.033,
      "step": 52955
    },
    {
      "epoch": 2.6884613432153914,
      "grad_norm": 0.27083489298820496,
      "learning_rate": 2.076924378564056e-06,
      "loss": 0.0344,
      "step": 52960
    },
    {
      "epoch": 2.688715163206254,
      "grad_norm": 0.2858365774154663,
      "learning_rate": 2.0752322452916394e-06,
      "loss": 0.0369,
      "step": 52965
    },
    {
      "epoch": 2.6889689831971166,
      "grad_norm": 0.4232316315174103,
      "learning_rate": 2.073540112019223e-06,
      "loss": 0.0317,
      "step": 52970
    },
    {
      "epoch": 2.6892228031879792,
      "grad_norm": 0.21906213462352753,
      "learning_rate": 2.071847978746806e-06,
      "loss": 0.03,
      "step": 52975
    },
    {
      "epoch": 2.689476623178842,
      "grad_norm": 0.27975738048553467,
      "learning_rate": 2.0701558454743896e-06,
      "loss": 0.0287,
      "step": 52980
    },
    {
      "epoch": 2.689730443169704,
      "grad_norm": 0.23709796369075775,
      "learning_rate": 2.068463712201973e-06,
      "loss": 0.0282,
      "step": 52985
    },
    {
      "epoch": 2.6899842631605666,
      "grad_norm": 0.23567497730255127,
      "learning_rate": 2.0667715789295567e-06,
      "loss": 0.0268,
      "step": 52990
    },
    {
      "epoch": 2.6902380831514288,
      "grad_norm": 0.2758468985557556,
      "learning_rate": 2.0650794456571402e-06,
      "loss": 0.0262,
      "step": 52995
    },
    {
      "epoch": 2.6904919031422914,
      "grad_norm": 0.3465255796909332,
      "learning_rate": 2.0633873123847238e-06,
      "loss": 0.0298,
      "step": 53000
    },
    {
      "epoch": 2.690745723133154,
      "grad_norm": 0.44373437762260437,
      "learning_rate": 2.0616951791123073e-06,
      "loss": 0.0269,
      "step": 53005
    },
    {
      "epoch": 2.6909995431240166,
      "grad_norm": 0.26769983768463135,
      "learning_rate": 2.0600030458398905e-06,
      "loss": 0.0326,
      "step": 53010
    },
    {
      "epoch": 2.691253363114879,
      "grad_norm": 0.38085025548934937,
      "learning_rate": 2.058310912567474e-06,
      "loss": 0.0337,
      "step": 53015
    },
    {
      "epoch": 2.6915071831057413,
      "grad_norm": 0.306956022977829,
      "learning_rate": 2.056618779295057e-06,
      "loss": 0.0285,
      "step": 53020
    },
    {
      "epoch": 2.691761003096604,
      "grad_norm": 0.3227962255477905,
      "learning_rate": 2.0549266460226407e-06,
      "loss": 0.0244,
      "step": 53025
    },
    {
      "epoch": 2.6920148230874665,
      "grad_norm": 0.3663029372692108,
      "learning_rate": 2.053234512750224e-06,
      "loss": 0.0341,
      "step": 53030
    },
    {
      "epoch": 2.6922686430783287,
      "grad_norm": 0.30087295174598694,
      "learning_rate": 2.0515423794778078e-06,
      "loss": 0.0226,
      "step": 53035
    },
    {
      "epoch": 2.6925224630691913,
      "grad_norm": 0.3504243493080139,
      "learning_rate": 2.0498502462053913e-06,
      "loss": 0.0256,
      "step": 53040
    },
    {
      "epoch": 2.692776283060054,
      "grad_norm": 0.28184401988983154,
      "learning_rate": 2.048158112932975e-06,
      "loss": 0.0293,
      "step": 53045
    },
    {
      "epoch": 2.6930301030509165,
      "grad_norm": 0.42434102296829224,
      "learning_rate": 2.0464659796605584e-06,
      "loss": 0.035,
      "step": 53050
    },
    {
      "epoch": 2.6932839230417787,
      "grad_norm": 0.3815975785255432,
      "learning_rate": 2.0447738463881415e-06,
      "loss": 0.0282,
      "step": 53055
    },
    {
      "epoch": 2.6935377430326413,
      "grad_norm": 0.25539329648017883,
      "learning_rate": 2.043081713115725e-06,
      "loss": 0.027,
      "step": 53060
    },
    {
      "epoch": 2.693791563023504,
      "grad_norm": 0.36869537830352783,
      "learning_rate": 2.0413895798433086e-06,
      "loss": 0.033,
      "step": 53065
    },
    {
      "epoch": 2.694045383014366,
      "grad_norm": 0.5211175680160522,
      "learning_rate": 2.039697446570892e-06,
      "loss": 0.0354,
      "step": 53070
    },
    {
      "epoch": 2.6942992030052286,
      "grad_norm": 0.4958113431930542,
      "learning_rate": 2.0380053132984757e-06,
      "loss": 0.0351,
      "step": 53075
    },
    {
      "epoch": 2.6945530229960912,
      "grad_norm": 0.38227608799934387,
      "learning_rate": 2.0363131800260592e-06,
      "loss": 0.0254,
      "step": 53080
    },
    {
      "epoch": 2.694806842986954,
      "grad_norm": 0.31782981753349304,
      "learning_rate": 2.0346210467536428e-06,
      "loss": 0.0357,
      "step": 53085
    },
    {
      "epoch": 2.695060662977816,
      "grad_norm": 0.344621866941452,
      "learning_rate": 2.032928913481226e-06,
      "loss": 0.0262,
      "step": 53090
    },
    {
      "epoch": 2.6953144829686786,
      "grad_norm": 0.2928641438484192,
      "learning_rate": 2.0312367802088094e-06,
      "loss": 0.0282,
      "step": 53095
    },
    {
      "epoch": 2.695568302959541,
      "grad_norm": 0.3690706193447113,
      "learning_rate": 2.0295446469363926e-06,
      "loss": 0.0326,
      "step": 53100
    },
    {
      "epoch": 2.6958221229504034,
      "grad_norm": 0.38905826210975647,
      "learning_rate": 2.027852513663976e-06,
      "loss": 0.0347,
      "step": 53105
    },
    {
      "epoch": 2.696075942941266,
      "grad_norm": 0.4815005660057068,
      "learning_rate": 2.0261603803915596e-06,
      "loss": 0.0324,
      "step": 53110
    },
    {
      "epoch": 2.6963297629321286,
      "grad_norm": 0.36236539483070374,
      "learning_rate": 2.024468247119143e-06,
      "loss": 0.0341,
      "step": 53115
    },
    {
      "epoch": 2.696583582922991,
      "grad_norm": 0.2581886947154999,
      "learning_rate": 2.0227761138467267e-06,
      "loss": 0.0253,
      "step": 53120
    },
    {
      "epoch": 2.6968374029138538,
      "grad_norm": 0.698701798915863,
      "learning_rate": 2.0210839805743103e-06,
      "loss": 0.0246,
      "step": 53125
    },
    {
      "epoch": 2.697091222904716,
      "grad_norm": 0.3778875172138214,
      "learning_rate": 2.019391847301894e-06,
      "loss": 0.0335,
      "step": 53130
    },
    {
      "epoch": 2.6973450428955785,
      "grad_norm": 0.3054998219013214,
      "learning_rate": 2.017699714029477e-06,
      "loss": 0.0272,
      "step": 53135
    },
    {
      "epoch": 2.6975988628864407,
      "grad_norm": 0.3761795163154602,
      "learning_rate": 2.0160075807570605e-06,
      "loss": 0.0313,
      "step": 53140
    },
    {
      "epoch": 2.6978526828773033,
      "grad_norm": 0.41128095984458923,
      "learning_rate": 2.014315447484644e-06,
      "loss": 0.0288,
      "step": 53145
    },
    {
      "epoch": 2.698106502868166,
      "grad_norm": 0.287852942943573,
      "learning_rate": 2.0126233142122276e-06,
      "loss": 0.0267,
      "step": 53150
    },
    {
      "epoch": 2.6983603228590285,
      "grad_norm": 0.32598933577537537,
      "learning_rate": 2.010931180939811e-06,
      "loss": 0.0271,
      "step": 53155
    },
    {
      "epoch": 2.698614142849891,
      "grad_norm": 0.4211830496788025,
      "learning_rate": 2.0092390476673947e-06,
      "loss": 0.0277,
      "step": 53160
    },
    {
      "epoch": 2.6988679628407533,
      "grad_norm": 0.308586448431015,
      "learning_rate": 2.0075469143949782e-06,
      "loss": 0.031,
      "step": 53165
    },
    {
      "epoch": 2.699121782831616,
      "grad_norm": 0.304868221282959,
      "learning_rate": 2.0058547811225613e-06,
      "loss": 0.0377,
      "step": 53170
    },
    {
      "epoch": 2.6993756028224785,
      "grad_norm": 0.2767963707447052,
      "learning_rate": 2.004162647850145e-06,
      "loss": 0.0338,
      "step": 53175
    },
    {
      "epoch": 2.6996294228133406,
      "grad_norm": 0.2587690055370331,
      "learning_rate": 2.002470514577728e-06,
      "loss": 0.026,
      "step": 53180
    },
    {
      "epoch": 2.699883242804203,
      "grad_norm": 0.25689199566841125,
      "learning_rate": 2.0007783813053115e-06,
      "loss": 0.0292,
      "step": 53185
    },
    {
      "epoch": 2.700137062795066,
      "grad_norm": 0.36869364976882935,
      "learning_rate": 1.999086248032895e-06,
      "loss": 0.0282,
      "step": 53190
    },
    {
      "epoch": 2.7003908827859284,
      "grad_norm": 0.34846585988998413,
      "learning_rate": 1.9973941147604786e-06,
      "loss": 0.0334,
      "step": 53195
    },
    {
      "epoch": 2.7006447027767906,
      "grad_norm": 0.31279513239860535,
      "learning_rate": 1.995701981488062e-06,
      "loss": 0.0291,
      "step": 53200
    },
    {
      "epoch": 2.700898522767653,
      "grad_norm": 0.2938686013221741,
      "learning_rate": 1.9940098482156457e-06,
      "loss": 0.0273,
      "step": 53205
    },
    {
      "epoch": 2.701152342758516,
      "grad_norm": 0.4499298930168152,
      "learning_rate": 1.9923177149432293e-06,
      "loss": 0.0297,
      "step": 53210
    },
    {
      "epoch": 2.701406162749378,
      "grad_norm": 0.5888147950172424,
      "learning_rate": 1.9906255816708124e-06,
      "loss": 0.0384,
      "step": 53215
    },
    {
      "epoch": 2.7016599827402406,
      "grad_norm": 0.2620474100112915,
      "learning_rate": 1.988933448398396e-06,
      "loss": 0.0301,
      "step": 53220
    },
    {
      "epoch": 2.701913802731103,
      "grad_norm": 0.4047901928424835,
      "learning_rate": 1.9872413151259795e-06,
      "loss": 0.0282,
      "step": 53225
    },
    {
      "epoch": 2.7021676227219658,
      "grad_norm": 0.42640551924705505,
      "learning_rate": 1.985549181853563e-06,
      "loss": 0.036,
      "step": 53230
    },
    {
      "epoch": 2.702421442712828,
      "grad_norm": 0.28090572357177734,
      "learning_rate": 1.9838570485811466e-06,
      "loss": 0.0255,
      "step": 53235
    },
    {
      "epoch": 2.7026752627036905,
      "grad_norm": 0.3164530396461487,
      "learning_rate": 1.98216491530873e-06,
      "loss": 0.0327,
      "step": 53240
    },
    {
      "epoch": 2.702929082694553,
      "grad_norm": 0.38603898882865906,
      "learning_rate": 1.9804727820363132e-06,
      "loss": 0.0295,
      "step": 53245
    },
    {
      "epoch": 2.7031829026854153,
      "grad_norm": 0.3389573395252228,
      "learning_rate": 1.9787806487638968e-06,
      "loss": 0.0362,
      "step": 53250
    },
    {
      "epoch": 2.703436722676278,
      "grad_norm": 0.2195272594690323,
      "learning_rate": 1.9770885154914803e-06,
      "loss": 0.0273,
      "step": 53255
    },
    {
      "epoch": 2.7036905426671405,
      "grad_norm": 0.505377471446991,
      "learning_rate": 1.9753963822190634e-06,
      "loss": 0.0335,
      "step": 53260
    },
    {
      "epoch": 2.703944362658003,
      "grad_norm": 0.32349297404289246,
      "learning_rate": 1.973704248946647e-06,
      "loss": 0.0317,
      "step": 53265
    },
    {
      "epoch": 2.7041981826488657,
      "grad_norm": 0.27195101976394653,
      "learning_rate": 1.9720121156742305e-06,
      "loss": 0.032,
      "step": 53270
    },
    {
      "epoch": 2.704452002639728,
      "grad_norm": 0.28886640071868896,
      "learning_rate": 1.970319982401814e-06,
      "loss": 0.0289,
      "step": 53275
    },
    {
      "epoch": 2.7047058226305905,
      "grad_norm": 0.44799792766571045,
      "learning_rate": 1.9686278491293976e-06,
      "loss": 0.0301,
      "step": 53280
    },
    {
      "epoch": 2.7049596426214526,
      "grad_norm": 0.44857335090637207,
      "learning_rate": 1.966935715856981e-06,
      "loss": 0.0387,
      "step": 53285
    },
    {
      "epoch": 2.705213462612315,
      "grad_norm": 0.29437506198883057,
      "learning_rate": 1.9652435825845643e-06,
      "loss": 0.0274,
      "step": 53290
    },
    {
      "epoch": 2.705467282603178,
      "grad_norm": 0.309415727853775,
      "learning_rate": 1.963551449312148e-06,
      "loss": 0.0368,
      "step": 53295
    },
    {
      "epoch": 2.7057211025940404,
      "grad_norm": 0.4497421979904175,
      "learning_rate": 1.9618593160397314e-06,
      "loss": 0.0299,
      "step": 53300
    },
    {
      "epoch": 2.705974922584903,
      "grad_norm": 0.3225899934768677,
      "learning_rate": 1.960167182767315e-06,
      "loss": 0.0273,
      "step": 53305
    },
    {
      "epoch": 2.706228742575765,
      "grad_norm": 0.34908631443977356,
      "learning_rate": 1.9584750494948985e-06,
      "loss": 0.033,
      "step": 53310
    },
    {
      "epoch": 2.706482562566628,
      "grad_norm": 0.33565932512283325,
      "learning_rate": 1.956782916222482e-06,
      "loss": 0.0309,
      "step": 53315
    },
    {
      "epoch": 2.7067363825574904,
      "grad_norm": 0.5139685273170471,
      "learning_rate": 1.9550907829500656e-06,
      "loss": 0.0297,
      "step": 53320
    },
    {
      "epoch": 2.7069902025483525,
      "grad_norm": 0.30862730741500854,
      "learning_rate": 1.9533986496776487e-06,
      "loss": 0.0273,
      "step": 53325
    },
    {
      "epoch": 2.707244022539215,
      "grad_norm": 0.4228881895542145,
      "learning_rate": 1.9517065164052322e-06,
      "loss": 0.0252,
      "step": 53330
    },
    {
      "epoch": 2.7074978425300777,
      "grad_norm": 0.278484970331192,
      "learning_rate": 1.9500143831328158e-06,
      "loss": 0.0254,
      "step": 53335
    },
    {
      "epoch": 2.7077516625209404,
      "grad_norm": 0.37600430846214294,
      "learning_rate": 1.948322249860399e-06,
      "loss": 0.0273,
      "step": 53340
    },
    {
      "epoch": 2.7080054825118025,
      "grad_norm": 0.20989076793193817,
      "learning_rate": 1.9466301165879824e-06,
      "loss": 0.0306,
      "step": 53345
    },
    {
      "epoch": 2.708259302502665,
      "grad_norm": 0.5742992162704468,
      "learning_rate": 1.944937983315566e-06,
      "loss": 0.0307,
      "step": 53350
    },
    {
      "epoch": 2.7085131224935277,
      "grad_norm": 0.28505074977874756,
      "learning_rate": 1.9432458500431495e-06,
      "loss": 0.0269,
      "step": 53355
    },
    {
      "epoch": 2.70876694248439,
      "grad_norm": 0.27916303277015686,
      "learning_rate": 1.941553716770733e-06,
      "loss": 0.0357,
      "step": 53360
    },
    {
      "epoch": 2.7090207624752525,
      "grad_norm": 0.261547714471817,
      "learning_rate": 1.9398615834983166e-06,
      "loss": 0.0297,
      "step": 53365
    },
    {
      "epoch": 2.709274582466115,
      "grad_norm": 0.370958149433136,
      "learning_rate": 1.9381694502258997e-06,
      "loss": 0.0266,
      "step": 53370
    },
    {
      "epoch": 2.7095284024569777,
      "grad_norm": 0.3699497580528259,
      "learning_rate": 1.9364773169534833e-06,
      "loss": 0.0305,
      "step": 53375
    },
    {
      "epoch": 2.70978222244784,
      "grad_norm": 0.2690362334251404,
      "learning_rate": 1.934785183681067e-06,
      "loss": 0.0262,
      "step": 53380
    },
    {
      "epoch": 2.7100360424387024,
      "grad_norm": 1.8637694120407104,
      "learning_rate": 1.9330930504086504e-06,
      "loss": 0.0365,
      "step": 53385
    },
    {
      "epoch": 2.710289862429565,
      "grad_norm": 0.32636716961860657,
      "learning_rate": 1.931400917136234e-06,
      "loss": 0.0338,
      "step": 53390
    },
    {
      "epoch": 2.710543682420427,
      "grad_norm": 0.3575417697429657,
      "learning_rate": 1.9297087838638175e-06,
      "loss": 0.031,
      "step": 53395
    },
    {
      "epoch": 2.71079750241129,
      "grad_norm": 0.3870401084423065,
      "learning_rate": 1.928016650591401e-06,
      "loss": 0.0277,
      "step": 53400
    },
    {
      "epoch": 2.7110513224021524,
      "grad_norm": 0.43011873960494995,
      "learning_rate": 1.926324517318984e-06,
      "loss": 0.0332,
      "step": 53405
    },
    {
      "epoch": 2.711305142393015,
      "grad_norm": 0.21395263075828552,
      "learning_rate": 1.9246323840465677e-06,
      "loss": 0.0241,
      "step": 53410
    },
    {
      "epoch": 2.7115589623838776,
      "grad_norm": 0.43874168395996094,
      "learning_rate": 1.922940250774151e-06,
      "loss": 0.0323,
      "step": 53415
    },
    {
      "epoch": 2.7118127823747398,
      "grad_norm": 0.32266518473625183,
      "learning_rate": 1.9212481175017343e-06,
      "loss": 0.0239,
      "step": 53420
    },
    {
      "epoch": 2.7120666023656024,
      "grad_norm": 0.34239718317985535,
      "learning_rate": 1.919555984229318e-06,
      "loss": 0.0363,
      "step": 53425
    },
    {
      "epoch": 2.7123204223564645,
      "grad_norm": 0.45117393136024475,
      "learning_rate": 1.9178638509569014e-06,
      "loss": 0.0298,
      "step": 53430
    },
    {
      "epoch": 2.712574242347327,
      "grad_norm": 0.3619763255119324,
      "learning_rate": 1.916171717684485e-06,
      "loss": 0.0256,
      "step": 53435
    },
    {
      "epoch": 2.7128280623381897,
      "grad_norm": 0.6177137494087219,
      "learning_rate": 1.9144795844120685e-06,
      "loss": 0.0242,
      "step": 53440
    },
    {
      "epoch": 2.7130818823290523,
      "grad_norm": 0.31061235070228577,
      "learning_rate": 1.912787451139652e-06,
      "loss": 0.0312,
      "step": 53445
    },
    {
      "epoch": 2.713335702319915,
      "grad_norm": 0.42823871970176697,
      "learning_rate": 1.911095317867235e-06,
      "loss": 0.0333,
      "step": 53450
    },
    {
      "epoch": 2.713589522310777,
      "grad_norm": 0.42373305559158325,
      "learning_rate": 1.9094031845948187e-06,
      "loss": 0.0257,
      "step": 53455
    },
    {
      "epoch": 2.7138433423016397,
      "grad_norm": 0.27779242396354675,
      "learning_rate": 1.9077110513224023e-06,
      "loss": 0.0247,
      "step": 53460
    },
    {
      "epoch": 2.7140971622925023,
      "grad_norm": 0.22847887873649597,
      "learning_rate": 1.9060189180499858e-06,
      "loss": 0.0272,
      "step": 53465
    },
    {
      "epoch": 2.7143509822833645,
      "grad_norm": 0.3980698883533478,
      "learning_rate": 1.9043267847775694e-06,
      "loss": 0.0298,
      "step": 53470
    },
    {
      "epoch": 2.714604802274227,
      "grad_norm": 0.2841477692127228,
      "learning_rate": 1.9026346515051527e-06,
      "loss": 0.0258,
      "step": 53475
    },
    {
      "epoch": 2.7148586222650897,
      "grad_norm": 0.38664472103118896,
      "learning_rate": 1.9009425182327362e-06,
      "loss": 0.03,
      "step": 53480
    },
    {
      "epoch": 2.7151124422559523,
      "grad_norm": 0.21626536548137665,
      "learning_rate": 1.8992503849603198e-06,
      "loss": 0.0284,
      "step": 53485
    },
    {
      "epoch": 2.7153662622468144,
      "grad_norm": 0.31044235825538635,
      "learning_rate": 1.897558251687903e-06,
      "loss": 0.0274,
      "step": 53490
    },
    {
      "epoch": 2.715620082237677,
      "grad_norm": 0.35324153304100037,
      "learning_rate": 1.8958661184154866e-06,
      "loss": 0.0274,
      "step": 53495
    },
    {
      "epoch": 2.7158739022285396,
      "grad_norm": 0.4163343608379364,
      "learning_rate": 1.8941739851430702e-06,
      "loss": 0.0293,
      "step": 53500
    },
    {
      "epoch": 2.716127722219402,
      "grad_norm": 0.561988890171051,
      "learning_rate": 1.8924818518706533e-06,
      "loss": 0.0295,
      "step": 53505
    },
    {
      "epoch": 2.7163815422102644,
      "grad_norm": 0.30611640214920044,
      "learning_rate": 1.8907897185982369e-06,
      "loss": 0.0317,
      "step": 53510
    },
    {
      "epoch": 2.716635362201127,
      "grad_norm": 0.3293723464012146,
      "learning_rate": 1.8890975853258204e-06,
      "loss": 0.0299,
      "step": 53515
    },
    {
      "epoch": 2.7168891821919896,
      "grad_norm": 1.3542070388793945,
      "learning_rate": 1.8874054520534037e-06,
      "loss": 0.0356,
      "step": 53520
    },
    {
      "epoch": 2.7171430021828518,
      "grad_norm": 0.25533461570739746,
      "learning_rate": 1.8857133187809873e-06,
      "loss": 0.035,
      "step": 53525
    },
    {
      "epoch": 2.7173968221737144,
      "grad_norm": 0.2728980481624603,
      "learning_rate": 1.8840211855085708e-06,
      "loss": 0.0321,
      "step": 53530
    },
    {
      "epoch": 2.717650642164577,
      "grad_norm": 0.3773408830165863,
      "learning_rate": 1.8823290522361542e-06,
      "loss": 0.0306,
      "step": 53535
    },
    {
      "epoch": 2.717904462155439,
      "grad_norm": 0.42744117975234985,
      "learning_rate": 1.8806369189637377e-06,
      "loss": 0.0274,
      "step": 53540
    },
    {
      "epoch": 2.7181582821463017,
      "grad_norm": 0.5459184646606445,
      "learning_rate": 1.8789447856913212e-06,
      "loss": 0.0354,
      "step": 53545
    },
    {
      "epoch": 2.7184121021371643,
      "grad_norm": 0.40570247173309326,
      "learning_rate": 1.8772526524189048e-06,
      "loss": 0.0303,
      "step": 53550
    },
    {
      "epoch": 2.718665922128027,
      "grad_norm": 0.18053355813026428,
      "learning_rate": 1.8755605191464881e-06,
      "loss": 0.0224,
      "step": 53555
    },
    {
      "epoch": 2.7189197421188895,
      "grad_norm": 0.3330439329147339,
      "learning_rate": 1.8738683858740717e-06,
      "loss": 0.0277,
      "step": 53560
    },
    {
      "epoch": 2.7191735621097517,
      "grad_norm": 0.3566732108592987,
      "learning_rate": 1.8721762526016552e-06,
      "loss": 0.0321,
      "step": 53565
    },
    {
      "epoch": 2.7194273821006143,
      "grad_norm": 0.2924792468547821,
      "learning_rate": 1.8704841193292385e-06,
      "loss": 0.0279,
      "step": 53570
    },
    {
      "epoch": 2.7196812020914765,
      "grad_norm": 0.35724127292633057,
      "learning_rate": 1.868791986056822e-06,
      "loss": 0.0277,
      "step": 53575
    },
    {
      "epoch": 2.719935022082339,
      "grad_norm": 0.4432433247566223,
      "learning_rate": 1.8670998527844056e-06,
      "loss": 0.035,
      "step": 53580
    },
    {
      "epoch": 2.7201888420732017,
      "grad_norm": 0.3439810276031494,
      "learning_rate": 1.8654077195119888e-06,
      "loss": 0.0321,
      "step": 53585
    },
    {
      "epoch": 2.7204426620640643,
      "grad_norm": 0.2902897298336029,
      "learning_rate": 1.8637155862395723e-06,
      "loss": 0.0339,
      "step": 53590
    },
    {
      "epoch": 2.720696482054927,
      "grad_norm": 0.565328061580658,
      "learning_rate": 1.8620234529671558e-06,
      "loss": 0.0368,
      "step": 53595
    },
    {
      "epoch": 2.720950302045789,
      "grad_norm": 0.23699818551540375,
      "learning_rate": 1.8603313196947392e-06,
      "loss": 0.0275,
      "step": 53600
    },
    {
      "epoch": 2.7212041220366516,
      "grad_norm": 0.4020783305168152,
      "learning_rate": 1.8586391864223227e-06,
      "loss": 0.0244,
      "step": 53605
    },
    {
      "epoch": 2.7214579420275142,
      "grad_norm": 0.6224753856658936,
      "learning_rate": 1.8569470531499063e-06,
      "loss": 0.03,
      "step": 53610
    },
    {
      "epoch": 2.7217117620183764,
      "grad_norm": 0.21802890300750732,
      "learning_rate": 1.8552549198774896e-06,
      "loss": 0.0261,
      "step": 53615
    },
    {
      "epoch": 2.721965582009239,
      "grad_norm": 0.25149235129356384,
      "learning_rate": 1.8535627866050731e-06,
      "loss": 0.0287,
      "step": 53620
    },
    {
      "epoch": 2.7222194020001016,
      "grad_norm": 0.2931320071220398,
      "learning_rate": 1.8518706533326567e-06,
      "loss": 0.0324,
      "step": 53625
    },
    {
      "epoch": 2.722473221990964,
      "grad_norm": 0.35715875029563904,
      "learning_rate": 1.85017852006024e-06,
      "loss": 0.0369,
      "step": 53630
    },
    {
      "epoch": 2.7227270419818264,
      "grad_norm": 0.31633713841438293,
      "learning_rate": 1.8484863867878236e-06,
      "loss": 0.0287,
      "step": 53635
    },
    {
      "epoch": 2.722980861972689,
      "grad_norm": 0.4361322522163391,
      "learning_rate": 1.8467942535154071e-06,
      "loss": 0.0323,
      "step": 53640
    },
    {
      "epoch": 2.7232346819635516,
      "grad_norm": 0.26381734013557434,
      "learning_rate": 1.8451021202429907e-06,
      "loss": 0.0274,
      "step": 53645
    },
    {
      "epoch": 2.7234885019544137,
      "grad_norm": 0.3084844946861267,
      "learning_rate": 1.843409986970574e-06,
      "loss": 0.0297,
      "step": 53650
    },
    {
      "epoch": 2.7237423219452763,
      "grad_norm": 0.6224455833435059,
      "learning_rate": 1.8417178536981575e-06,
      "loss": 0.0349,
      "step": 53655
    },
    {
      "epoch": 2.723996141936139,
      "grad_norm": 0.44383668899536133,
      "learning_rate": 1.840025720425741e-06,
      "loss": 0.0309,
      "step": 53660
    },
    {
      "epoch": 2.7242499619270015,
      "grad_norm": 0.4933556914329529,
      "learning_rate": 1.8383335871533242e-06,
      "loss": 0.036,
      "step": 53665
    },
    {
      "epoch": 2.7245037819178637,
      "grad_norm": 0.23321112990379333,
      "learning_rate": 1.8366414538809077e-06,
      "loss": 0.0227,
      "step": 53670
    },
    {
      "epoch": 2.7247576019087263,
      "grad_norm": 0.3232824504375458,
      "learning_rate": 1.834949320608491e-06,
      "loss": 0.0339,
      "step": 53675
    },
    {
      "epoch": 2.725011421899589,
      "grad_norm": 0.2447776347398758,
      "learning_rate": 1.8332571873360746e-06,
      "loss": 0.0308,
      "step": 53680
    },
    {
      "epoch": 2.725265241890451,
      "grad_norm": 0.26831647753715515,
      "learning_rate": 1.8315650540636582e-06,
      "loss": 0.03,
      "step": 53685
    },
    {
      "epoch": 2.7255190618813137,
      "grad_norm": 0.2714548110961914,
      "learning_rate": 1.8298729207912417e-06,
      "loss": 0.0315,
      "step": 53690
    },
    {
      "epoch": 2.7257728818721763,
      "grad_norm": 0.2372501641511917,
      "learning_rate": 1.828180787518825e-06,
      "loss": 0.0254,
      "step": 53695
    },
    {
      "epoch": 2.726026701863039,
      "grad_norm": 0.18035678565502167,
      "learning_rate": 1.8264886542464086e-06,
      "loss": 0.0257,
      "step": 53700
    },
    {
      "epoch": 2.7262805218539015,
      "grad_norm": 0.3743029832839966,
      "learning_rate": 1.8247965209739921e-06,
      "loss": 0.0365,
      "step": 53705
    },
    {
      "epoch": 2.7265343418447636,
      "grad_norm": 0.4348129630088806,
      "learning_rate": 1.8231043877015755e-06,
      "loss": 0.0311,
      "step": 53710
    },
    {
      "epoch": 2.7267881618356262,
      "grad_norm": 0.3668404817581177,
      "learning_rate": 1.821412254429159e-06,
      "loss": 0.0312,
      "step": 53715
    },
    {
      "epoch": 2.7270419818264884,
      "grad_norm": 0.35408517718315125,
      "learning_rate": 1.8197201211567426e-06,
      "loss": 0.0291,
      "step": 53720
    },
    {
      "epoch": 2.727295801817351,
      "grad_norm": 0.32698673009872437,
      "learning_rate": 1.818027987884326e-06,
      "loss": 0.0252,
      "step": 53725
    },
    {
      "epoch": 2.7275496218082136,
      "grad_norm": 0.4094547927379608,
      "learning_rate": 1.8163358546119094e-06,
      "loss": 0.0381,
      "step": 53730
    },
    {
      "epoch": 2.727803441799076,
      "grad_norm": 0.38025033473968506,
      "learning_rate": 1.814643721339493e-06,
      "loss": 0.0291,
      "step": 53735
    },
    {
      "epoch": 2.728057261789939,
      "grad_norm": 0.45479652285575867,
      "learning_rate": 1.8129515880670765e-06,
      "loss": 0.034,
      "step": 53740
    },
    {
      "epoch": 2.728311081780801,
      "grad_norm": 0.726309061050415,
      "learning_rate": 1.8112594547946596e-06,
      "loss": 0.0354,
      "step": 53745
    },
    {
      "epoch": 2.7285649017716636,
      "grad_norm": 0.30821505188941956,
      "learning_rate": 1.8095673215222432e-06,
      "loss": 0.031,
      "step": 53750
    },
    {
      "epoch": 2.728818721762526,
      "grad_norm": 0.2646833658218384,
      "learning_rate": 1.8078751882498265e-06,
      "loss": 0.0285,
      "step": 53755
    },
    {
      "epoch": 2.7290725417533883,
      "grad_norm": 0.32509294152259827,
      "learning_rate": 1.80618305497741e-06,
      "loss": 0.0295,
      "step": 53760
    },
    {
      "epoch": 2.729326361744251,
      "grad_norm": 0.2838088274002075,
      "learning_rate": 1.8044909217049936e-06,
      "loss": 0.0293,
      "step": 53765
    },
    {
      "epoch": 2.7295801817351135,
      "grad_norm": 0.34795206785202026,
      "learning_rate": 1.8027987884325772e-06,
      "loss": 0.0296,
      "step": 53770
    },
    {
      "epoch": 2.729834001725976,
      "grad_norm": 0.17932498455047607,
      "learning_rate": 1.8011066551601605e-06,
      "loss": 0.0278,
      "step": 53775
    },
    {
      "epoch": 2.7300878217168383,
      "grad_norm": 0.29193925857543945,
      "learning_rate": 1.799414521887744e-06,
      "loss": 0.0348,
      "step": 53780
    },
    {
      "epoch": 2.730341641707701,
      "grad_norm": 1.6499922275543213,
      "learning_rate": 1.7977223886153276e-06,
      "loss": 0.0318,
      "step": 53785
    },
    {
      "epoch": 2.7305954616985635,
      "grad_norm": 0.2674989402294159,
      "learning_rate": 1.796030255342911e-06,
      "loss": 0.0301,
      "step": 53790
    },
    {
      "epoch": 2.7308492816894256,
      "grad_norm": 0.3045971691608429,
      "learning_rate": 1.7943381220704945e-06,
      "loss": 0.0286,
      "step": 53795
    },
    {
      "epoch": 2.7311031016802882,
      "grad_norm": 0.25646665692329407,
      "learning_rate": 1.792645988798078e-06,
      "loss": 0.0254,
      "step": 53800
    },
    {
      "epoch": 2.731356921671151,
      "grad_norm": 0.2805609703063965,
      "learning_rate": 1.7909538555256613e-06,
      "loss": 0.0326,
      "step": 53805
    },
    {
      "epoch": 2.7316107416620135,
      "grad_norm": 0.3846939504146576,
      "learning_rate": 1.7892617222532449e-06,
      "loss": 0.0287,
      "step": 53810
    },
    {
      "epoch": 2.731864561652876,
      "grad_norm": 0.30910441279411316,
      "learning_rate": 1.7875695889808284e-06,
      "loss": 0.0338,
      "step": 53815
    },
    {
      "epoch": 2.732118381643738,
      "grad_norm": 0.26713868975639343,
      "learning_rate": 1.785877455708412e-06,
      "loss": 0.0244,
      "step": 53820
    },
    {
      "epoch": 2.732372201634601,
      "grad_norm": 0.3901999592781067,
      "learning_rate": 1.784185322435995e-06,
      "loss": 0.0247,
      "step": 53825
    },
    {
      "epoch": 2.732626021625463,
      "grad_norm": 0.260437935590744,
      "learning_rate": 1.7824931891635786e-06,
      "loss": 0.0339,
      "step": 53830
    },
    {
      "epoch": 2.7328798416163256,
      "grad_norm": 0.418204665184021,
      "learning_rate": 1.780801055891162e-06,
      "loss": 0.028,
      "step": 53835
    },
    {
      "epoch": 2.733133661607188,
      "grad_norm": 0.32420992851257324,
      "learning_rate": 1.7791089226187455e-06,
      "loss": 0.0309,
      "step": 53840
    },
    {
      "epoch": 2.733387481598051,
      "grad_norm": 0.32176563143730164,
      "learning_rate": 1.777416789346329e-06,
      "loss": 0.025,
      "step": 53845
    },
    {
      "epoch": 2.7336413015889134,
      "grad_norm": 0.332557737827301,
      "learning_rate": 1.7757246560739124e-06,
      "loss": 0.032,
      "step": 53850
    },
    {
      "epoch": 2.7338951215797755,
      "grad_norm": 0.31119346618652344,
      "learning_rate": 1.774032522801496e-06,
      "loss": 0.0231,
      "step": 53855
    },
    {
      "epoch": 2.734148941570638,
      "grad_norm": 0.33499324321746826,
      "learning_rate": 1.7723403895290795e-06,
      "loss": 0.0305,
      "step": 53860
    },
    {
      "epoch": 2.7344027615615007,
      "grad_norm": 0.4173754155635834,
      "learning_rate": 1.770648256256663e-06,
      "loss": 0.0289,
      "step": 53865
    },
    {
      "epoch": 2.734656581552363,
      "grad_norm": 0.32819196581840515,
      "learning_rate": 1.7689561229842463e-06,
      "loss": 0.0312,
      "step": 53870
    },
    {
      "epoch": 2.7349104015432255,
      "grad_norm": 0.3945378363132477,
      "learning_rate": 1.7672639897118299e-06,
      "loss": 0.0383,
      "step": 53875
    },
    {
      "epoch": 2.735164221534088,
      "grad_norm": 0.34407007694244385,
      "learning_rate": 1.7655718564394134e-06,
      "loss": 0.0313,
      "step": 53880
    },
    {
      "epoch": 2.7354180415249507,
      "grad_norm": 0.37155938148498535,
      "learning_rate": 1.7638797231669968e-06,
      "loss": 0.0348,
      "step": 53885
    },
    {
      "epoch": 2.735671861515813,
      "grad_norm": 0.3493196666240692,
      "learning_rate": 1.7621875898945803e-06,
      "loss": 0.0262,
      "step": 53890
    },
    {
      "epoch": 2.7359256815066755,
      "grad_norm": 0.39568087458610535,
      "learning_rate": 1.7604954566221639e-06,
      "loss": 0.0327,
      "step": 53895
    },
    {
      "epoch": 2.736179501497538,
      "grad_norm": 0.3887282907962799,
      "learning_rate": 1.7588033233497472e-06,
      "loss": 0.0314,
      "step": 53900
    },
    {
      "epoch": 2.7364333214884002,
      "grad_norm": 0.2152509093284607,
      "learning_rate": 1.7571111900773305e-06,
      "loss": 0.0319,
      "step": 53905
    },
    {
      "epoch": 2.736687141479263,
      "grad_norm": 0.24241326749324799,
      "learning_rate": 1.755419056804914e-06,
      "loss": 0.0241,
      "step": 53910
    },
    {
      "epoch": 2.7369409614701254,
      "grad_norm": 0.6015492677688599,
      "learning_rate": 1.7537269235324974e-06,
      "loss": 0.0287,
      "step": 53915
    },
    {
      "epoch": 2.737194781460988,
      "grad_norm": 0.7322681546211243,
      "learning_rate": 1.752034790260081e-06,
      "loss": 0.0336,
      "step": 53920
    },
    {
      "epoch": 2.73744860145185,
      "grad_norm": 0.27525922656059265,
      "learning_rate": 1.7503426569876645e-06,
      "loss": 0.0284,
      "step": 53925
    },
    {
      "epoch": 2.737702421442713,
      "grad_norm": 0.24374821782112122,
      "learning_rate": 1.7486505237152478e-06,
      "loss": 0.035,
      "step": 53930
    },
    {
      "epoch": 2.7379562414335754,
      "grad_norm": 0.2927287518978119,
      "learning_rate": 1.7469583904428314e-06,
      "loss": 0.0293,
      "step": 53935
    },
    {
      "epoch": 2.7382100614244376,
      "grad_norm": 0.44183194637298584,
      "learning_rate": 1.745266257170415e-06,
      "loss": 0.0257,
      "step": 53940
    },
    {
      "epoch": 2.7384638814153,
      "grad_norm": 0.970026433467865,
      "learning_rate": 1.7435741238979982e-06,
      "loss": 0.0286,
      "step": 53945
    },
    {
      "epoch": 2.7387177014061628,
      "grad_norm": 0.5048357844352722,
      "learning_rate": 1.7418819906255818e-06,
      "loss": 0.0292,
      "step": 53950
    },
    {
      "epoch": 2.7389715213970254,
      "grad_norm": 0.46825069189071655,
      "learning_rate": 1.7401898573531653e-06,
      "loss": 0.0404,
      "step": 53955
    },
    {
      "epoch": 2.739225341387888,
      "grad_norm": 0.3677990138530731,
      "learning_rate": 1.7384977240807489e-06,
      "loss": 0.0266,
      "step": 53960
    },
    {
      "epoch": 2.73947916137875,
      "grad_norm": 0.312150776386261,
      "learning_rate": 1.7368055908083322e-06,
      "loss": 0.028,
      "step": 53965
    },
    {
      "epoch": 2.7397329813696127,
      "grad_norm": 0.3144506812095642,
      "learning_rate": 1.7351134575359158e-06,
      "loss": 0.026,
      "step": 53970
    },
    {
      "epoch": 2.739986801360475,
      "grad_norm": 1.072723627090454,
      "learning_rate": 1.7334213242634993e-06,
      "loss": 0.0408,
      "step": 53975
    },
    {
      "epoch": 2.7402406213513375,
      "grad_norm": 0.2998804748058319,
      "learning_rate": 1.7317291909910826e-06,
      "loss": 0.0257,
      "step": 53980
    },
    {
      "epoch": 2.7404944413422,
      "grad_norm": 0.3485616147518158,
      "learning_rate": 1.7300370577186662e-06,
      "loss": 0.0289,
      "step": 53985
    },
    {
      "epoch": 2.7407482613330627,
      "grad_norm": 0.36675891280174255,
      "learning_rate": 1.7283449244462493e-06,
      "loss": 0.0333,
      "step": 53990
    },
    {
      "epoch": 2.7410020813239253,
      "grad_norm": 0.48393315076828003,
      "learning_rate": 1.7266527911738328e-06,
      "loss": 0.0316,
      "step": 53995
    },
    {
      "epoch": 2.7412559013147875,
      "grad_norm": 0.35300666093826294,
      "learning_rate": 1.7249606579014164e-06,
      "loss": 0.0271,
      "step": 54000
    },
    {
      "epoch": 2.74150972130565,
      "grad_norm": 0.2933576703071594,
      "learning_rate": 1.723268524629e-06,
      "loss": 0.0377,
      "step": 54005
    },
    {
      "epoch": 2.7417635412965127,
      "grad_norm": 0.4027237296104431,
      "learning_rate": 1.7215763913565833e-06,
      "loss": 0.0307,
      "step": 54010
    },
    {
      "epoch": 2.742017361287375,
      "grad_norm": 0.2828601598739624,
      "learning_rate": 1.7198842580841668e-06,
      "loss": 0.0271,
      "step": 54015
    },
    {
      "epoch": 2.7422711812782374,
      "grad_norm": 0.3389827609062195,
      "learning_rate": 1.7181921248117504e-06,
      "loss": 0.0276,
      "step": 54020
    },
    {
      "epoch": 2.7425250012691,
      "grad_norm": 0.5130777955055237,
      "learning_rate": 1.7164999915393337e-06,
      "loss": 0.0279,
      "step": 54025
    },
    {
      "epoch": 2.7427788212599626,
      "grad_norm": 0.32711169123649597,
      "learning_rate": 1.7148078582669172e-06,
      "loss": 0.0328,
      "step": 54030
    },
    {
      "epoch": 2.743032641250825,
      "grad_norm": 0.45256665349006653,
      "learning_rate": 1.7131157249945008e-06,
      "loss": 0.0365,
      "step": 54035
    },
    {
      "epoch": 2.7432864612416874,
      "grad_norm": 0.27658596634864807,
      "learning_rate": 1.7114235917220843e-06,
      "loss": 0.0289,
      "step": 54040
    },
    {
      "epoch": 2.74354028123255,
      "grad_norm": 0.35465267300605774,
      "learning_rate": 1.7097314584496677e-06,
      "loss": 0.0401,
      "step": 54045
    },
    {
      "epoch": 2.743794101223412,
      "grad_norm": 0.2526685893535614,
      "learning_rate": 1.7080393251772512e-06,
      "loss": 0.0335,
      "step": 54050
    },
    {
      "epoch": 2.7440479212142748,
      "grad_norm": 0.43355873227119446,
      "learning_rate": 1.7063471919048347e-06,
      "loss": 0.03,
      "step": 54055
    },
    {
      "epoch": 2.7443017412051374,
      "grad_norm": 0.2964344918727875,
      "learning_rate": 1.704655058632418e-06,
      "loss": 0.0299,
      "step": 54060
    },
    {
      "epoch": 2.744555561196,
      "grad_norm": 0.46827706694602966,
      "learning_rate": 1.7029629253600016e-06,
      "loss": 0.0427,
      "step": 54065
    },
    {
      "epoch": 2.744809381186862,
      "grad_norm": 1.2639143466949463,
      "learning_rate": 1.7012707920875847e-06,
      "loss": 0.027,
      "step": 54070
    },
    {
      "epoch": 2.7450632011777247,
      "grad_norm": 0.2572653293609619,
      "learning_rate": 1.6995786588151683e-06,
      "loss": 0.0322,
      "step": 54075
    },
    {
      "epoch": 2.7453170211685873,
      "grad_norm": 0.37225455045700073,
      "learning_rate": 1.6978865255427518e-06,
      "loss": 0.0297,
      "step": 54080
    },
    {
      "epoch": 2.7455708411594495,
      "grad_norm": 0.3000081777572632,
      "learning_rate": 1.6961943922703354e-06,
      "loss": 0.0344,
      "step": 54085
    },
    {
      "epoch": 2.745824661150312,
      "grad_norm": 0.36823397874832153,
      "learning_rate": 1.6945022589979187e-06,
      "loss": 0.0302,
      "step": 54090
    },
    {
      "epoch": 2.7460784811411747,
      "grad_norm": 0.37122178077697754,
      "learning_rate": 1.6928101257255023e-06,
      "loss": 0.0316,
      "step": 54095
    },
    {
      "epoch": 2.7463323011320373,
      "grad_norm": 0.3983829617500305,
      "learning_rate": 1.6911179924530858e-06,
      "loss": 0.0296,
      "step": 54100
    },
    {
      "epoch": 2.7465861211229,
      "grad_norm": 0.40506190061569214,
      "learning_rate": 1.6894258591806691e-06,
      "loss": 0.031,
      "step": 54105
    },
    {
      "epoch": 2.746839941113762,
      "grad_norm": 0.45326289534568787,
      "learning_rate": 1.6877337259082527e-06,
      "loss": 0.0232,
      "step": 54110
    },
    {
      "epoch": 2.7470937611046247,
      "grad_norm": 0.23854999244213104,
      "learning_rate": 1.6860415926358362e-06,
      "loss": 0.0361,
      "step": 54115
    },
    {
      "epoch": 2.747347581095487,
      "grad_norm": 0.3053087890148163,
      "learning_rate": 1.6843494593634196e-06,
      "loss": 0.0328,
      "step": 54120
    },
    {
      "epoch": 2.7476014010863494,
      "grad_norm": 0.47979798913002014,
      "learning_rate": 1.682657326091003e-06,
      "loss": 0.0255,
      "step": 54125
    },
    {
      "epoch": 2.747855221077212,
      "grad_norm": 0.35441768169403076,
      "learning_rate": 1.6809651928185866e-06,
      "loss": 0.0294,
      "step": 54130
    },
    {
      "epoch": 2.7481090410680746,
      "grad_norm": 0.40479037165641785,
      "learning_rate": 1.6792730595461702e-06,
      "loss": 0.0313,
      "step": 54135
    },
    {
      "epoch": 2.7483628610589372,
      "grad_norm": 0.27825799584388733,
      "learning_rate": 1.6775809262737535e-06,
      "loss": 0.0245,
      "step": 54140
    },
    {
      "epoch": 2.7486166810497994,
      "grad_norm": 0.2868492901325226,
      "learning_rate": 1.675888793001337e-06,
      "loss": 0.0342,
      "step": 54145
    },
    {
      "epoch": 2.748870501040662,
      "grad_norm": 0.3349640667438507,
      "learning_rate": 1.6741966597289202e-06,
      "loss": 0.027,
      "step": 54150
    },
    {
      "epoch": 2.7491243210315246,
      "grad_norm": 0.29142335057258606,
      "learning_rate": 1.6725045264565037e-06,
      "loss": 0.0333,
      "step": 54155
    },
    {
      "epoch": 2.7493781410223868,
      "grad_norm": 0.327721506357193,
      "learning_rate": 1.6708123931840873e-06,
      "loss": 0.0307,
      "step": 54160
    },
    {
      "epoch": 2.7496319610132494,
      "grad_norm": 0.3892741799354553,
      "learning_rate": 1.6691202599116706e-06,
      "loss": 0.0313,
      "step": 54165
    },
    {
      "epoch": 2.749885781004112,
      "grad_norm": 0.3699101209640503,
      "learning_rate": 1.6674281266392542e-06,
      "loss": 0.0257,
      "step": 54170
    },
    {
      "epoch": 2.7501396009949746,
      "grad_norm": 0.30111369490623474,
      "learning_rate": 1.6657359933668377e-06,
      "loss": 0.032,
      "step": 54175
    },
    {
      "epoch": 2.7503934209858367,
      "grad_norm": 0.3970666229724884,
      "learning_rate": 1.6640438600944212e-06,
      "loss": 0.027,
      "step": 54180
    },
    {
      "epoch": 2.7506472409766993,
      "grad_norm": 0.600459635257721,
      "learning_rate": 1.6623517268220046e-06,
      "loss": 0.0373,
      "step": 54185
    },
    {
      "epoch": 2.750901060967562,
      "grad_norm": 0.5128913521766663,
      "learning_rate": 1.6606595935495881e-06,
      "loss": 0.0366,
      "step": 54190
    },
    {
      "epoch": 2.751154880958424,
      "grad_norm": 0.4742211103439331,
      "learning_rate": 1.6589674602771717e-06,
      "loss": 0.0414,
      "step": 54195
    },
    {
      "epoch": 2.7514087009492867,
      "grad_norm": 0.5436285734176636,
      "learning_rate": 1.657275327004755e-06,
      "loss": 0.0302,
      "step": 54200
    },
    {
      "epoch": 2.7516625209401493,
      "grad_norm": 0.19674985110759735,
      "learning_rate": 1.6555831937323385e-06,
      "loss": 0.0332,
      "step": 54205
    },
    {
      "epoch": 2.751916340931012,
      "grad_norm": 0.341109961271286,
      "learning_rate": 1.653891060459922e-06,
      "loss": 0.0322,
      "step": 54210
    },
    {
      "epoch": 2.752170160921874,
      "grad_norm": 0.2867673337459564,
      "learning_rate": 1.6521989271875054e-06,
      "loss": 0.0313,
      "step": 54215
    },
    {
      "epoch": 2.7524239809127367,
      "grad_norm": 0.24930787086486816,
      "learning_rate": 1.650506793915089e-06,
      "loss": 0.0294,
      "step": 54220
    },
    {
      "epoch": 2.7526778009035993,
      "grad_norm": 0.3115566074848175,
      "learning_rate": 1.6488146606426725e-06,
      "loss": 0.0261,
      "step": 54225
    },
    {
      "epoch": 2.7529316208944614,
      "grad_norm": 0.36223429441452026,
      "learning_rate": 1.6471225273702556e-06,
      "loss": 0.0422,
      "step": 54230
    },
    {
      "epoch": 2.753185440885324,
      "grad_norm": 0.3967058062553406,
      "learning_rate": 1.6454303940978392e-06,
      "loss": 0.0316,
      "step": 54235
    },
    {
      "epoch": 2.7534392608761866,
      "grad_norm": 0.2525790333747864,
      "learning_rate": 1.6437382608254227e-06,
      "loss": 0.029,
      "step": 54240
    },
    {
      "epoch": 2.7536930808670492,
      "grad_norm": 0.4637283384799957,
      "learning_rate": 1.642046127553006e-06,
      "loss": 0.0283,
      "step": 54245
    },
    {
      "epoch": 2.753946900857912,
      "grad_norm": 0.34502989053726196,
      "learning_rate": 1.6403539942805896e-06,
      "loss": 0.0317,
      "step": 54250
    },
    {
      "epoch": 2.754200720848774,
      "grad_norm": 0.2657451331615448,
      "learning_rate": 1.6386618610081731e-06,
      "loss": 0.0304,
      "step": 54255
    },
    {
      "epoch": 2.7544545408396366,
      "grad_norm": 0.2720555365085602,
      "learning_rate": 1.6369697277357565e-06,
      "loss": 0.0266,
      "step": 54260
    },
    {
      "epoch": 2.7547083608304987,
      "grad_norm": 0.31259456276893616,
      "learning_rate": 1.63527759446334e-06,
      "loss": 0.0306,
      "step": 54265
    },
    {
      "epoch": 2.7549621808213614,
      "grad_norm": 0.23816870152950287,
      "learning_rate": 1.6335854611909236e-06,
      "loss": 0.031,
      "step": 54270
    },
    {
      "epoch": 2.755216000812224,
      "grad_norm": 0.3377796709537506,
      "learning_rate": 1.631893327918507e-06,
      "loss": 0.0259,
      "step": 54275
    },
    {
      "epoch": 2.7554698208030866,
      "grad_norm": 0.4004886746406555,
      "learning_rate": 1.6302011946460904e-06,
      "loss": 0.0311,
      "step": 54280
    },
    {
      "epoch": 2.755723640793949,
      "grad_norm": 0.40718159079551697,
      "learning_rate": 1.628509061373674e-06,
      "loss": 0.0317,
      "step": 54285
    },
    {
      "epoch": 2.7559774607848113,
      "grad_norm": 0.32053378224372864,
      "learning_rate": 1.6268169281012575e-06,
      "loss": 0.0399,
      "step": 54290
    },
    {
      "epoch": 2.756231280775674,
      "grad_norm": 0.24005572497844696,
      "learning_rate": 1.6251247948288409e-06,
      "loss": 0.0293,
      "step": 54295
    },
    {
      "epoch": 2.7564851007665365,
      "grad_norm": 0.20497609674930573,
      "learning_rate": 1.6234326615564244e-06,
      "loss": 0.0265,
      "step": 54300
    },
    {
      "epoch": 2.7567389207573987,
      "grad_norm": 0.23367075622081757,
      "learning_rate": 1.621740528284008e-06,
      "loss": 0.0284,
      "step": 54305
    },
    {
      "epoch": 2.7569927407482613,
      "grad_norm": 0.2720516622066498,
      "learning_rate": 1.620048395011591e-06,
      "loss": 0.0291,
      "step": 54310
    },
    {
      "epoch": 2.757246560739124,
      "grad_norm": 0.5106813907623291,
      "learning_rate": 1.6183562617391746e-06,
      "loss": 0.0322,
      "step": 54315
    },
    {
      "epoch": 2.7575003807299865,
      "grad_norm": 0.36109301447868347,
      "learning_rate": 1.6166641284667582e-06,
      "loss": 0.026,
      "step": 54320
    },
    {
      "epoch": 2.7577542007208486,
      "grad_norm": 0.29374897480010986,
      "learning_rate": 1.6149719951943415e-06,
      "loss": 0.0272,
      "step": 54325
    },
    {
      "epoch": 2.7580080207117113,
      "grad_norm": 0.35226473212242126,
      "learning_rate": 1.613279861921925e-06,
      "loss": 0.0372,
      "step": 54330
    },
    {
      "epoch": 2.758261840702574,
      "grad_norm": 0.36463406682014465,
      "learning_rate": 1.6115877286495086e-06,
      "loss": 0.0308,
      "step": 54335
    },
    {
      "epoch": 2.758515660693436,
      "grad_norm": 0.5041033029556274,
      "learning_rate": 1.609895595377092e-06,
      "loss": 0.0309,
      "step": 54340
    },
    {
      "epoch": 2.7587694806842986,
      "grad_norm": 0.24311576783657074,
      "learning_rate": 1.6082034621046755e-06,
      "loss": 0.0296,
      "step": 54345
    },
    {
      "epoch": 2.759023300675161,
      "grad_norm": 0.5578087568283081,
      "learning_rate": 1.606511328832259e-06,
      "loss": 0.0323,
      "step": 54350
    },
    {
      "epoch": 2.759277120666024,
      "grad_norm": 0.2777861952781677,
      "learning_rate": 1.6048191955598425e-06,
      "loss": 0.0297,
      "step": 54355
    },
    {
      "epoch": 2.759530940656886,
      "grad_norm": 0.3467789590358734,
      "learning_rate": 1.6031270622874259e-06,
      "loss": 0.0308,
      "step": 54360
    },
    {
      "epoch": 2.7597847606477486,
      "grad_norm": 0.3258620500564575,
      "learning_rate": 1.6014349290150094e-06,
      "loss": 0.0259,
      "step": 54365
    },
    {
      "epoch": 2.760038580638611,
      "grad_norm": 0.32327863574028015,
      "learning_rate": 1.599742795742593e-06,
      "loss": 0.0298,
      "step": 54370
    },
    {
      "epoch": 2.7602924006294733,
      "grad_norm": 0.2921620309352875,
      "learning_rate": 1.5980506624701763e-06,
      "loss": 0.0314,
      "step": 54375
    },
    {
      "epoch": 2.760546220620336,
      "grad_norm": 0.3468886911869049,
      "learning_rate": 1.5963585291977598e-06,
      "loss": 0.0328,
      "step": 54380
    },
    {
      "epoch": 2.7608000406111985,
      "grad_norm": 0.4291483759880066,
      "learning_rate": 1.5946663959253434e-06,
      "loss": 0.0328,
      "step": 54385
    },
    {
      "epoch": 2.761053860602061,
      "grad_norm": 0.42713791131973267,
      "learning_rate": 1.5929742626529265e-06,
      "loss": 0.0256,
      "step": 54390
    },
    {
      "epoch": 2.7613076805929238,
      "grad_norm": 0.3330911099910736,
      "learning_rate": 1.59128212938051e-06,
      "loss": 0.0306,
      "step": 54395
    },
    {
      "epoch": 2.761561500583786,
      "grad_norm": 0.23655596375465393,
      "learning_rate": 1.5895899961080936e-06,
      "loss": 0.0303,
      "step": 54400
    },
    {
      "epoch": 2.7618153205746485,
      "grad_norm": 0.22869274020195007,
      "learning_rate": 1.587897862835677e-06,
      "loss": 0.0279,
      "step": 54405
    },
    {
      "epoch": 2.7620691405655107,
      "grad_norm": 0.37720853090286255,
      "learning_rate": 1.5862057295632605e-06,
      "loss": 0.0356,
      "step": 54410
    },
    {
      "epoch": 2.7623229605563733,
      "grad_norm": 0.4346856474876404,
      "learning_rate": 1.584513596290844e-06,
      "loss": 0.0294,
      "step": 54415
    },
    {
      "epoch": 2.762576780547236,
      "grad_norm": 0.3101567029953003,
      "learning_rate": 1.5828214630184274e-06,
      "loss": 0.0287,
      "step": 54420
    },
    {
      "epoch": 2.7628306005380985,
      "grad_norm": 0.46257123351097107,
      "learning_rate": 1.581129329746011e-06,
      "loss": 0.0243,
      "step": 54425
    },
    {
      "epoch": 2.763084420528961,
      "grad_norm": 0.42117181420326233,
      "learning_rate": 1.5794371964735944e-06,
      "loss": 0.0328,
      "step": 54430
    },
    {
      "epoch": 2.7633382405198232,
      "grad_norm": 0.2318694293498993,
      "learning_rate": 1.5777450632011778e-06,
      "loss": 0.0335,
      "step": 54435
    },
    {
      "epoch": 2.763592060510686,
      "grad_norm": 0.1912282556295395,
      "learning_rate": 1.5760529299287613e-06,
      "loss": 0.0294,
      "step": 54440
    },
    {
      "epoch": 2.7638458805015484,
      "grad_norm": 0.42659279704093933,
      "learning_rate": 1.5743607966563449e-06,
      "loss": 0.0234,
      "step": 54445
    },
    {
      "epoch": 2.7640997004924106,
      "grad_norm": 0.3059816360473633,
      "learning_rate": 1.5726686633839284e-06,
      "loss": 0.0234,
      "step": 54450
    },
    {
      "epoch": 2.764353520483273,
      "grad_norm": 0.43603745102882385,
      "learning_rate": 1.5709765301115117e-06,
      "loss": 0.0371,
      "step": 54455
    },
    {
      "epoch": 2.764607340474136,
      "grad_norm": 0.38762128353118896,
      "learning_rate": 1.5692843968390953e-06,
      "loss": 0.0337,
      "step": 54460
    },
    {
      "epoch": 2.7648611604649984,
      "grad_norm": 0.1974819451570511,
      "learning_rate": 1.5675922635666788e-06,
      "loss": 0.0254,
      "step": 54465
    },
    {
      "epoch": 2.7651149804558606,
      "grad_norm": 0.3987434506416321,
      "learning_rate": 1.565900130294262e-06,
      "loss": 0.0273,
      "step": 54470
    },
    {
      "epoch": 2.765368800446723,
      "grad_norm": 0.312901109457016,
      "learning_rate": 1.5642079970218455e-06,
      "loss": 0.0346,
      "step": 54475
    },
    {
      "epoch": 2.7656226204375858,
      "grad_norm": 0.3683776259422302,
      "learning_rate": 1.5625158637494288e-06,
      "loss": 0.0341,
      "step": 54480
    },
    {
      "epoch": 2.765876440428448,
      "grad_norm": 0.3106054365634918,
      "learning_rate": 1.5608237304770124e-06,
      "loss": 0.0288,
      "step": 54485
    },
    {
      "epoch": 2.7661302604193105,
      "grad_norm": 0.275949627161026,
      "learning_rate": 1.559131597204596e-06,
      "loss": 0.0275,
      "step": 54490
    },
    {
      "epoch": 2.766384080410173,
      "grad_norm": 0.39345717430114746,
      "learning_rate": 1.5574394639321795e-06,
      "loss": 0.0312,
      "step": 54495
    },
    {
      "epoch": 2.7666379004010357,
      "grad_norm": 0.4222765862941742,
      "learning_rate": 1.5557473306597628e-06,
      "loss": 0.0295,
      "step": 54500
    },
    {
      "epoch": 2.766891720391898,
      "grad_norm": 0.5655775666236877,
      "learning_rate": 1.5540551973873463e-06,
      "loss": 0.0336,
      "step": 54505
    },
    {
      "epoch": 2.7671455403827605,
      "grad_norm": 0.34256479144096375,
      "learning_rate": 1.5523630641149299e-06,
      "loss": 0.0305,
      "step": 54510
    },
    {
      "epoch": 2.767399360373623,
      "grad_norm": 0.31936198472976685,
      "learning_rate": 1.5506709308425132e-06,
      "loss": 0.035,
      "step": 54515
    },
    {
      "epoch": 2.7676531803644853,
      "grad_norm": 0.5186873078346252,
      "learning_rate": 1.5489787975700968e-06,
      "loss": 0.0349,
      "step": 54520
    },
    {
      "epoch": 2.767907000355348,
      "grad_norm": 0.3005698323249817,
      "learning_rate": 1.5472866642976803e-06,
      "loss": 0.0384,
      "step": 54525
    },
    {
      "epoch": 2.7681608203462105,
      "grad_norm": 0.41957610845565796,
      "learning_rate": 1.5455945310252636e-06,
      "loss": 0.033,
      "step": 54530
    },
    {
      "epoch": 2.768414640337073,
      "grad_norm": 0.21110914647579193,
      "learning_rate": 1.5439023977528472e-06,
      "loss": 0.034,
      "step": 54535
    },
    {
      "epoch": 2.7686684603279357,
      "grad_norm": 0.4496798813343048,
      "learning_rate": 1.5422102644804307e-06,
      "loss": 0.0327,
      "step": 54540
    },
    {
      "epoch": 2.768922280318798,
      "grad_norm": 0.33650708198547363,
      "learning_rate": 1.5405181312080143e-06,
      "loss": 0.0265,
      "step": 54545
    },
    {
      "epoch": 2.7691761003096604,
      "grad_norm": 0.31067219376564026,
      "learning_rate": 1.5388259979355976e-06,
      "loss": 0.031,
      "step": 54550
    },
    {
      "epoch": 2.7694299203005226,
      "grad_norm": 0.4140172600746155,
      "learning_rate": 1.537133864663181e-06,
      "loss": 0.0309,
      "step": 54555
    },
    {
      "epoch": 2.769683740291385,
      "grad_norm": 0.5254555940628052,
      "learning_rate": 1.5354417313907643e-06,
      "loss": 0.0329,
      "step": 54560
    },
    {
      "epoch": 2.769937560282248,
      "grad_norm": 0.3984996974468231,
      "learning_rate": 1.5337495981183478e-06,
      "loss": 0.0373,
      "step": 54565
    },
    {
      "epoch": 2.7701913802731104,
      "grad_norm": 0.3657810091972351,
      "learning_rate": 1.5320574648459314e-06,
      "loss": 0.0234,
      "step": 54570
    },
    {
      "epoch": 2.770445200263973,
      "grad_norm": 0.4152500033378601,
      "learning_rate": 1.5303653315735147e-06,
      "loss": 0.0374,
      "step": 54575
    },
    {
      "epoch": 2.770699020254835,
      "grad_norm": 0.3096051812171936,
      "learning_rate": 1.5286731983010982e-06,
      "loss": 0.029,
      "step": 54580
    },
    {
      "epoch": 2.7709528402456978,
      "grad_norm": 0.27113279700279236,
      "learning_rate": 1.5269810650286818e-06,
      "loss": 0.0353,
      "step": 54585
    },
    {
      "epoch": 2.7712066602365604,
      "grad_norm": 0.8830772638320923,
      "learning_rate": 1.5252889317562653e-06,
      "loss": 0.0293,
      "step": 54590
    },
    {
      "epoch": 2.7714604802274225,
      "grad_norm": 0.2550186812877655,
      "learning_rate": 1.5235967984838487e-06,
      "loss": 0.0345,
      "step": 54595
    },
    {
      "epoch": 2.771714300218285,
      "grad_norm": 0.5828475952148438,
      "learning_rate": 1.5219046652114322e-06,
      "loss": 0.0303,
      "step": 54600
    },
    {
      "epoch": 2.7719681202091477,
      "grad_norm": 0.27280479669570923,
      "learning_rate": 1.5202125319390158e-06,
      "loss": 0.0247,
      "step": 54605
    },
    {
      "epoch": 2.7722219402000103,
      "grad_norm": 0.3151116967201233,
      "learning_rate": 1.518520398666599e-06,
      "loss": 0.0344,
      "step": 54610
    },
    {
      "epoch": 2.7724757601908725,
      "grad_norm": 0.279140442609787,
      "learning_rate": 1.5168282653941826e-06,
      "loss": 0.0292,
      "step": 54615
    },
    {
      "epoch": 2.772729580181735,
      "grad_norm": 0.37659916281700134,
      "learning_rate": 1.5151361321217662e-06,
      "loss": 0.0321,
      "step": 54620
    },
    {
      "epoch": 2.7729834001725977,
      "grad_norm": 0.3131263256072998,
      "learning_rate": 1.5134439988493497e-06,
      "loss": 0.026,
      "step": 54625
    },
    {
      "epoch": 2.77323722016346,
      "grad_norm": 0.20809635519981384,
      "learning_rate": 1.511751865576933e-06,
      "loss": 0.0307,
      "step": 54630
    },
    {
      "epoch": 2.7734910401543225,
      "grad_norm": 0.30809369683265686,
      "learning_rate": 1.5100597323045164e-06,
      "loss": 0.0343,
      "step": 54635
    },
    {
      "epoch": 2.773744860145185,
      "grad_norm": 0.2556045949459076,
      "learning_rate": 1.5083675990320997e-06,
      "loss": 0.0235,
      "step": 54640
    },
    {
      "epoch": 2.7739986801360477,
      "grad_norm": 0.1649613231420517,
      "learning_rate": 1.5066754657596833e-06,
      "loss": 0.0305,
      "step": 54645
    },
    {
      "epoch": 2.7742525001269103,
      "grad_norm": 0.23043885827064514,
      "learning_rate": 1.5049833324872668e-06,
      "loss": 0.0304,
      "step": 54650
    },
    {
      "epoch": 2.7745063201177724,
      "grad_norm": 0.5607897043228149,
      "learning_rate": 1.5032911992148501e-06,
      "loss": 0.0281,
      "step": 54655
    },
    {
      "epoch": 2.774760140108635,
      "grad_norm": 0.28994089365005493,
      "learning_rate": 1.5015990659424337e-06,
      "loss": 0.029,
      "step": 54660
    },
    {
      "epoch": 2.775013960099497,
      "grad_norm": 0.3250032961368561,
      "learning_rate": 1.4999069326700172e-06,
      "loss": 0.0268,
      "step": 54665
    },
    {
      "epoch": 2.77526778009036,
      "grad_norm": 0.5446613430976868,
      "learning_rate": 1.4982147993976008e-06,
      "loss": 0.0253,
      "step": 54670
    },
    {
      "epoch": 2.7755216000812224,
      "grad_norm": 0.33558404445648193,
      "learning_rate": 1.496522666125184e-06,
      "loss": 0.0323,
      "step": 54675
    },
    {
      "epoch": 2.775775420072085,
      "grad_norm": 0.37360864877700806,
      "learning_rate": 1.4948305328527676e-06,
      "loss": 0.0341,
      "step": 54680
    },
    {
      "epoch": 2.7760292400629476,
      "grad_norm": 0.39937227964401245,
      "learning_rate": 1.4931383995803512e-06,
      "loss": 0.0308,
      "step": 54685
    },
    {
      "epoch": 2.7762830600538098,
      "grad_norm": 0.4435569941997528,
      "learning_rate": 1.4914462663079345e-06,
      "loss": 0.0347,
      "step": 54690
    },
    {
      "epoch": 2.7765368800446724,
      "grad_norm": 0.2976076602935791,
      "learning_rate": 1.489754133035518e-06,
      "loss": 0.0308,
      "step": 54695
    },
    {
      "epoch": 2.776790700035535,
      "grad_norm": 0.5093637704849243,
      "learning_rate": 1.4880619997631016e-06,
      "loss": 0.0256,
      "step": 54700
    },
    {
      "epoch": 2.777044520026397,
      "grad_norm": 0.2233898788690567,
      "learning_rate": 1.486369866490685e-06,
      "loss": 0.0328,
      "step": 54705
    },
    {
      "epoch": 2.7772983400172597,
      "grad_norm": 0.3220416009426117,
      "learning_rate": 1.4846777332182685e-06,
      "loss": 0.0316,
      "step": 54710
    },
    {
      "epoch": 2.7775521600081223,
      "grad_norm": 0.3018849194049835,
      "learning_rate": 1.4829855999458518e-06,
      "loss": 0.0328,
      "step": 54715
    },
    {
      "epoch": 2.777805979998985,
      "grad_norm": 0.29990217089653015,
      "learning_rate": 1.4812934666734352e-06,
      "loss": 0.025,
      "step": 54720
    },
    {
      "epoch": 2.778059799989847,
      "grad_norm": 0.2418762594461441,
      "learning_rate": 1.4796013334010187e-06,
      "loss": 0.0323,
      "step": 54725
    },
    {
      "epoch": 2.7783136199807097,
      "grad_norm": 0.46849316358566284,
      "learning_rate": 1.4779092001286022e-06,
      "loss": 0.0352,
      "step": 54730
    },
    {
      "epoch": 2.7785674399715723,
      "grad_norm": 0.4002370536327362,
      "learning_rate": 1.4762170668561856e-06,
      "loss": 0.0331,
      "step": 54735
    },
    {
      "epoch": 2.7788212599624345,
      "grad_norm": 0.37340641021728516,
      "learning_rate": 1.4745249335837691e-06,
      "loss": 0.028,
      "step": 54740
    },
    {
      "epoch": 2.779075079953297,
      "grad_norm": 0.20235440135002136,
      "learning_rate": 1.4728328003113527e-06,
      "loss": 0.0298,
      "step": 54745
    },
    {
      "epoch": 2.7793288999441597,
      "grad_norm": 0.379058301448822,
      "learning_rate": 1.471140667038936e-06,
      "loss": 0.0319,
      "step": 54750
    },
    {
      "epoch": 2.7795827199350223,
      "grad_norm": 0.7213035225868225,
      "learning_rate": 1.4694485337665195e-06,
      "loss": 0.0343,
      "step": 54755
    },
    {
      "epoch": 2.7798365399258844,
      "grad_norm": 0.4405682384967804,
      "learning_rate": 1.467756400494103e-06,
      "loss": 0.0266,
      "step": 54760
    },
    {
      "epoch": 2.780090359916747,
      "grad_norm": 0.43502214550971985,
      "learning_rate": 1.4660642672216866e-06,
      "loss": 0.026,
      "step": 54765
    },
    {
      "epoch": 2.7803441799076096,
      "grad_norm": 0.3652034103870392,
      "learning_rate": 1.46437213394927e-06,
      "loss": 0.0295,
      "step": 54770
    },
    {
      "epoch": 2.780597999898472,
      "grad_norm": 0.2877683937549591,
      "learning_rate": 1.4626800006768535e-06,
      "loss": 0.0298,
      "step": 54775
    },
    {
      "epoch": 2.7808518198893344,
      "grad_norm": 0.24761992692947388,
      "learning_rate": 1.460987867404437e-06,
      "loss": 0.0292,
      "step": 54780
    },
    {
      "epoch": 2.781105639880197,
      "grad_norm": 0.2824486792087555,
      "learning_rate": 1.4592957341320204e-06,
      "loss": 0.0317,
      "step": 54785
    },
    {
      "epoch": 2.7813594598710596,
      "grad_norm": 0.33154594898223877,
      "learning_rate": 1.457603600859604e-06,
      "loss": 0.036,
      "step": 54790
    },
    {
      "epoch": 2.781613279861922,
      "grad_norm": 0.3138459324836731,
      "learning_rate": 1.455911467587187e-06,
      "loss": 0.0263,
      "step": 54795
    },
    {
      "epoch": 2.7818670998527844,
      "grad_norm": 0.39758846163749695,
      "learning_rate": 1.4542193343147706e-06,
      "loss": 0.0324,
      "step": 54800
    },
    {
      "epoch": 2.782120919843647,
      "grad_norm": 0.2929428219795227,
      "learning_rate": 1.4525272010423541e-06,
      "loss": 0.0262,
      "step": 54805
    },
    {
      "epoch": 2.782374739834509,
      "grad_norm": 0.3221243619918823,
      "learning_rate": 1.4508350677699377e-06,
      "loss": 0.0363,
      "step": 54810
    },
    {
      "epoch": 2.7826285598253717,
      "grad_norm": 0.4146369397640228,
      "learning_rate": 1.449142934497521e-06,
      "loss": 0.0335,
      "step": 54815
    },
    {
      "epoch": 2.7828823798162343,
      "grad_norm": 0.48562976717948914,
      "learning_rate": 1.4474508012251046e-06,
      "loss": 0.0375,
      "step": 54820
    },
    {
      "epoch": 2.783136199807097,
      "grad_norm": 0.7758735418319702,
      "learning_rate": 1.4457586679526881e-06,
      "loss": 0.0289,
      "step": 54825
    },
    {
      "epoch": 2.7833900197979595,
      "grad_norm": 0.33634909987449646,
      "learning_rate": 1.4440665346802714e-06,
      "loss": 0.0261,
      "step": 54830
    },
    {
      "epoch": 2.7836438397888217,
      "grad_norm": 0.3155843913555145,
      "learning_rate": 1.442374401407855e-06,
      "loss": 0.0344,
      "step": 54835
    },
    {
      "epoch": 2.7838976597796843,
      "grad_norm": 0.2868061363697052,
      "learning_rate": 1.4406822681354385e-06,
      "loss": 0.0326,
      "step": 54840
    },
    {
      "epoch": 2.784151479770547,
      "grad_norm": 0.35042792558670044,
      "learning_rate": 1.4389901348630219e-06,
      "loss": 0.0284,
      "step": 54845
    },
    {
      "epoch": 2.784405299761409,
      "grad_norm": 0.2201160043478012,
      "learning_rate": 1.4372980015906054e-06,
      "loss": 0.0221,
      "step": 54850
    },
    {
      "epoch": 2.7846591197522716,
      "grad_norm": 0.3119716942310333,
      "learning_rate": 1.435605868318189e-06,
      "loss": 0.0265,
      "step": 54855
    },
    {
      "epoch": 2.7849129397431343,
      "grad_norm": 0.4090033173561096,
      "learning_rate": 1.4339137350457725e-06,
      "loss": 0.0345,
      "step": 54860
    },
    {
      "epoch": 2.785166759733997,
      "grad_norm": 0.2550966143608093,
      "learning_rate": 1.4322216017733558e-06,
      "loss": 0.0311,
      "step": 54865
    },
    {
      "epoch": 2.785420579724859,
      "grad_norm": 0.27337196469306946,
      "learning_rate": 1.4305294685009394e-06,
      "loss": 0.0361,
      "step": 54870
    },
    {
      "epoch": 2.7856743997157216,
      "grad_norm": 0.26968151330947876,
      "learning_rate": 1.4288373352285225e-06,
      "loss": 0.0265,
      "step": 54875
    },
    {
      "epoch": 2.785928219706584,
      "grad_norm": 0.3220800757408142,
      "learning_rate": 1.427145201956106e-06,
      "loss": 0.0278,
      "step": 54880
    },
    {
      "epoch": 2.7861820396974464,
      "grad_norm": 0.2650590240955353,
      "learning_rate": 1.4254530686836896e-06,
      "loss": 0.0317,
      "step": 54885
    },
    {
      "epoch": 2.786435859688309,
      "grad_norm": 0.2614637017250061,
      "learning_rate": 1.4237609354112731e-06,
      "loss": 0.0273,
      "step": 54890
    },
    {
      "epoch": 2.7866896796791716,
      "grad_norm": 0.34072941541671753,
      "learning_rate": 1.4220688021388565e-06,
      "loss": 0.0249,
      "step": 54895
    },
    {
      "epoch": 2.786943499670034,
      "grad_norm": 0.2870447337627411,
      "learning_rate": 1.42037666886644e-06,
      "loss": 0.0274,
      "step": 54900
    },
    {
      "epoch": 2.7871973196608963,
      "grad_norm": 0.28836342692375183,
      "learning_rate": 1.4186845355940236e-06,
      "loss": 0.0328,
      "step": 54905
    },
    {
      "epoch": 2.787451139651759,
      "grad_norm": 0.4721055030822754,
      "learning_rate": 1.4169924023216069e-06,
      "loss": 0.0343,
      "step": 54910
    },
    {
      "epoch": 2.7877049596426215,
      "grad_norm": 0.23974347114562988,
      "learning_rate": 1.4153002690491904e-06,
      "loss": 0.027,
      "step": 54915
    },
    {
      "epoch": 2.7879587796334837,
      "grad_norm": 0.31404948234558105,
      "learning_rate": 1.413608135776774e-06,
      "loss": 0.0383,
      "step": 54920
    },
    {
      "epoch": 2.7882125996243463,
      "grad_norm": 0.26835718750953674,
      "learning_rate": 1.4119160025043573e-06,
      "loss": 0.0234,
      "step": 54925
    },
    {
      "epoch": 2.788466419615209,
      "grad_norm": 0.4135149419307709,
      "learning_rate": 1.4102238692319409e-06,
      "loss": 0.0293,
      "step": 54930
    },
    {
      "epoch": 2.7887202396060715,
      "grad_norm": 0.3677108883857727,
      "learning_rate": 1.4085317359595244e-06,
      "loss": 0.026,
      "step": 54935
    },
    {
      "epoch": 2.788974059596934,
      "grad_norm": 0.28418686985969543,
      "learning_rate": 1.406839602687108e-06,
      "loss": 0.0255,
      "step": 54940
    },
    {
      "epoch": 2.7892278795877963,
      "grad_norm": 0.369150847196579,
      "learning_rate": 1.4051474694146913e-06,
      "loss": 0.0335,
      "step": 54945
    },
    {
      "epoch": 2.789481699578659,
      "grad_norm": 0.23390887677669525,
      "learning_rate": 1.4034553361422748e-06,
      "loss": 0.0245,
      "step": 54950
    },
    {
      "epoch": 2.789735519569521,
      "grad_norm": 0.5765233039855957,
      "learning_rate": 1.401763202869858e-06,
      "loss": 0.026,
      "step": 54955
    },
    {
      "epoch": 2.7899893395603836,
      "grad_norm": 0.35066676139831543,
      "learning_rate": 1.4000710695974415e-06,
      "loss": 0.0316,
      "step": 54960
    },
    {
      "epoch": 2.7902431595512462,
      "grad_norm": 0.282090425491333,
      "learning_rate": 1.398378936325025e-06,
      "loss": 0.029,
      "step": 54965
    },
    {
      "epoch": 2.790496979542109,
      "grad_norm": 0.24391506612300873,
      "learning_rate": 1.3966868030526084e-06,
      "loss": 0.0289,
      "step": 54970
    },
    {
      "epoch": 2.7907507995329714,
      "grad_norm": 0.3046450912952423,
      "learning_rate": 1.394994669780192e-06,
      "loss": 0.0325,
      "step": 54975
    },
    {
      "epoch": 2.7910046195238336,
      "grad_norm": 0.2391621321439743,
      "learning_rate": 1.3933025365077755e-06,
      "loss": 0.0363,
      "step": 54980
    },
    {
      "epoch": 2.791258439514696,
      "grad_norm": 0.3334785997867584,
      "learning_rate": 1.391610403235359e-06,
      "loss": 0.0321,
      "step": 54985
    },
    {
      "epoch": 2.791512259505559,
      "grad_norm": 0.2553236782550812,
      "learning_rate": 1.3899182699629423e-06,
      "loss": 0.0232,
      "step": 54990
    },
    {
      "epoch": 2.791766079496421,
      "grad_norm": 0.2845880687236786,
      "learning_rate": 1.3882261366905259e-06,
      "loss": 0.0403,
      "step": 54995
    },
    {
      "epoch": 2.7920198994872836,
      "grad_norm": 0.23673295974731445,
      "learning_rate": 1.3865340034181094e-06,
      "loss": 0.0299,
      "step": 55000
    },
    {
      "epoch": 2.792273719478146,
      "grad_norm": 0.37878087162971497,
      "learning_rate": 1.3848418701456927e-06,
      "loss": 0.0454,
      "step": 55005
    },
    {
      "epoch": 2.792527539469009,
      "grad_norm": 0.3362523019313812,
      "learning_rate": 1.3831497368732763e-06,
      "loss": 0.0273,
      "step": 55010
    },
    {
      "epoch": 2.792781359459871,
      "grad_norm": 0.3414081335067749,
      "learning_rate": 1.3814576036008598e-06,
      "loss": 0.0353,
      "step": 55015
    },
    {
      "epoch": 2.7930351794507335,
      "grad_norm": 0.5782923102378845,
      "learning_rate": 1.3797654703284432e-06,
      "loss": 0.0391,
      "step": 55020
    },
    {
      "epoch": 2.793288999441596,
      "grad_norm": 0.33037516474723816,
      "learning_rate": 1.3780733370560267e-06,
      "loss": 0.0296,
      "step": 55025
    },
    {
      "epoch": 2.7935428194324583,
      "grad_norm": 0.3172570466995239,
      "learning_rate": 1.3763812037836103e-06,
      "loss": 0.0353,
      "step": 55030
    },
    {
      "epoch": 2.793796639423321,
      "grad_norm": 0.3240835964679718,
      "learning_rate": 1.3746890705111934e-06,
      "loss": 0.0338,
      "step": 55035
    },
    {
      "epoch": 2.7940504594141835,
      "grad_norm": 0.2889838218688965,
      "learning_rate": 1.372996937238777e-06,
      "loss": 0.0278,
      "step": 55040
    },
    {
      "epoch": 2.794304279405046,
      "grad_norm": 0.5404177308082581,
      "learning_rate": 1.3713048039663605e-06,
      "loss": 0.0267,
      "step": 55045
    },
    {
      "epoch": 2.7945580993959083,
      "grad_norm": 0.27731940150260925,
      "learning_rate": 1.3696126706939438e-06,
      "loss": 0.0269,
      "step": 55050
    },
    {
      "epoch": 2.794811919386771,
      "grad_norm": 0.46202221512794495,
      "learning_rate": 1.3679205374215273e-06,
      "loss": 0.0344,
      "step": 55055
    },
    {
      "epoch": 2.7950657393776335,
      "grad_norm": 0.4258192181587219,
      "learning_rate": 1.3662284041491109e-06,
      "loss": 0.0289,
      "step": 55060
    },
    {
      "epoch": 2.7953195593684956,
      "grad_norm": 0.574909508228302,
      "learning_rate": 1.3645362708766942e-06,
      "loss": 0.0306,
      "step": 55065
    },
    {
      "epoch": 2.7955733793593582,
      "grad_norm": 0.5580008625984192,
      "learning_rate": 1.3628441376042778e-06,
      "loss": 0.0283,
      "step": 55070
    },
    {
      "epoch": 2.795827199350221,
      "grad_norm": 0.24519678950309753,
      "learning_rate": 1.3611520043318613e-06,
      "loss": 0.0294,
      "step": 55075
    },
    {
      "epoch": 2.7960810193410834,
      "grad_norm": 0.351688414812088,
      "learning_rate": 1.3594598710594449e-06,
      "loss": 0.0326,
      "step": 55080
    },
    {
      "epoch": 2.796334839331946,
      "grad_norm": 0.544750452041626,
      "learning_rate": 1.3577677377870282e-06,
      "loss": 0.0308,
      "step": 55085
    },
    {
      "epoch": 2.796588659322808,
      "grad_norm": 0.2852821350097656,
      "learning_rate": 1.3560756045146117e-06,
      "loss": 0.0298,
      "step": 55090
    },
    {
      "epoch": 2.796842479313671,
      "grad_norm": 0.2739110589027405,
      "learning_rate": 1.3543834712421953e-06,
      "loss": 0.0313,
      "step": 55095
    },
    {
      "epoch": 2.797096299304533,
      "grad_norm": 0.26846620440483093,
      "learning_rate": 1.3526913379697786e-06,
      "loss": 0.0255,
      "step": 55100
    },
    {
      "epoch": 2.7973501192953956,
      "grad_norm": 0.3372206389904022,
      "learning_rate": 1.3509992046973622e-06,
      "loss": 0.0261,
      "step": 55105
    },
    {
      "epoch": 2.797603939286258,
      "grad_norm": 0.2873419523239136,
      "learning_rate": 1.3493070714249457e-06,
      "loss": 0.0322,
      "step": 55110
    },
    {
      "epoch": 2.7978577592771208,
      "grad_norm": 0.3785225749015808,
      "learning_rate": 1.347614938152529e-06,
      "loss": 0.0282,
      "step": 55115
    },
    {
      "epoch": 2.7981115792679834,
      "grad_norm": 0.3132956624031067,
      "learning_rate": 1.3459228048801124e-06,
      "loss": 0.0271,
      "step": 55120
    },
    {
      "epoch": 2.7983653992588455,
      "grad_norm": 0.22603264451026917,
      "learning_rate": 1.344230671607696e-06,
      "loss": 0.0328,
      "step": 55125
    },
    {
      "epoch": 2.798619219249708,
      "grad_norm": 0.2586395740509033,
      "learning_rate": 1.3425385383352792e-06,
      "loss": 0.0207,
      "step": 55130
    },
    {
      "epoch": 2.7988730392405707,
      "grad_norm": 0.8957403898239136,
      "learning_rate": 1.3408464050628628e-06,
      "loss": 0.0314,
      "step": 55135
    },
    {
      "epoch": 2.799126859231433,
      "grad_norm": 0.24698592722415924,
      "learning_rate": 1.3391542717904463e-06,
      "loss": 0.0292,
      "step": 55140
    },
    {
      "epoch": 2.7993806792222955,
      "grad_norm": 0.2640494704246521,
      "learning_rate": 1.3374621385180297e-06,
      "loss": 0.0289,
      "step": 55145
    },
    {
      "epoch": 2.799634499213158,
      "grad_norm": 0.2573885917663574,
      "learning_rate": 1.3357700052456132e-06,
      "loss": 0.0295,
      "step": 55150
    },
    {
      "epoch": 2.7998883192040207,
      "grad_norm": 0.2399040311574936,
      "learning_rate": 1.3340778719731968e-06,
      "loss": 0.028,
      "step": 55155
    },
    {
      "epoch": 2.800142139194883,
      "grad_norm": 0.41929957270622253,
      "learning_rate": 1.3323857387007803e-06,
      "loss": 0.0251,
      "step": 55160
    },
    {
      "epoch": 2.8003959591857455,
      "grad_norm": 0.40195733308792114,
      "learning_rate": 1.3306936054283636e-06,
      "loss": 0.022,
      "step": 55165
    },
    {
      "epoch": 2.800649779176608,
      "grad_norm": 0.31589677929878235,
      "learning_rate": 1.3290014721559472e-06,
      "loss": 0.0308,
      "step": 55170
    },
    {
      "epoch": 2.8009035991674702,
      "grad_norm": 0.3632115423679352,
      "learning_rate": 1.3273093388835307e-06,
      "loss": 0.0315,
      "step": 55175
    },
    {
      "epoch": 2.801157419158333,
      "grad_norm": 0.37809327244758606,
      "learning_rate": 1.325617205611114e-06,
      "loss": 0.0344,
      "step": 55180
    },
    {
      "epoch": 2.8014112391491954,
      "grad_norm": 0.3532311022281647,
      "learning_rate": 1.3239250723386976e-06,
      "loss": 0.0318,
      "step": 55185
    },
    {
      "epoch": 2.801665059140058,
      "grad_norm": 0.3482745885848999,
      "learning_rate": 1.3222329390662811e-06,
      "loss": 0.0354,
      "step": 55190
    },
    {
      "epoch": 2.80191887913092,
      "grad_norm": 0.3260227143764496,
      "learning_rate": 1.3205408057938645e-06,
      "loss": 0.0296,
      "step": 55195
    },
    {
      "epoch": 2.802172699121783,
      "grad_norm": 0.2941901385784149,
      "learning_rate": 1.3188486725214478e-06,
      "loss": 0.028,
      "step": 55200
    },
    {
      "epoch": 2.8024265191126454,
      "grad_norm": 0.3940282464027405,
      "learning_rate": 1.3171565392490314e-06,
      "loss": 0.0374,
      "step": 55205
    },
    {
      "epoch": 2.8026803391035076,
      "grad_norm": 0.2064363658428192,
      "learning_rate": 1.3154644059766147e-06,
      "loss": 0.027,
      "step": 55210
    },
    {
      "epoch": 2.80293415909437,
      "grad_norm": 0.303488165140152,
      "learning_rate": 1.3137722727041982e-06,
      "loss": 0.0314,
      "step": 55215
    },
    {
      "epoch": 2.8031879790852328,
      "grad_norm": 0.49318596720695496,
      "learning_rate": 1.3120801394317818e-06,
      "loss": 0.0298,
      "step": 55220
    },
    {
      "epoch": 2.8034417990760954,
      "grad_norm": 0.2664399743080139,
      "learning_rate": 1.3103880061593651e-06,
      "loss": 0.0303,
      "step": 55225
    },
    {
      "epoch": 2.803695619066958,
      "grad_norm": 0.2834978699684143,
      "learning_rate": 1.3086958728869487e-06,
      "loss": 0.0252,
      "step": 55230
    },
    {
      "epoch": 2.80394943905782,
      "grad_norm": 0.3490089178085327,
      "learning_rate": 1.3070037396145322e-06,
      "loss": 0.0221,
      "step": 55235
    },
    {
      "epoch": 2.8042032590486827,
      "grad_norm": 0.4803531765937805,
      "learning_rate": 1.3053116063421155e-06,
      "loss": 0.0222,
      "step": 55240
    },
    {
      "epoch": 2.804457079039545,
      "grad_norm": 0.4122765064239502,
      "learning_rate": 1.303619473069699e-06,
      "loss": 0.0318,
      "step": 55245
    },
    {
      "epoch": 2.8047108990304075,
      "grad_norm": 0.4039084315299988,
      "learning_rate": 1.3019273397972826e-06,
      "loss": 0.0321,
      "step": 55250
    },
    {
      "epoch": 2.80496471902127,
      "grad_norm": 0.32391849160194397,
      "learning_rate": 1.3002352065248662e-06,
      "loss": 0.0344,
      "step": 55255
    },
    {
      "epoch": 2.8052185390121327,
      "grad_norm": 0.3236011564731598,
      "learning_rate": 1.2985430732524495e-06,
      "loss": 0.0245,
      "step": 55260
    },
    {
      "epoch": 2.8054723590029953,
      "grad_norm": 0.41068461537361145,
      "learning_rate": 1.296850939980033e-06,
      "loss": 0.0278,
      "step": 55265
    },
    {
      "epoch": 2.8057261789938575,
      "grad_norm": 0.2617715895175934,
      "learning_rate": 1.2951588067076166e-06,
      "loss": 0.0337,
      "step": 55270
    },
    {
      "epoch": 2.80597999898472,
      "grad_norm": 0.32993102073669434,
      "learning_rate": 1.2934666734352e-06,
      "loss": 0.0315,
      "step": 55275
    },
    {
      "epoch": 2.8062338189755827,
      "grad_norm": 0.29497578740119934,
      "learning_rate": 1.2917745401627833e-06,
      "loss": 0.0282,
      "step": 55280
    },
    {
      "epoch": 2.806487638966445,
      "grad_norm": 0.3863518238067627,
      "learning_rate": 1.2900824068903666e-06,
      "loss": 0.0311,
      "step": 55285
    },
    {
      "epoch": 2.8067414589573074,
      "grad_norm": 0.29784294962882996,
      "learning_rate": 1.2883902736179501e-06,
      "loss": 0.0233,
      "step": 55290
    },
    {
      "epoch": 2.80699527894817,
      "grad_norm": 0.2937968671321869,
      "learning_rate": 1.2866981403455337e-06,
      "loss": 0.0351,
      "step": 55295
    },
    {
      "epoch": 2.8072490989390326,
      "grad_norm": 0.3985872268676758,
      "learning_rate": 1.2850060070731172e-06,
      "loss": 0.0356,
      "step": 55300
    },
    {
      "epoch": 2.807502918929895,
      "grad_norm": 0.3850856423377991,
      "learning_rate": 1.2833138738007006e-06,
      "loss": 0.0315,
      "step": 55305
    },
    {
      "epoch": 2.8077567389207574,
      "grad_norm": 0.3013561964035034,
      "learning_rate": 1.281621740528284e-06,
      "loss": 0.0286,
      "step": 55310
    },
    {
      "epoch": 2.80801055891162,
      "grad_norm": 0.40981554985046387,
      "learning_rate": 1.2799296072558676e-06,
      "loss": 0.0333,
      "step": 55315
    },
    {
      "epoch": 2.808264378902482,
      "grad_norm": 0.2424926608800888,
      "learning_rate": 1.278237473983451e-06,
      "loss": 0.0323,
      "step": 55320
    },
    {
      "epoch": 2.8085181988933448,
      "grad_norm": 0.4015446901321411,
      "learning_rate": 1.2765453407110345e-06,
      "loss": 0.0266,
      "step": 55325
    },
    {
      "epoch": 2.8087720188842074,
      "grad_norm": 0.7582650780677795,
      "learning_rate": 1.274853207438618e-06,
      "loss": 0.0392,
      "step": 55330
    },
    {
      "epoch": 2.80902583887507,
      "grad_norm": 1.0067224502563477,
      "learning_rate": 1.2731610741662014e-06,
      "loss": 0.0278,
      "step": 55335
    },
    {
      "epoch": 2.809279658865932,
      "grad_norm": 0.27800679206848145,
      "learning_rate": 1.271468940893785e-06,
      "loss": 0.0315,
      "step": 55340
    },
    {
      "epoch": 2.8095334788567947,
      "grad_norm": 0.5860305428504944,
      "learning_rate": 1.2697768076213685e-06,
      "loss": 0.0311,
      "step": 55345
    },
    {
      "epoch": 2.8097872988476573,
      "grad_norm": 0.34320294857025146,
      "learning_rate": 1.268084674348952e-06,
      "loss": 0.0297,
      "step": 55350
    },
    {
      "epoch": 2.8100411188385195,
      "grad_norm": 0.3073110580444336,
      "learning_rate": 1.2663925410765354e-06,
      "loss": 0.0262,
      "step": 55355
    },
    {
      "epoch": 2.810294938829382,
      "grad_norm": 0.6585887670516968,
      "learning_rate": 1.2647004078041187e-06,
      "loss": 0.0306,
      "step": 55360
    },
    {
      "epoch": 2.8105487588202447,
      "grad_norm": 0.9610297083854675,
      "learning_rate": 1.263008274531702e-06,
      "loss": 0.0327,
      "step": 55365
    },
    {
      "epoch": 2.8108025788111073,
      "grad_norm": 0.2548528015613556,
      "learning_rate": 1.2613161412592856e-06,
      "loss": 0.0327,
      "step": 55370
    },
    {
      "epoch": 2.81105639880197,
      "grad_norm": 0.4964401423931122,
      "learning_rate": 1.2596240079868691e-06,
      "loss": 0.0359,
      "step": 55375
    },
    {
      "epoch": 2.811310218792832,
      "grad_norm": 0.27559715509414673,
      "learning_rate": 1.2579318747144524e-06,
      "loss": 0.0302,
      "step": 55380
    },
    {
      "epoch": 2.8115640387836947,
      "grad_norm": 0.6710149645805359,
      "learning_rate": 1.256239741442036e-06,
      "loss": 0.0304,
      "step": 55385
    },
    {
      "epoch": 2.811817858774557,
      "grad_norm": 1.0265967845916748,
      "learning_rate": 1.2545476081696195e-06,
      "loss": 0.029,
      "step": 55390
    },
    {
      "epoch": 2.8120716787654194,
      "grad_norm": 0.46441468596458435,
      "learning_rate": 1.252855474897203e-06,
      "loss": 0.0356,
      "step": 55395
    },
    {
      "epoch": 2.812325498756282,
      "grad_norm": 0.21976347267627716,
      "learning_rate": 1.2511633416247864e-06,
      "loss": 0.0228,
      "step": 55400
    },
    {
      "epoch": 2.8125793187471446,
      "grad_norm": 0.7399079203605652,
      "learning_rate": 1.24947120835237e-06,
      "loss": 0.0308,
      "step": 55405
    },
    {
      "epoch": 2.812833138738007,
      "grad_norm": 0.38101959228515625,
      "learning_rate": 1.2477790750799535e-06,
      "loss": 0.0259,
      "step": 55410
    },
    {
      "epoch": 2.8130869587288694,
      "grad_norm": 0.30995088815689087,
      "learning_rate": 1.2460869418075368e-06,
      "loss": 0.0305,
      "step": 55415
    },
    {
      "epoch": 2.813340778719732,
      "grad_norm": 0.4461381137371063,
      "learning_rate": 1.2443948085351202e-06,
      "loss": 0.0239,
      "step": 55420
    },
    {
      "epoch": 2.8135945987105946,
      "grad_norm": 0.33550941944122314,
      "learning_rate": 1.2427026752627037e-06,
      "loss": 0.0292,
      "step": 55425
    },
    {
      "epoch": 2.8138484187014567,
      "grad_norm": 0.25769856572151184,
      "learning_rate": 1.2410105419902873e-06,
      "loss": 0.0264,
      "step": 55430
    },
    {
      "epoch": 2.8141022386923193,
      "grad_norm": 0.39704206585884094,
      "learning_rate": 1.2393184087178708e-06,
      "loss": 0.0312,
      "step": 55435
    },
    {
      "epoch": 2.814356058683182,
      "grad_norm": 0.4254438877105713,
      "learning_rate": 1.2376262754454541e-06,
      "loss": 0.0329,
      "step": 55440
    },
    {
      "epoch": 2.8146098786740446,
      "grad_norm": 0.2788540720939636,
      "learning_rate": 1.2359341421730377e-06,
      "loss": 0.0289,
      "step": 55445
    },
    {
      "epoch": 2.8148636986649067,
      "grad_norm": 0.37466317415237427,
      "learning_rate": 1.2342420089006212e-06,
      "loss": 0.0301,
      "step": 55450
    },
    {
      "epoch": 2.8151175186557693,
      "grad_norm": 0.2767610251903534,
      "learning_rate": 1.2325498756282046e-06,
      "loss": 0.0343,
      "step": 55455
    },
    {
      "epoch": 2.815371338646632,
      "grad_norm": 0.30491724610328674,
      "learning_rate": 1.2308577423557879e-06,
      "loss": 0.0295,
      "step": 55460
    },
    {
      "epoch": 2.815625158637494,
      "grad_norm": 0.3644537031650543,
      "learning_rate": 1.2291656090833714e-06,
      "loss": 0.027,
      "step": 55465
    },
    {
      "epoch": 2.8158789786283567,
      "grad_norm": 0.5963749885559082,
      "learning_rate": 1.227473475810955e-06,
      "loss": 0.0304,
      "step": 55470
    },
    {
      "epoch": 2.8161327986192193,
      "grad_norm": 0.31018519401550293,
      "learning_rate": 1.2257813425385385e-06,
      "loss": 0.0256,
      "step": 55475
    },
    {
      "epoch": 2.816386618610082,
      "grad_norm": 0.311709463596344,
      "learning_rate": 1.2240892092661219e-06,
      "loss": 0.0264,
      "step": 55480
    },
    {
      "epoch": 2.8166404386009445,
      "grad_norm": 0.4451717734336853,
      "learning_rate": 1.2223970759937054e-06,
      "loss": 0.0304,
      "step": 55485
    },
    {
      "epoch": 2.8168942585918066,
      "grad_norm": 0.3190285265445709,
      "learning_rate": 1.220704942721289e-06,
      "loss": 0.0262,
      "step": 55490
    },
    {
      "epoch": 2.8171480785826692,
      "grad_norm": 0.29641619324684143,
      "learning_rate": 1.2190128094488723e-06,
      "loss": 0.0332,
      "step": 55495
    },
    {
      "epoch": 2.8174018985735314,
      "grad_norm": 0.3630962371826172,
      "learning_rate": 1.2173206761764556e-06,
      "loss": 0.0196,
      "step": 55500
    },
    {
      "epoch": 2.817655718564394,
      "grad_norm": 0.26542800664901733,
      "learning_rate": 1.2156285429040392e-06,
      "loss": 0.0379,
      "step": 55505
    },
    {
      "epoch": 2.8179095385552566,
      "grad_norm": 0.45730018615722656,
      "learning_rate": 1.2139364096316227e-06,
      "loss": 0.0365,
      "step": 55510
    },
    {
      "epoch": 2.818163358546119,
      "grad_norm": 0.3015592694282532,
      "learning_rate": 1.212244276359206e-06,
      "loss": 0.0382,
      "step": 55515
    },
    {
      "epoch": 2.818417178536982,
      "grad_norm": 0.2231670320034027,
      "learning_rate": 1.2105521430867896e-06,
      "loss": 0.0276,
      "step": 55520
    },
    {
      "epoch": 2.818670998527844,
      "grad_norm": 0.3285776674747467,
      "learning_rate": 1.2088600098143731e-06,
      "loss": 0.0272,
      "step": 55525
    },
    {
      "epoch": 2.8189248185187066,
      "grad_norm": 0.23628677427768707,
      "learning_rate": 1.2071678765419567e-06,
      "loss": 0.0328,
      "step": 55530
    },
    {
      "epoch": 2.819178638509569,
      "grad_norm": 0.5668171644210815,
      "learning_rate": 1.20547574326954e-06,
      "loss": 0.0319,
      "step": 55535
    },
    {
      "epoch": 2.8194324585004313,
      "grad_norm": 0.3623851537704468,
      "learning_rate": 1.2037836099971233e-06,
      "loss": 0.0282,
      "step": 55540
    },
    {
      "epoch": 2.819686278491294,
      "grad_norm": 0.2985495626926422,
      "learning_rate": 1.2020914767247069e-06,
      "loss": 0.0288,
      "step": 55545
    },
    {
      "epoch": 2.8199400984821565,
      "grad_norm": 0.32171109318733215,
      "learning_rate": 1.2003993434522904e-06,
      "loss": 0.03,
      "step": 55550
    },
    {
      "epoch": 2.820193918473019,
      "grad_norm": 0.3371548056602478,
      "learning_rate": 1.1987072101798738e-06,
      "loss": 0.0345,
      "step": 55555
    },
    {
      "epoch": 2.8204477384638813,
      "grad_norm": 0.3954242467880249,
      "learning_rate": 1.1970150769074573e-06,
      "loss": 0.0306,
      "step": 55560
    },
    {
      "epoch": 2.820701558454744,
      "grad_norm": 0.34994304180145264,
      "learning_rate": 1.1953229436350408e-06,
      "loss": 0.0276,
      "step": 55565
    },
    {
      "epoch": 2.8209553784456065,
      "grad_norm": 0.34854093194007874,
      "learning_rate": 1.1936308103626244e-06,
      "loss": 0.031,
      "step": 55570
    },
    {
      "epoch": 2.8212091984364687,
      "grad_norm": 0.3053123652935028,
      "learning_rate": 1.1919386770902077e-06,
      "loss": 0.0298,
      "step": 55575
    },
    {
      "epoch": 2.8214630184273313,
      "grad_norm": 0.3606565296649933,
      "learning_rate": 1.1902465438177913e-06,
      "loss": 0.0282,
      "step": 55580
    },
    {
      "epoch": 2.821716838418194,
      "grad_norm": 0.3627634048461914,
      "learning_rate": 1.1885544105453746e-06,
      "loss": 0.031,
      "step": 55585
    },
    {
      "epoch": 2.8219706584090565,
      "grad_norm": 0.30210357904434204,
      "learning_rate": 1.1868622772729581e-06,
      "loss": 0.0233,
      "step": 55590
    },
    {
      "epoch": 2.8222244783999186,
      "grad_norm": 0.318068265914917,
      "learning_rate": 1.1851701440005415e-06,
      "loss": 0.0287,
      "step": 55595
    },
    {
      "epoch": 2.8224782983907812,
      "grad_norm": 0.24310652911663055,
      "learning_rate": 1.183478010728125e-06,
      "loss": 0.0317,
      "step": 55600
    },
    {
      "epoch": 2.822732118381644,
      "grad_norm": 0.33054736256599426,
      "learning_rate": 1.1817858774557086e-06,
      "loss": 0.0297,
      "step": 55605
    },
    {
      "epoch": 2.822985938372506,
      "grad_norm": 0.33596691489219666,
      "learning_rate": 1.1800937441832921e-06,
      "loss": 0.0331,
      "step": 55610
    },
    {
      "epoch": 2.8232397583633686,
      "grad_norm": 0.3044825792312622,
      "learning_rate": 1.1784016109108754e-06,
      "loss": 0.0281,
      "step": 55615
    },
    {
      "epoch": 2.823493578354231,
      "grad_norm": 0.3748861253261566,
      "learning_rate": 1.176709477638459e-06,
      "loss": 0.0272,
      "step": 55620
    },
    {
      "epoch": 2.823747398345094,
      "grad_norm": 0.28317007422447205,
      "learning_rate": 1.1750173443660423e-06,
      "loss": 0.027,
      "step": 55625
    },
    {
      "epoch": 2.8240012183359564,
      "grad_norm": 0.3502991795539856,
      "learning_rate": 1.1733252110936259e-06,
      "loss": 0.0318,
      "step": 55630
    },
    {
      "epoch": 2.8242550383268186,
      "grad_norm": 0.42141517996788025,
      "learning_rate": 1.1716330778212092e-06,
      "loss": 0.0389,
      "step": 55635
    },
    {
      "epoch": 2.824508858317681,
      "grad_norm": 0.30312493443489075,
      "learning_rate": 1.1699409445487927e-06,
      "loss": 0.0276,
      "step": 55640
    },
    {
      "epoch": 2.8247626783085433,
      "grad_norm": 0.34598830342292786,
      "learning_rate": 1.1682488112763763e-06,
      "loss": 0.0298,
      "step": 55645
    },
    {
      "epoch": 2.825016498299406,
      "grad_norm": 0.5238077640533447,
      "learning_rate": 1.1665566780039596e-06,
      "loss": 0.0283,
      "step": 55650
    },
    {
      "epoch": 2.8252703182902685,
      "grad_norm": 0.2969505190849304,
      "learning_rate": 1.1648645447315432e-06,
      "loss": 0.0256,
      "step": 55655
    },
    {
      "epoch": 2.825524138281131,
      "grad_norm": 0.18817229568958282,
      "learning_rate": 1.1631724114591267e-06,
      "loss": 0.0243,
      "step": 55660
    },
    {
      "epoch": 2.8257779582719937,
      "grad_norm": 0.2759462594985962,
      "learning_rate": 1.16148027818671e-06,
      "loss": 0.0291,
      "step": 55665
    },
    {
      "epoch": 2.826031778262856,
      "grad_norm": 0.26524046063423157,
      "learning_rate": 1.1597881449142936e-06,
      "loss": 0.0374,
      "step": 55670
    },
    {
      "epoch": 2.8262855982537185,
      "grad_norm": 0.2896076440811157,
      "learning_rate": 1.158096011641877e-06,
      "loss": 0.0283,
      "step": 55675
    },
    {
      "epoch": 2.826539418244581,
      "grad_norm": 0.598273754119873,
      "learning_rate": 1.1564038783694605e-06,
      "loss": 0.0348,
      "step": 55680
    },
    {
      "epoch": 2.8267932382354433,
      "grad_norm": 0.38251566886901855,
      "learning_rate": 1.154711745097044e-06,
      "loss": 0.0281,
      "step": 55685
    },
    {
      "epoch": 2.827047058226306,
      "grad_norm": 0.32450079917907715,
      "learning_rate": 1.1530196118246273e-06,
      "loss": 0.0327,
      "step": 55690
    },
    {
      "epoch": 2.8273008782171685,
      "grad_norm": 0.24178165197372437,
      "learning_rate": 1.1513274785522109e-06,
      "loss": 0.0259,
      "step": 55695
    },
    {
      "epoch": 2.827554698208031,
      "grad_norm": 0.4267941117286682,
      "learning_rate": 1.1496353452797944e-06,
      "loss": 0.0364,
      "step": 55700
    },
    {
      "epoch": 2.8278085181988932,
      "grad_norm": 0.2857091426849365,
      "learning_rate": 1.1479432120073778e-06,
      "loss": 0.0295,
      "step": 55705
    },
    {
      "epoch": 2.828062338189756,
      "grad_norm": 0.24375779926776886,
      "learning_rate": 1.1462510787349613e-06,
      "loss": 0.0227,
      "step": 55710
    },
    {
      "epoch": 2.8283161581806184,
      "grad_norm": 0.3215489685535431,
      "learning_rate": 1.1445589454625446e-06,
      "loss": 0.0341,
      "step": 55715
    },
    {
      "epoch": 2.8285699781714806,
      "grad_norm": 0.23902076482772827,
      "learning_rate": 1.1428668121901282e-06,
      "loss": 0.0302,
      "step": 55720
    },
    {
      "epoch": 2.828823798162343,
      "grad_norm": 0.3301446735858917,
      "learning_rate": 1.1411746789177117e-06,
      "loss": 0.0301,
      "step": 55725
    },
    {
      "epoch": 2.829077618153206,
      "grad_norm": 0.516697108745575,
      "learning_rate": 1.139482545645295e-06,
      "loss": 0.0381,
      "step": 55730
    },
    {
      "epoch": 2.8293314381440684,
      "grad_norm": 0.30288365483283997,
      "learning_rate": 1.1377904123728786e-06,
      "loss": 0.0317,
      "step": 55735
    },
    {
      "epoch": 2.8295852581349306,
      "grad_norm": 0.42075833678245544,
      "learning_rate": 1.1360982791004622e-06,
      "loss": 0.0278,
      "step": 55740
    },
    {
      "epoch": 2.829839078125793,
      "grad_norm": 0.5544710755348206,
      "learning_rate": 1.1344061458280455e-06,
      "loss": 0.0245,
      "step": 55745
    },
    {
      "epoch": 2.8300928981166558,
      "grad_norm": 0.5366796851158142,
      "learning_rate": 1.132714012555629e-06,
      "loss": 0.0281,
      "step": 55750
    },
    {
      "epoch": 2.830346718107518,
      "grad_norm": 0.2638997435569763,
      "learning_rate": 1.1310218792832124e-06,
      "loss": 0.0318,
      "step": 55755
    },
    {
      "epoch": 2.8306005380983805,
      "grad_norm": 0.449851393699646,
      "learning_rate": 1.129329746010796e-06,
      "loss": 0.0354,
      "step": 55760
    },
    {
      "epoch": 2.830854358089243,
      "grad_norm": 0.4421868324279785,
      "learning_rate": 1.1276376127383794e-06,
      "loss": 0.0262,
      "step": 55765
    },
    {
      "epoch": 2.8311081780801057,
      "grad_norm": 0.18787184357643127,
      "learning_rate": 1.1259454794659628e-06,
      "loss": 0.0308,
      "step": 55770
    },
    {
      "epoch": 2.8313619980709683,
      "grad_norm": 0.7075026631355286,
      "learning_rate": 1.1242533461935463e-06,
      "loss": 0.0309,
      "step": 55775
    },
    {
      "epoch": 2.8316158180618305,
      "grad_norm": 0.17095202207565308,
      "learning_rate": 1.1225612129211299e-06,
      "loss": 0.0243,
      "step": 55780
    },
    {
      "epoch": 2.831869638052693,
      "grad_norm": 0.38210830092430115,
      "learning_rate": 1.1208690796487132e-06,
      "loss": 0.0353,
      "step": 55785
    },
    {
      "epoch": 2.8321234580435553,
      "grad_norm": 0.23966944217681885,
      "learning_rate": 1.1191769463762967e-06,
      "loss": 0.0292,
      "step": 55790
    },
    {
      "epoch": 2.832377278034418,
      "grad_norm": 0.18793806433677673,
      "learning_rate": 1.11748481310388e-06,
      "loss": 0.0326,
      "step": 55795
    },
    {
      "epoch": 2.8326310980252805,
      "grad_norm": 0.652092695236206,
      "learning_rate": 1.1157926798314636e-06,
      "loss": 0.0278,
      "step": 55800
    },
    {
      "epoch": 2.832884918016143,
      "grad_norm": 0.30326351523399353,
      "learning_rate": 1.1141005465590472e-06,
      "loss": 0.0267,
      "step": 55805
    },
    {
      "epoch": 2.8331387380070057,
      "grad_norm": 0.34887462854385376,
      "learning_rate": 1.1124084132866305e-06,
      "loss": 0.0313,
      "step": 55810
    },
    {
      "epoch": 2.833392557997868,
      "grad_norm": 0.48567840456962585,
      "learning_rate": 1.110716280014214e-06,
      "loss": 0.0318,
      "step": 55815
    },
    {
      "epoch": 2.8336463779887304,
      "grad_norm": 0.22675849497318268,
      "learning_rate": 1.1090241467417976e-06,
      "loss": 0.0284,
      "step": 55820
    },
    {
      "epoch": 2.833900197979593,
      "grad_norm": 0.36345478892326355,
      "learning_rate": 1.107332013469381e-06,
      "loss": 0.028,
      "step": 55825
    },
    {
      "epoch": 2.834154017970455,
      "grad_norm": 0.24584625661373138,
      "learning_rate": 1.1056398801969643e-06,
      "loss": 0.0317,
      "step": 55830
    },
    {
      "epoch": 2.834407837961318,
      "grad_norm": 0.18756209313869476,
      "learning_rate": 1.1039477469245478e-06,
      "loss": 0.0279,
      "step": 55835
    },
    {
      "epoch": 2.8346616579521804,
      "grad_norm": 0.2258630394935608,
      "learning_rate": 1.1022556136521313e-06,
      "loss": 0.0256,
      "step": 55840
    },
    {
      "epoch": 2.834915477943043,
      "grad_norm": 0.39259475469589233,
      "learning_rate": 1.1005634803797149e-06,
      "loss": 0.0267,
      "step": 55845
    },
    {
      "epoch": 2.835169297933905,
      "grad_norm": 0.43295443058013916,
      "learning_rate": 1.0988713471072982e-06,
      "loss": 0.0394,
      "step": 55850
    },
    {
      "epoch": 2.8354231179247678,
      "grad_norm": 0.5019866228103638,
      "learning_rate": 1.0971792138348818e-06,
      "loss": 0.0359,
      "step": 55855
    },
    {
      "epoch": 2.8356769379156304,
      "grad_norm": 0.3963499069213867,
      "learning_rate": 1.0954870805624653e-06,
      "loss": 0.0316,
      "step": 55860
    },
    {
      "epoch": 2.8359307579064925,
      "grad_norm": 0.5169487595558167,
      "learning_rate": 1.0937949472900486e-06,
      "loss": 0.0368,
      "step": 55865
    },
    {
      "epoch": 2.836184577897355,
      "grad_norm": 0.3360297679901123,
      "learning_rate": 1.092102814017632e-06,
      "loss": 0.0338,
      "step": 55870
    },
    {
      "epoch": 2.8364383978882177,
      "grad_norm": 0.3319953680038452,
      "learning_rate": 1.0904106807452155e-06,
      "loss": 0.0289,
      "step": 55875
    },
    {
      "epoch": 2.8366922178790803,
      "grad_norm": 0.24195292592048645,
      "learning_rate": 1.088718547472799e-06,
      "loss": 0.0266,
      "step": 55880
    },
    {
      "epoch": 2.8369460378699425,
      "grad_norm": 0.327484130859375,
      "learning_rate": 1.0870264142003826e-06,
      "loss": 0.0274,
      "step": 55885
    },
    {
      "epoch": 2.837199857860805,
      "grad_norm": 0.3096347451210022,
      "learning_rate": 1.085334280927966e-06,
      "loss": 0.0255,
      "step": 55890
    },
    {
      "epoch": 2.8374536778516677,
      "grad_norm": 0.2832336723804474,
      "learning_rate": 1.0836421476555495e-06,
      "loss": 0.0422,
      "step": 55895
    },
    {
      "epoch": 2.83770749784253,
      "grad_norm": 0.38523250818252563,
      "learning_rate": 1.081950014383133e-06,
      "loss": 0.0344,
      "step": 55900
    },
    {
      "epoch": 2.8379613178333924,
      "grad_norm": 0.331540584564209,
      "learning_rate": 1.0802578811107164e-06,
      "loss": 0.0305,
      "step": 55905
    },
    {
      "epoch": 2.838215137824255,
      "grad_norm": 0.25240784883499146,
      "learning_rate": 1.0785657478382997e-06,
      "loss": 0.0321,
      "step": 55910
    },
    {
      "epoch": 2.8384689578151177,
      "grad_norm": 0.26983770728111267,
      "learning_rate": 1.0768736145658832e-06,
      "loss": 0.027,
      "step": 55915
    },
    {
      "epoch": 2.8387227778059803,
      "grad_norm": 0.30436670780181885,
      "learning_rate": 1.0751814812934668e-06,
      "loss": 0.0244,
      "step": 55920
    },
    {
      "epoch": 2.8389765977968424,
      "grad_norm": 0.2570286989212036,
      "learning_rate": 1.0734893480210503e-06,
      "loss": 0.0286,
      "step": 55925
    },
    {
      "epoch": 2.839230417787705,
      "grad_norm": 0.3244872987270355,
      "learning_rate": 1.0717972147486337e-06,
      "loss": 0.0287,
      "step": 55930
    },
    {
      "epoch": 2.839484237778567,
      "grad_norm": 0.2520430088043213,
      "learning_rate": 1.0701050814762172e-06,
      "loss": 0.0339,
      "step": 55935
    },
    {
      "epoch": 2.83973805776943,
      "grad_norm": 0.2672327160835266,
      "learning_rate": 1.0684129482038008e-06,
      "loss": 0.0219,
      "step": 55940
    },
    {
      "epoch": 2.8399918777602924,
      "grad_norm": 0.2933124899864197,
      "learning_rate": 1.066720814931384e-06,
      "loss": 0.0357,
      "step": 55945
    },
    {
      "epoch": 2.840245697751155,
      "grad_norm": 0.43850472569465637,
      "learning_rate": 1.0650286816589674e-06,
      "loss": 0.0321,
      "step": 55950
    },
    {
      "epoch": 2.8404995177420176,
      "grad_norm": 0.412503719329834,
      "learning_rate": 1.063336548386551e-06,
      "loss": 0.0319,
      "step": 55955
    },
    {
      "epoch": 2.8407533377328797,
      "grad_norm": 0.3230607509613037,
      "learning_rate": 1.0616444151141345e-06,
      "loss": 0.0291,
      "step": 55960
    },
    {
      "epoch": 2.8410071577237423,
      "grad_norm": 0.25596120953559875,
      "learning_rate": 1.0599522818417178e-06,
      "loss": 0.0347,
      "step": 55965
    },
    {
      "epoch": 2.841260977714605,
      "grad_norm": 0.2851473093032837,
      "learning_rate": 1.0582601485693014e-06,
      "loss": 0.0307,
      "step": 55970
    },
    {
      "epoch": 2.841514797705467,
      "grad_norm": 0.3201811611652374,
      "learning_rate": 1.056568015296885e-06,
      "loss": 0.0296,
      "step": 55975
    },
    {
      "epoch": 2.8417686176963297,
      "grad_norm": 0.3185413181781769,
      "learning_rate": 1.0548758820244685e-06,
      "loss": 0.0304,
      "step": 55980
    },
    {
      "epoch": 2.8420224376871923,
      "grad_norm": 0.3229440748691559,
      "learning_rate": 1.0531837487520518e-06,
      "loss": 0.0314,
      "step": 55985
    },
    {
      "epoch": 2.842276257678055,
      "grad_norm": 0.39506053924560547,
      "learning_rate": 1.0514916154796351e-06,
      "loss": 0.0296,
      "step": 55990
    },
    {
      "epoch": 2.842530077668917,
      "grad_norm": 0.37298551201820374,
      "learning_rate": 1.0497994822072187e-06,
      "loss": 0.0321,
      "step": 55995
    },
    {
      "epoch": 2.8427838976597797,
      "grad_norm": 0.4710085690021515,
      "learning_rate": 1.0481073489348022e-06,
      "loss": 0.0377,
      "step": 56000
    },
    {
      "epoch": 2.8430377176506423,
      "grad_norm": 0.2344180792570114,
      "learning_rate": 1.0464152156623856e-06,
      "loss": 0.034,
      "step": 56005
    },
    {
      "epoch": 2.8432915376415044,
      "grad_norm": 0.42132675647735596,
      "learning_rate": 1.0447230823899691e-06,
      "loss": 0.0272,
      "step": 56010
    },
    {
      "epoch": 2.843545357632367,
      "grad_norm": 0.29779860377311707,
      "learning_rate": 1.0430309491175527e-06,
      "loss": 0.0289,
      "step": 56015
    },
    {
      "epoch": 2.8437991776232296,
      "grad_norm": 0.1912485510110855,
      "learning_rate": 1.0413388158451362e-06,
      "loss": 0.025,
      "step": 56020
    },
    {
      "epoch": 2.8440529976140922,
      "grad_norm": 0.33178281784057617,
      "learning_rate": 1.0396466825727195e-06,
      "loss": 0.029,
      "step": 56025
    },
    {
      "epoch": 2.8443068176049544,
      "grad_norm": 0.28970810770988464,
      "learning_rate": 1.0379545493003029e-06,
      "loss": 0.0282,
      "step": 56030
    },
    {
      "epoch": 2.844560637595817,
      "grad_norm": 0.3231714069843292,
      "learning_rate": 1.0362624160278864e-06,
      "loss": 0.0353,
      "step": 56035
    },
    {
      "epoch": 2.8448144575866796,
      "grad_norm": 0.4571632742881775,
      "learning_rate": 1.03457028275547e-06,
      "loss": 0.0315,
      "step": 56040
    },
    {
      "epoch": 2.8450682775775418,
      "grad_norm": 0.31172847747802734,
      "learning_rate": 1.0328781494830533e-06,
      "loss": 0.036,
      "step": 56045
    },
    {
      "epoch": 2.8453220975684044,
      "grad_norm": 0.5730637907981873,
      "learning_rate": 1.0311860162106368e-06,
      "loss": 0.0227,
      "step": 56050
    },
    {
      "epoch": 2.845575917559267,
      "grad_norm": 0.3472713232040405,
      "learning_rate": 1.0294938829382204e-06,
      "loss": 0.0233,
      "step": 56055
    },
    {
      "epoch": 2.8458297375501296,
      "grad_norm": 0.554363489151001,
      "learning_rate": 1.027801749665804e-06,
      "loss": 0.0312,
      "step": 56060
    },
    {
      "epoch": 2.846083557540992,
      "grad_norm": 0.38508570194244385,
      "learning_rate": 1.0261096163933873e-06,
      "loss": 0.0311,
      "step": 56065
    },
    {
      "epoch": 2.8463373775318543,
      "grad_norm": 0.3509955108165741,
      "learning_rate": 1.0244174831209706e-06,
      "loss": 0.0279,
      "step": 56070
    },
    {
      "epoch": 2.846591197522717,
      "grad_norm": 0.2738310992717743,
      "learning_rate": 1.0227253498485541e-06,
      "loss": 0.0352,
      "step": 56075
    },
    {
      "epoch": 2.846845017513579,
      "grad_norm": 0.2927628755569458,
      "learning_rate": 1.0210332165761377e-06,
      "loss": 0.028,
      "step": 56080
    },
    {
      "epoch": 2.8470988375044417,
      "grad_norm": 0.3419632017612457,
      "learning_rate": 1.019341083303721e-06,
      "loss": 0.0321,
      "step": 56085
    },
    {
      "epoch": 2.8473526574953043,
      "grad_norm": 0.26105278730392456,
      "learning_rate": 1.0176489500313046e-06,
      "loss": 0.0235,
      "step": 56090
    },
    {
      "epoch": 2.847606477486167,
      "grad_norm": 0.35791948437690735,
      "learning_rate": 1.015956816758888e-06,
      "loss": 0.0211,
      "step": 56095
    },
    {
      "epoch": 2.8478602974770295,
      "grad_norm": 0.4351104199886322,
      "learning_rate": 1.0142646834864714e-06,
      "loss": 0.0249,
      "step": 56100
    },
    {
      "epoch": 2.8481141174678917,
      "grad_norm": 0.31804075837135315,
      "learning_rate": 1.012572550214055e-06,
      "loss": 0.0288,
      "step": 56105
    },
    {
      "epoch": 2.8483679374587543,
      "grad_norm": 0.3467938005924225,
      "learning_rate": 1.0108804169416383e-06,
      "loss": 0.033,
      "step": 56110
    },
    {
      "epoch": 2.848621757449617,
      "grad_norm": 0.2748010754585266,
      "learning_rate": 1.0091882836692218e-06,
      "loss": 0.0314,
      "step": 56115
    },
    {
      "epoch": 2.848875577440479,
      "grad_norm": 0.23437300324440002,
      "learning_rate": 1.0074961503968054e-06,
      "loss": 0.0271,
      "step": 56120
    },
    {
      "epoch": 2.8491293974313416,
      "grad_norm": 0.3459556996822357,
      "learning_rate": 1.0058040171243887e-06,
      "loss": 0.0329,
      "step": 56125
    },
    {
      "epoch": 2.8493832174222042,
      "grad_norm": 0.33637312054634094,
      "learning_rate": 1.0041118838519723e-06,
      "loss": 0.0289,
      "step": 56130
    },
    {
      "epoch": 2.849637037413067,
      "grad_norm": 0.27937188744544983,
      "learning_rate": 1.0024197505795558e-06,
      "loss": 0.0321,
      "step": 56135
    },
    {
      "epoch": 2.849890857403929,
      "grad_norm": 0.21973158419132233,
      "learning_rate": 1.0007276173071391e-06,
      "loss": 0.0301,
      "step": 56140
    },
    {
      "epoch": 2.8501446773947916,
      "grad_norm": 0.5233383774757385,
      "learning_rate": 9.990354840347227e-07,
      "loss": 0.0283,
      "step": 56145
    },
    {
      "epoch": 2.850398497385654,
      "grad_norm": 0.3020778000354767,
      "learning_rate": 9.97343350762306e-07,
      "loss": 0.0265,
      "step": 56150
    },
    {
      "epoch": 2.8506523173765164,
      "grad_norm": 0.7818123698234558,
      "learning_rate": 9.956512174898896e-07,
      "loss": 0.0327,
      "step": 56155
    },
    {
      "epoch": 2.850906137367379,
      "grad_norm": 0.41146236658096313,
      "learning_rate": 9.939590842174731e-07,
      "loss": 0.0399,
      "step": 56160
    },
    {
      "epoch": 2.8511599573582416,
      "grad_norm": 0.3017003536224365,
      "learning_rate": 9.922669509450564e-07,
      "loss": 0.0287,
      "step": 56165
    },
    {
      "epoch": 2.851413777349104,
      "grad_norm": 0.2691115438938141,
      "learning_rate": 9.9057481767264e-07,
      "loss": 0.0261,
      "step": 56170
    },
    {
      "epoch": 2.8516675973399663,
      "grad_norm": 0.4630514979362488,
      "learning_rate": 9.888826844002235e-07,
      "loss": 0.0304,
      "step": 56175
    },
    {
      "epoch": 2.851921417330829,
      "grad_norm": 0.4147006869316101,
      "learning_rate": 9.871905511278069e-07,
      "loss": 0.0279,
      "step": 56180
    },
    {
      "epoch": 2.8521752373216915,
      "grad_norm": 0.3111046254634857,
      "learning_rate": 9.854984178553904e-07,
      "loss": 0.0283,
      "step": 56185
    },
    {
      "epoch": 2.8524290573125537,
      "grad_norm": 0.42208629846572876,
      "learning_rate": 9.838062845829737e-07,
      "loss": 0.0317,
      "step": 56190
    },
    {
      "epoch": 2.8526828773034163,
      "grad_norm": 0.3102574348449707,
      "learning_rate": 9.821141513105573e-07,
      "loss": 0.0269,
      "step": 56195
    },
    {
      "epoch": 2.852936697294279,
      "grad_norm": 0.29505234956741333,
      "learning_rate": 9.804220180381408e-07,
      "loss": 0.0266,
      "step": 56200
    },
    {
      "epoch": 2.8531905172851415,
      "grad_norm": 0.2698964476585388,
      "learning_rate": 9.787298847657242e-07,
      "loss": 0.031,
      "step": 56205
    },
    {
      "epoch": 2.853444337276004,
      "grad_norm": 0.722200334072113,
      "learning_rate": 9.770377514933077e-07,
      "loss": 0.0329,
      "step": 56210
    },
    {
      "epoch": 2.8536981572668663,
      "grad_norm": 0.41778144240379333,
      "learning_rate": 9.753456182208913e-07,
      "loss": 0.0313,
      "step": 56215
    },
    {
      "epoch": 2.853951977257729,
      "grad_norm": 0.33847302198410034,
      "learning_rate": 9.736534849484746e-07,
      "loss": 0.0264,
      "step": 56220
    },
    {
      "epoch": 2.854205797248591,
      "grad_norm": 0.36504846811294556,
      "learning_rate": 9.719613516760581e-07,
      "loss": 0.0334,
      "step": 56225
    },
    {
      "epoch": 2.8544596172394536,
      "grad_norm": 0.3869740664958954,
      "learning_rate": 9.702692184036415e-07,
      "loss": 0.0328,
      "step": 56230
    },
    {
      "epoch": 2.8547134372303162,
      "grad_norm": 0.41756024956703186,
      "learning_rate": 9.68577085131225e-07,
      "loss": 0.0298,
      "step": 56235
    },
    {
      "epoch": 2.854967257221179,
      "grad_norm": 0.3898419439792633,
      "learning_rate": 9.668849518588086e-07,
      "loss": 0.0318,
      "step": 56240
    },
    {
      "epoch": 2.8552210772120414,
      "grad_norm": 0.2647220194339752,
      "learning_rate": 9.651928185863919e-07,
      "loss": 0.0311,
      "step": 56245
    },
    {
      "epoch": 2.8554748972029036,
      "grad_norm": 0.4688816964626312,
      "learning_rate": 9.635006853139754e-07,
      "loss": 0.0319,
      "step": 56250
    },
    {
      "epoch": 2.855728717193766,
      "grad_norm": 0.36906033754348755,
      "learning_rate": 9.61808552041559e-07,
      "loss": 0.0275,
      "step": 56255
    },
    {
      "epoch": 2.855982537184629,
      "grad_norm": 0.4400111138820648,
      "learning_rate": 9.601164187691423e-07,
      "loss": 0.0286,
      "step": 56260
    },
    {
      "epoch": 2.856236357175491,
      "grad_norm": 0.2533106505870819,
      "learning_rate": 9.584242854967259e-07,
      "loss": 0.0241,
      "step": 56265
    },
    {
      "epoch": 2.8564901771663536,
      "grad_norm": 0.29996147751808167,
      "learning_rate": 9.567321522243092e-07,
      "loss": 0.0299,
      "step": 56270
    },
    {
      "epoch": 2.856743997157216,
      "grad_norm": 0.2994527220726013,
      "learning_rate": 9.550400189518927e-07,
      "loss": 0.0298,
      "step": 56275
    },
    {
      "epoch": 2.8569978171480788,
      "grad_norm": 0.34880349040031433,
      "learning_rate": 9.533478856794762e-07,
      "loss": 0.0247,
      "step": 56280
    },
    {
      "epoch": 2.857251637138941,
      "grad_norm": 0.43103355169296265,
      "learning_rate": 9.516557524070596e-07,
      "loss": 0.037,
      "step": 56285
    },
    {
      "epoch": 2.8575054571298035,
      "grad_norm": 0.7479355931282043,
      "learning_rate": 9.499636191346432e-07,
      "loss": 0.0298,
      "step": 56290
    },
    {
      "epoch": 2.857759277120666,
      "grad_norm": 0.27349087595939636,
      "learning_rate": 9.482714858622266e-07,
      "loss": 0.0244,
      "step": 56295
    },
    {
      "epoch": 2.8580130971115283,
      "grad_norm": 0.6969117522239685,
      "learning_rate": 9.465793525898101e-07,
      "loss": 0.0278,
      "step": 56300
    },
    {
      "epoch": 2.858266917102391,
      "grad_norm": 0.30732694268226624,
      "learning_rate": 9.448872193173936e-07,
      "loss": 0.0295,
      "step": 56305
    },
    {
      "epoch": 2.8585207370932535,
      "grad_norm": 0.3011256158351898,
      "learning_rate": 9.431950860449769e-07,
      "loss": 0.0267,
      "step": 56310
    },
    {
      "epoch": 2.858774557084116,
      "grad_norm": 1.600549578666687,
      "learning_rate": 9.415029527725603e-07,
      "loss": 0.0254,
      "step": 56315
    },
    {
      "epoch": 2.8590283770749787,
      "grad_norm": 0.5621958374977112,
      "learning_rate": 9.398108195001439e-07,
      "loss": 0.0373,
      "step": 56320
    },
    {
      "epoch": 2.859282197065841,
      "grad_norm": 0.2704627513885498,
      "learning_rate": 9.381186862277273e-07,
      "loss": 0.0272,
      "step": 56325
    },
    {
      "epoch": 2.8595360170567035,
      "grad_norm": 0.3137623965740204,
      "learning_rate": 9.364265529553109e-07,
      "loss": 0.0319,
      "step": 56330
    },
    {
      "epoch": 2.8597898370475656,
      "grad_norm": 0.3069267272949219,
      "learning_rate": 9.347344196828943e-07,
      "loss": 0.0285,
      "step": 56335
    },
    {
      "epoch": 2.860043657038428,
      "grad_norm": 0.2506422698497772,
      "learning_rate": 9.330422864104779e-07,
      "loss": 0.0217,
      "step": 56340
    },
    {
      "epoch": 2.860297477029291,
      "grad_norm": 0.30039626359939575,
      "learning_rate": 9.313501531380613e-07,
      "loss": 0.0241,
      "step": 56345
    },
    {
      "epoch": 2.8605512970201534,
      "grad_norm": 0.3857613503932953,
      "learning_rate": 9.296580198656446e-07,
      "loss": 0.0286,
      "step": 56350
    },
    {
      "epoch": 2.860805117011016,
      "grad_norm": 0.4006558656692505,
      "learning_rate": 9.279658865932281e-07,
      "loss": 0.0284,
      "step": 56355
    },
    {
      "epoch": 2.861058937001878,
      "grad_norm": 0.28864726424217224,
      "learning_rate": 9.262737533208116e-07,
      "loss": 0.0362,
      "step": 56360
    },
    {
      "epoch": 2.861312756992741,
      "grad_norm": 0.31609058380126953,
      "learning_rate": 9.245816200483951e-07,
      "loss": 0.0254,
      "step": 56365
    },
    {
      "epoch": 2.8615665769836034,
      "grad_norm": 0.46757835149765015,
      "learning_rate": 9.228894867759786e-07,
      "loss": 0.0291,
      "step": 56370
    },
    {
      "epoch": 2.8618203969744656,
      "grad_norm": 0.46682101488113403,
      "learning_rate": 9.21197353503562e-07,
      "loss": 0.0265,
      "step": 56375
    },
    {
      "epoch": 2.862074216965328,
      "grad_norm": 0.599969208240509,
      "learning_rate": 9.195052202311455e-07,
      "loss": 0.0289,
      "step": 56380
    },
    {
      "epoch": 2.8623280369561908,
      "grad_norm": 0.39302515983581543,
      "learning_rate": 9.17813086958729e-07,
      "loss": 0.0307,
      "step": 56385
    },
    {
      "epoch": 2.8625818569470534,
      "grad_norm": 0.2010660022497177,
      "learning_rate": 9.161209536863124e-07,
      "loss": 0.0275,
      "step": 56390
    },
    {
      "epoch": 2.8628356769379155,
      "grad_norm": 0.30817094445228577,
      "learning_rate": 9.144288204138958e-07,
      "loss": 0.026,
      "step": 56395
    },
    {
      "epoch": 2.863089496928778,
      "grad_norm": 0.543203592300415,
      "learning_rate": 9.127366871414793e-07,
      "loss": 0.03,
      "step": 56400
    },
    {
      "epoch": 2.8633433169196407,
      "grad_norm": 0.3976742923259735,
      "learning_rate": 9.110445538690628e-07,
      "loss": 0.0294,
      "step": 56405
    },
    {
      "epoch": 2.863597136910503,
      "grad_norm": 0.28443464636802673,
      "learning_rate": 9.093524205966463e-07,
      "loss": 0.0315,
      "step": 56410
    },
    {
      "epoch": 2.8638509569013655,
      "grad_norm": 0.3199726343154907,
      "learning_rate": 9.076602873242298e-07,
      "loss": 0.0293,
      "step": 56415
    },
    {
      "epoch": 2.864104776892228,
      "grad_norm": 0.27055320143699646,
      "learning_rate": 9.059681540518132e-07,
      "loss": 0.026,
      "step": 56420
    },
    {
      "epoch": 2.8643585968830907,
      "grad_norm": 0.336502343416214,
      "learning_rate": 9.042760207793967e-07,
      "loss": 0.0198,
      "step": 56425
    },
    {
      "epoch": 2.864612416873953,
      "grad_norm": 0.33029425144195557,
      "learning_rate": 9.025838875069801e-07,
      "loss": 0.0288,
      "step": 56430
    },
    {
      "epoch": 2.8648662368648155,
      "grad_norm": 0.3340581953525543,
      "learning_rate": 9.008917542345635e-07,
      "loss": 0.0248,
      "step": 56435
    },
    {
      "epoch": 2.865120056855678,
      "grad_norm": 0.47649848461151123,
      "learning_rate": 8.991996209621471e-07,
      "loss": 0.0283,
      "step": 56440
    },
    {
      "epoch": 2.86537387684654,
      "grad_norm": 0.6935096979141235,
      "learning_rate": 8.975074876897305e-07,
      "loss": 0.0334,
      "step": 56445
    },
    {
      "epoch": 2.865627696837403,
      "grad_norm": 0.30507203936576843,
      "learning_rate": 8.958153544173139e-07,
      "loss": 0.0286,
      "step": 56450
    },
    {
      "epoch": 2.8658815168282654,
      "grad_norm": 0.30034294724464417,
      "learning_rate": 8.941232211448975e-07,
      "loss": 0.0333,
      "step": 56455
    },
    {
      "epoch": 2.866135336819128,
      "grad_norm": 0.3122558295726776,
      "learning_rate": 8.924310878724809e-07,
      "loss": 0.0338,
      "step": 56460
    },
    {
      "epoch": 2.8663891568099906,
      "grad_norm": 0.5069664716720581,
      "learning_rate": 8.907389546000645e-07,
      "loss": 0.027,
      "step": 56465
    },
    {
      "epoch": 2.866642976800853,
      "grad_norm": 0.26773950457572937,
      "learning_rate": 8.890468213276478e-07,
      "loss": 0.0275,
      "step": 56470
    },
    {
      "epoch": 2.8668967967917154,
      "grad_norm": 0.46711573004722595,
      "learning_rate": 8.873546880552312e-07,
      "loss": 0.03,
      "step": 56475
    },
    {
      "epoch": 2.8671506167825775,
      "grad_norm": 0.5285343527793884,
      "learning_rate": 8.856625547828148e-07,
      "loss": 0.0329,
      "step": 56480
    },
    {
      "epoch": 2.86740443677344,
      "grad_norm": 0.32522931694984436,
      "learning_rate": 8.839704215103982e-07,
      "loss": 0.0326,
      "step": 56485
    },
    {
      "epoch": 2.8676582567643027,
      "grad_norm": 0.21851252019405365,
      "learning_rate": 8.822782882379817e-07,
      "loss": 0.0244,
      "step": 56490
    },
    {
      "epoch": 2.8679120767551654,
      "grad_norm": 0.3424266576766968,
      "learning_rate": 8.805861549655652e-07,
      "loss": 0.0306,
      "step": 56495
    },
    {
      "epoch": 2.868165896746028,
      "grad_norm": 0.27859312295913696,
      "learning_rate": 8.788940216931486e-07,
      "loss": 0.0323,
      "step": 56500
    },
    {
      "epoch": 2.86841971673689,
      "grad_norm": 0.3467095196247101,
      "learning_rate": 8.772018884207322e-07,
      "loss": 0.0296,
      "step": 56505
    },
    {
      "epoch": 2.8686735367277527,
      "grad_norm": 0.2763430178165436,
      "learning_rate": 8.755097551483155e-07,
      "loss": 0.0241,
      "step": 56510
    },
    {
      "epoch": 2.8689273567186153,
      "grad_norm": 0.5489804744720459,
      "learning_rate": 8.73817621875899e-07,
      "loss": 0.0351,
      "step": 56515
    },
    {
      "epoch": 2.8691811767094775,
      "grad_norm": 0.22821855545043945,
      "learning_rate": 8.721254886034825e-07,
      "loss": 0.0299,
      "step": 56520
    },
    {
      "epoch": 2.86943499670034,
      "grad_norm": 0.26971396803855896,
      "learning_rate": 8.704333553310659e-07,
      "loss": 0.029,
      "step": 56525
    },
    {
      "epoch": 2.8696888166912027,
      "grad_norm": 0.35413697361946106,
      "learning_rate": 8.687412220586494e-07,
      "loss": 0.0328,
      "step": 56530
    },
    {
      "epoch": 2.8699426366820653,
      "grad_norm": 0.58110511302948,
      "learning_rate": 8.670490887862329e-07,
      "loss": 0.0304,
      "step": 56535
    },
    {
      "epoch": 2.8701964566729274,
      "grad_norm": 0.40751639008522034,
      "learning_rate": 8.653569555138164e-07,
      "loss": 0.0277,
      "step": 56540
    },
    {
      "epoch": 2.87045027666379,
      "grad_norm": 0.27954888343811035,
      "learning_rate": 8.636648222413999e-07,
      "loss": 0.0326,
      "step": 56545
    },
    {
      "epoch": 2.8707040966546526,
      "grad_norm": 0.29035764932632446,
      "learning_rate": 8.619726889689832e-07,
      "loss": 0.0219,
      "step": 56550
    },
    {
      "epoch": 2.870957916645515,
      "grad_norm": 0.30947911739349365,
      "learning_rate": 8.602805556965667e-07,
      "loss": 0.0294,
      "step": 56555
    },
    {
      "epoch": 2.8712117366363774,
      "grad_norm": 0.22087480127811432,
      "learning_rate": 8.585884224241501e-07,
      "loss": 0.0347,
      "step": 56560
    },
    {
      "epoch": 2.87146555662724,
      "grad_norm": 0.2977849245071411,
      "learning_rate": 8.568962891517337e-07,
      "loss": 0.0283,
      "step": 56565
    },
    {
      "epoch": 2.8717193766181026,
      "grad_norm": 0.34214112162590027,
      "learning_rate": 8.552041558793171e-07,
      "loss": 0.0275,
      "step": 56570
    },
    {
      "epoch": 2.8719731966089648,
      "grad_norm": 0.2554374039173126,
      "learning_rate": 8.535120226069006e-07,
      "loss": 0.0325,
      "step": 56575
    },
    {
      "epoch": 2.8722270165998274,
      "grad_norm": 0.4054310917854309,
      "learning_rate": 8.518198893344841e-07,
      "loss": 0.0283,
      "step": 56580
    },
    {
      "epoch": 2.87248083659069,
      "grad_norm": 0.26180627942085266,
      "learning_rate": 8.501277560620675e-07,
      "loss": 0.0242,
      "step": 56585
    },
    {
      "epoch": 2.872734656581552,
      "grad_norm": 0.41069480776786804,
      "learning_rate": 8.48435622789651e-07,
      "loss": 0.035,
      "step": 56590
    },
    {
      "epoch": 2.8729884765724147,
      "grad_norm": 0.2548157870769501,
      "learning_rate": 8.467434895172344e-07,
      "loss": 0.0236,
      "step": 56595
    },
    {
      "epoch": 2.8732422965632773,
      "grad_norm": 0.35240301489830017,
      "learning_rate": 8.450513562448178e-07,
      "loss": 0.0239,
      "step": 56600
    },
    {
      "epoch": 2.87349611655414,
      "grad_norm": 0.38352060317993164,
      "learning_rate": 8.433592229724014e-07,
      "loss": 0.034,
      "step": 56605
    },
    {
      "epoch": 2.8737499365450025,
      "grad_norm": 0.3092139661312103,
      "learning_rate": 8.416670896999848e-07,
      "loss": 0.0348,
      "step": 56610
    },
    {
      "epoch": 2.8740037565358647,
      "grad_norm": 0.3200393319129944,
      "learning_rate": 8.399749564275684e-07,
      "loss": 0.034,
      "step": 56615
    },
    {
      "epoch": 2.8742575765267273,
      "grad_norm": 0.31458497047424316,
      "learning_rate": 8.382828231551518e-07,
      "loss": 0.0338,
      "step": 56620
    },
    {
      "epoch": 2.8745113965175895,
      "grad_norm": 0.4197658896446228,
      "learning_rate": 8.365906898827352e-07,
      "loss": 0.0319,
      "step": 56625
    },
    {
      "epoch": 2.874765216508452,
      "grad_norm": 0.2996726632118225,
      "learning_rate": 8.348985566103187e-07,
      "loss": 0.0334,
      "step": 56630
    },
    {
      "epoch": 2.8750190364993147,
      "grad_norm": 0.3455791771411896,
      "learning_rate": 8.332064233379021e-07,
      "loss": 0.0277,
      "step": 56635
    },
    {
      "epoch": 2.8752728564901773,
      "grad_norm": 0.30782240629196167,
      "learning_rate": 8.315142900654856e-07,
      "loss": 0.0331,
      "step": 56640
    },
    {
      "epoch": 2.87552667648104,
      "grad_norm": 0.584852397441864,
      "learning_rate": 8.298221567930691e-07,
      "loss": 0.036,
      "step": 56645
    },
    {
      "epoch": 2.875780496471902,
      "grad_norm": 0.41536372900009155,
      "learning_rate": 8.281300235206525e-07,
      "loss": 0.0309,
      "step": 56650
    },
    {
      "epoch": 2.8760343164627646,
      "grad_norm": 0.3044881522655487,
      "learning_rate": 8.264378902482361e-07,
      "loss": 0.0262,
      "step": 56655
    },
    {
      "epoch": 2.8762881364536272,
      "grad_norm": 0.2610401213169098,
      "learning_rate": 8.247457569758195e-07,
      "loss": 0.0226,
      "step": 56660
    },
    {
      "epoch": 2.8765419564444894,
      "grad_norm": 0.3302207589149475,
      "learning_rate": 8.23053623703403e-07,
      "loss": 0.0249,
      "step": 56665
    },
    {
      "epoch": 2.876795776435352,
      "grad_norm": 0.3605993092060089,
      "learning_rate": 8.213614904309865e-07,
      "loss": 0.0303,
      "step": 56670
    },
    {
      "epoch": 2.8770495964262146,
      "grad_norm": 0.38024815917015076,
      "learning_rate": 8.196693571585698e-07,
      "loss": 0.0245,
      "step": 56675
    },
    {
      "epoch": 2.877303416417077,
      "grad_norm": 0.33744293451309204,
      "learning_rate": 8.179772238861533e-07,
      "loss": 0.0309,
      "step": 56680
    },
    {
      "epoch": 2.8775572364079394,
      "grad_norm": 0.3316429853439331,
      "learning_rate": 8.162850906137368e-07,
      "loss": 0.0289,
      "step": 56685
    },
    {
      "epoch": 2.877811056398802,
      "grad_norm": 0.37818485498428345,
      "learning_rate": 8.145929573413203e-07,
      "loss": 0.0296,
      "step": 56690
    },
    {
      "epoch": 2.8780648763896646,
      "grad_norm": 0.4408175051212311,
      "learning_rate": 8.129008240689037e-07,
      "loss": 0.035,
      "step": 56695
    },
    {
      "epoch": 2.8783186963805267,
      "grad_norm": 0.3286103308200836,
      "learning_rate": 8.112086907964872e-07,
      "loss": 0.0305,
      "step": 56700
    },
    {
      "epoch": 2.8785725163713893,
      "grad_norm": 0.6499526500701904,
      "learning_rate": 8.095165575240707e-07,
      "loss": 0.0297,
      "step": 56705
    },
    {
      "epoch": 2.878826336362252,
      "grad_norm": 0.8811410069465637,
      "learning_rate": 8.078244242516542e-07,
      "loss": 0.0308,
      "step": 56710
    },
    {
      "epoch": 2.8790801563531145,
      "grad_norm": 0.6163033843040466,
      "learning_rate": 8.061322909792376e-07,
      "loss": 0.0359,
      "step": 56715
    },
    {
      "epoch": 2.8793339763439767,
      "grad_norm": 0.48688793182373047,
      "learning_rate": 8.04440157706821e-07,
      "loss": 0.0281,
      "step": 56720
    },
    {
      "epoch": 2.8795877963348393,
      "grad_norm": 0.5693960189819336,
      "learning_rate": 8.027480244344045e-07,
      "loss": 0.026,
      "step": 56725
    },
    {
      "epoch": 2.879841616325702,
      "grad_norm": 0.24559521675109863,
      "learning_rate": 8.01055891161988e-07,
      "loss": 0.029,
      "step": 56730
    },
    {
      "epoch": 2.880095436316564,
      "grad_norm": 0.38832828402519226,
      "learning_rate": 7.993637578895714e-07,
      "loss": 0.0371,
      "step": 56735
    },
    {
      "epoch": 2.8803492563074267,
      "grad_norm": 0.32193049788475037,
      "learning_rate": 7.97671624617155e-07,
      "loss": 0.0353,
      "step": 56740
    },
    {
      "epoch": 2.8806030762982893,
      "grad_norm": 0.2351982444524765,
      "learning_rate": 7.959794913447384e-07,
      "loss": 0.0237,
      "step": 56745
    },
    {
      "epoch": 2.880856896289152,
      "grad_norm": 0.24744842946529388,
      "learning_rate": 7.94287358072322e-07,
      "loss": 0.0244,
      "step": 56750
    },
    {
      "epoch": 2.8811107162800145,
      "grad_norm": 0.4948106110095978,
      "learning_rate": 7.925952247999053e-07,
      "loss": 0.042,
      "step": 56755
    },
    {
      "epoch": 2.8813645362708766,
      "grad_norm": 0.36634719371795654,
      "learning_rate": 7.909030915274887e-07,
      "loss": 0.0388,
      "step": 56760
    },
    {
      "epoch": 2.8816183562617392,
      "grad_norm": 0.3451341390609741,
      "learning_rate": 7.892109582550723e-07,
      "loss": 0.035,
      "step": 56765
    },
    {
      "epoch": 2.8818721762526014,
      "grad_norm": 0.3345429599285126,
      "learning_rate": 7.875188249826557e-07,
      "loss": 0.0323,
      "step": 56770
    },
    {
      "epoch": 2.882125996243464,
      "grad_norm": 0.35555848479270935,
      "learning_rate": 7.858266917102391e-07,
      "loss": 0.0284,
      "step": 56775
    },
    {
      "epoch": 2.8823798162343266,
      "grad_norm": 0.36172765493392944,
      "learning_rate": 7.841345584378227e-07,
      "loss": 0.0362,
      "step": 56780
    },
    {
      "epoch": 2.882633636225189,
      "grad_norm": 0.262870192527771,
      "learning_rate": 7.824424251654061e-07,
      "loss": 0.03,
      "step": 56785
    },
    {
      "epoch": 2.882887456216052,
      "grad_norm": 0.5119378566741943,
      "learning_rate": 7.807502918929897e-07,
      "loss": 0.0331,
      "step": 56790
    },
    {
      "epoch": 2.883141276206914,
      "grad_norm": 0.7077087759971619,
      "learning_rate": 7.79058158620573e-07,
      "loss": 0.0265,
      "step": 56795
    },
    {
      "epoch": 2.8833950961977766,
      "grad_norm": 0.31209030747413635,
      "learning_rate": 7.773660253481564e-07,
      "loss": 0.0311,
      "step": 56800
    },
    {
      "epoch": 2.883648916188639,
      "grad_norm": 0.5953086018562317,
      "learning_rate": 7.756738920757399e-07,
      "loss": 0.0343,
      "step": 56805
    },
    {
      "epoch": 2.8839027361795013,
      "grad_norm": 0.38027554750442505,
      "learning_rate": 7.739817588033234e-07,
      "loss": 0.0277,
      "step": 56810
    },
    {
      "epoch": 2.884156556170364,
      "grad_norm": 0.3770337700843811,
      "learning_rate": 7.722896255309069e-07,
      "loss": 0.0287,
      "step": 56815
    },
    {
      "epoch": 2.8844103761612265,
      "grad_norm": 0.3363008201122284,
      "learning_rate": 7.705974922584904e-07,
      "loss": 0.0334,
      "step": 56820
    },
    {
      "epoch": 2.884664196152089,
      "grad_norm": 1.4316260814666748,
      "learning_rate": 7.689053589860738e-07,
      "loss": 0.0302,
      "step": 56825
    },
    {
      "epoch": 2.8849180161429513,
      "grad_norm": 0.36405399441719055,
      "learning_rate": 7.672132257136573e-07,
      "loss": 0.0239,
      "step": 56830
    },
    {
      "epoch": 2.885171836133814,
      "grad_norm": 0.36887475848197937,
      "learning_rate": 7.655210924412407e-07,
      "loss": 0.0328,
      "step": 56835
    },
    {
      "epoch": 2.8854256561246765,
      "grad_norm": 0.25625044107437134,
      "learning_rate": 7.638289591688242e-07,
      "loss": 0.0287,
      "step": 56840
    },
    {
      "epoch": 2.8856794761155387,
      "grad_norm": 0.25802773237228394,
      "learning_rate": 7.621368258964076e-07,
      "loss": 0.0221,
      "step": 56845
    },
    {
      "epoch": 2.8859332961064013,
      "grad_norm": 0.3497041165828705,
      "learning_rate": 7.604446926239911e-07,
      "loss": 0.0292,
      "step": 56850
    },
    {
      "epoch": 2.886187116097264,
      "grad_norm": 0.3454679548740387,
      "learning_rate": 7.587525593515746e-07,
      "loss": 0.0331,
      "step": 56855
    },
    {
      "epoch": 2.8864409360881265,
      "grad_norm": 0.5030285716056824,
      "learning_rate": 7.570604260791581e-07,
      "loss": 0.0359,
      "step": 56860
    },
    {
      "epoch": 2.8866947560789886,
      "grad_norm": 0.44325920939445496,
      "learning_rate": 7.553682928067416e-07,
      "loss": 0.0296,
      "step": 56865
    },
    {
      "epoch": 2.8869485760698512,
      "grad_norm": 0.4056631326675415,
      "learning_rate": 7.53676159534325e-07,
      "loss": 0.0272,
      "step": 56870
    },
    {
      "epoch": 2.887202396060714,
      "grad_norm": 0.26784148812294006,
      "learning_rate": 7.519840262619083e-07,
      "loss": 0.0262,
      "step": 56875
    },
    {
      "epoch": 2.887456216051576,
      "grad_norm": 0.33540356159210205,
      "learning_rate": 7.502918929894919e-07,
      "loss": 0.035,
      "step": 56880
    },
    {
      "epoch": 2.8877100360424386,
      "grad_norm": 0.30647772550582886,
      "learning_rate": 7.485997597170753e-07,
      "loss": 0.0313,
      "step": 56885
    },
    {
      "epoch": 2.887963856033301,
      "grad_norm": 0.25061747431755066,
      "learning_rate": 7.469076264446589e-07,
      "loss": 0.0249,
      "step": 56890
    },
    {
      "epoch": 2.888217676024164,
      "grad_norm": 0.5146888494491577,
      "learning_rate": 7.452154931722423e-07,
      "loss": 0.0272,
      "step": 56895
    },
    {
      "epoch": 2.8884714960150264,
      "grad_norm": 0.2966390550136566,
      "learning_rate": 7.435233598998258e-07,
      "loss": 0.0299,
      "step": 56900
    },
    {
      "epoch": 2.8887253160058886,
      "grad_norm": 0.37970107793807983,
      "learning_rate": 7.418312266274093e-07,
      "loss": 0.0322,
      "step": 56905
    },
    {
      "epoch": 2.888979135996751,
      "grad_norm": 0.2674563527107239,
      "learning_rate": 7.401390933549927e-07,
      "loss": 0.0245,
      "step": 56910
    },
    {
      "epoch": 2.8892329559876133,
      "grad_norm": 0.44810232520103455,
      "learning_rate": 7.384469600825761e-07,
      "loss": 0.0248,
      "step": 56915
    },
    {
      "epoch": 2.889486775978476,
      "grad_norm": 0.25780561566352844,
      "learning_rate": 7.367548268101596e-07,
      "loss": 0.0242,
      "step": 56920
    },
    {
      "epoch": 2.8897405959693385,
      "grad_norm": 0.35203060507774353,
      "learning_rate": 7.35062693537743e-07,
      "loss": 0.0284,
      "step": 56925
    },
    {
      "epoch": 2.889994415960201,
      "grad_norm": 0.26767706871032715,
      "learning_rate": 7.333705602653266e-07,
      "loss": 0.0262,
      "step": 56930
    },
    {
      "epoch": 2.8902482359510637,
      "grad_norm": 0.3971843719482422,
      "learning_rate": 7.3167842699291e-07,
      "loss": 0.0454,
      "step": 56935
    },
    {
      "epoch": 2.890502055941926,
      "grad_norm": 0.22796855866909027,
      "learning_rate": 7.299862937204935e-07,
      "loss": 0.0319,
      "step": 56940
    },
    {
      "epoch": 2.8907558759327885,
      "grad_norm": 0.32220014929771423,
      "learning_rate": 7.28294160448077e-07,
      "loss": 0.0233,
      "step": 56945
    },
    {
      "epoch": 2.891009695923651,
      "grad_norm": 0.27687567472457886,
      "learning_rate": 7.266020271756604e-07,
      "loss": 0.0278,
      "step": 56950
    },
    {
      "epoch": 2.8912635159145132,
      "grad_norm": 0.2914799749851227,
      "learning_rate": 7.249098939032438e-07,
      "loss": 0.0265,
      "step": 56955
    },
    {
      "epoch": 2.891517335905376,
      "grad_norm": 0.3295184373855591,
      "learning_rate": 7.232177606308273e-07,
      "loss": 0.0264,
      "step": 56960
    },
    {
      "epoch": 2.8917711558962385,
      "grad_norm": 0.2902180850505829,
      "learning_rate": 7.215256273584108e-07,
      "loss": 0.0259,
      "step": 56965
    },
    {
      "epoch": 2.892024975887101,
      "grad_norm": 0.21984949707984924,
      "learning_rate": 7.198334940859943e-07,
      "loss": 0.0201,
      "step": 56970
    },
    {
      "epoch": 2.892278795877963,
      "grad_norm": 0.2158031314611435,
      "learning_rate": 7.181413608135777e-07,
      "loss": 0.0327,
      "step": 56975
    },
    {
      "epoch": 2.892532615868826,
      "grad_norm": 0.27695468068122864,
      "learning_rate": 7.164492275411612e-07,
      "loss": 0.0281,
      "step": 56980
    },
    {
      "epoch": 2.8927864358596884,
      "grad_norm": 0.23929698765277863,
      "learning_rate": 7.147570942687447e-07,
      "loss": 0.0273,
      "step": 56985
    },
    {
      "epoch": 2.8930402558505506,
      "grad_norm": 0.3177143335342407,
      "learning_rate": 7.130649609963282e-07,
      "loss": 0.0367,
      "step": 56990
    },
    {
      "epoch": 2.893294075841413,
      "grad_norm": 0.3839656114578247,
      "learning_rate": 7.113728277239115e-07,
      "loss": 0.0342,
      "step": 56995
    },
    {
      "epoch": 2.893547895832276,
      "grad_norm": 0.30691877007484436,
      "learning_rate": 7.09680694451495e-07,
      "loss": 0.0257,
      "step": 57000
    },
    {
      "epoch": 2.8938017158231384,
      "grad_norm": 0.2793743312358856,
      "learning_rate": 7.079885611790785e-07,
      "loss": 0.0404,
      "step": 57005
    },
    {
      "epoch": 2.8940555358140005,
      "grad_norm": 0.5192366242408752,
      "learning_rate": 7.062964279066619e-07,
      "loss": 0.0299,
      "step": 57010
    },
    {
      "epoch": 2.894309355804863,
      "grad_norm": 0.3496776521205902,
      "learning_rate": 7.046042946342455e-07,
      "loss": 0.0346,
      "step": 57015
    },
    {
      "epoch": 2.8945631757957258,
      "grad_norm": 0.24379898607730865,
      "learning_rate": 7.029121613618289e-07,
      "loss": 0.0283,
      "step": 57020
    },
    {
      "epoch": 2.894816995786588,
      "grad_norm": 0.41297459602355957,
      "learning_rate": 7.012200280894125e-07,
      "loss": 0.023,
      "step": 57025
    },
    {
      "epoch": 2.8950708157774505,
      "grad_norm": 0.2837767004966736,
      "learning_rate": 6.995278948169959e-07,
      "loss": 0.0295,
      "step": 57030
    },
    {
      "epoch": 2.895324635768313,
      "grad_norm": 0.33692866563796997,
      "learning_rate": 6.978357615445792e-07,
      "loss": 0.0325,
      "step": 57035
    },
    {
      "epoch": 2.8955784557591757,
      "grad_norm": 0.32200172543525696,
      "learning_rate": 6.961436282721628e-07,
      "loss": 0.0299,
      "step": 57040
    },
    {
      "epoch": 2.8958322757500383,
      "grad_norm": 0.3966142535209656,
      "learning_rate": 6.944514949997462e-07,
      "loss": 0.0302,
      "step": 57045
    },
    {
      "epoch": 2.8960860957409005,
      "grad_norm": 0.36363843083381653,
      "learning_rate": 6.927593617273296e-07,
      "loss": 0.0283,
      "step": 57050
    },
    {
      "epoch": 2.896339915731763,
      "grad_norm": 0.3507053852081299,
      "learning_rate": 6.910672284549132e-07,
      "loss": 0.028,
      "step": 57055
    },
    {
      "epoch": 2.8965937357226252,
      "grad_norm": 0.2757592499256134,
      "learning_rate": 6.893750951824966e-07,
      "loss": 0.0231,
      "step": 57060
    },
    {
      "epoch": 2.896847555713488,
      "grad_norm": 0.3220710754394531,
      "learning_rate": 6.876829619100802e-07,
      "loss": 0.0277,
      "step": 57065
    },
    {
      "epoch": 2.8971013757043504,
      "grad_norm": 0.25192391872406006,
      "learning_rate": 6.859908286376636e-07,
      "loss": 0.0283,
      "step": 57070
    },
    {
      "epoch": 2.897355195695213,
      "grad_norm": 0.40557751059532166,
      "learning_rate": 6.842986953652469e-07,
      "loss": 0.0357,
      "step": 57075
    },
    {
      "epoch": 2.8976090156860757,
      "grad_norm": 0.24660883843898773,
      "learning_rate": 6.826065620928305e-07,
      "loss": 0.0326,
      "step": 57080
    },
    {
      "epoch": 2.897862835676938,
      "grad_norm": 0.3630150556564331,
      "learning_rate": 6.809144288204139e-07,
      "loss": 0.0296,
      "step": 57085
    },
    {
      "epoch": 2.8981166556678004,
      "grad_norm": 0.5821228623390198,
      "learning_rate": 6.792222955479974e-07,
      "loss": 0.03,
      "step": 57090
    },
    {
      "epoch": 2.898370475658663,
      "grad_norm": 0.3000786602497101,
      "learning_rate": 6.775301622755809e-07,
      "loss": 0.0325,
      "step": 57095
    },
    {
      "epoch": 2.898624295649525,
      "grad_norm": 0.5774555206298828,
      "learning_rate": 6.758380290031643e-07,
      "loss": 0.0313,
      "step": 57100
    },
    {
      "epoch": 2.8988781156403878,
      "grad_norm": 0.23967984318733215,
      "learning_rate": 6.741458957307479e-07,
      "loss": 0.0317,
      "step": 57105
    },
    {
      "epoch": 2.8991319356312504,
      "grad_norm": 0.2716747522354126,
      "learning_rate": 6.724537624583313e-07,
      "loss": 0.0365,
      "step": 57110
    },
    {
      "epoch": 2.899385755622113,
      "grad_norm": 0.20131956040859222,
      "learning_rate": 6.707616291859147e-07,
      "loss": 0.029,
      "step": 57115
    },
    {
      "epoch": 2.899639575612975,
      "grad_norm": 0.2939135432243347,
      "learning_rate": 6.690694959134981e-07,
      "loss": 0.0263,
      "step": 57120
    },
    {
      "epoch": 2.8998933956038377,
      "grad_norm": 0.4463021755218506,
      "learning_rate": 6.673773626410816e-07,
      "loss": 0.0323,
      "step": 57125
    },
    {
      "epoch": 2.9001472155947003,
      "grad_norm": 0.36133143305778503,
      "learning_rate": 6.656852293686651e-07,
      "loss": 0.037,
      "step": 57130
    },
    {
      "epoch": 2.9004010355855625,
      "grad_norm": 0.28862321376800537,
      "learning_rate": 6.639930960962486e-07,
      "loss": 0.0256,
      "step": 57135
    },
    {
      "epoch": 2.900654855576425,
      "grad_norm": 0.3055947422981262,
      "learning_rate": 6.623009628238321e-07,
      "loss": 0.0334,
      "step": 57140
    },
    {
      "epoch": 2.9009086755672877,
      "grad_norm": 0.31747281551361084,
      "learning_rate": 6.606088295514155e-07,
      "loss": 0.0257,
      "step": 57145
    },
    {
      "epoch": 2.9011624955581503,
      "grad_norm": 0.3737865388393402,
      "learning_rate": 6.589166962789991e-07,
      "loss": 0.0325,
      "step": 57150
    },
    {
      "epoch": 2.901416315549013,
      "grad_norm": 0.31869587302207947,
      "learning_rate": 6.572245630065824e-07,
      "loss": 0.0358,
      "step": 57155
    },
    {
      "epoch": 2.901670135539875,
      "grad_norm": 0.33540067076683044,
      "learning_rate": 6.555324297341658e-07,
      "loss": 0.0277,
      "step": 57160
    },
    {
      "epoch": 2.9019239555307377,
      "grad_norm": 0.42595553398132324,
      "learning_rate": 6.538402964617494e-07,
      "loss": 0.0333,
      "step": 57165
    },
    {
      "epoch": 2.9021777755216,
      "grad_norm": 0.2767331302165985,
      "learning_rate": 6.521481631893328e-07,
      "loss": 0.0278,
      "step": 57170
    },
    {
      "epoch": 2.9024315955124624,
      "grad_norm": 0.31812888383865356,
      "learning_rate": 6.504560299169164e-07,
      "loss": 0.0263,
      "step": 57175
    },
    {
      "epoch": 2.902685415503325,
      "grad_norm": 0.27926456928253174,
      "learning_rate": 6.487638966444998e-07,
      "loss": 0.0285,
      "step": 57180
    },
    {
      "epoch": 2.9029392354941876,
      "grad_norm": 0.5825649499893188,
      "learning_rate": 6.470717633720832e-07,
      "loss": 0.0298,
      "step": 57185
    },
    {
      "epoch": 2.9031930554850502,
      "grad_norm": 0.35669177770614624,
      "learning_rate": 6.453796300996668e-07,
      "loss": 0.0377,
      "step": 57190
    },
    {
      "epoch": 2.9034468754759124,
      "grad_norm": 0.3390977382659912,
      "learning_rate": 6.436874968272502e-07,
      "loss": 0.0281,
      "step": 57195
    },
    {
      "epoch": 2.903700695466775,
      "grad_norm": 0.25372812151908875,
      "learning_rate": 6.419953635548335e-07,
      "loss": 0.0325,
      "step": 57200
    },
    {
      "epoch": 2.9039545154576376,
      "grad_norm": 0.48622772097587585,
      "learning_rate": 6.403032302824171e-07,
      "loss": 0.0322,
      "step": 57205
    },
    {
      "epoch": 2.9042083354484998,
      "grad_norm": 0.27143144607543945,
      "learning_rate": 6.386110970100005e-07,
      "loss": 0.0299,
      "step": 57210
    },
    {
      "epoch": 2.9044621554393624,
      "grad_norm": 0.3860352039337158,
      "learning_rate": 6.369189637375841e-07,
      "loss": 0.0412,
      "step": 57215
    },
    {
      "epoch": 2.904715975430225,
      "grad_norm": 0.34994158148765564,
      "learning_rate": 6.352268304651675e-07,
      "loss": 0.029,
      "step": 57220
    },
    {
      "epoch": 2.9049697954210876,
      "grad_norm": 0.3273712694644928,
      "learning_rate": 6.33534697192751e-07,
      "loss": 0.0279,
      "step": 57225
    },
    {
      "epoch": 2.9052236154119497,
      "grad_norm": 0.5091918110847473,
      "learning_rate": 6.318425639203345e-07,
      "loss": 0.0296,
      "step": 57230
    },
    {
      "epoch": 2.9054774354028123,
      "grad_norm": 0.33736932277679443,
      "learning_rate": 6.301504306479179e-07,
      "loss": 0.0319,
      "step": 57235
    },
    {
      "epoch": 2.905731255393675,
      "grad_norm": 0.19080248475074768,
      "learning_rate": 6.284582973755013e-07,
      "loss": 0.0273,
      "step": 57240
    },
    {
      "epoch": 2.905985075384537,
      "grad_norm": 0.4544998109340668,
      "learning_rate": 6.267661641030848e-07,
      "loss": 0.0302,
      "step": 57245
    },
    {
      "epoch": 2.9062388953753997,
      "grad_norm": 0.342433363199234,
      "learning_rate": 6.250740308306682e-07,
      "loss": 0.0297,
      "step": 57250
    },
    {
      "epoch": 2.9064927153662623,
      "grad_norm": 0.28809770941734314,
      "learning_rate": 6.233818975582517e-07,
      "loss": 0.0266,
      "step": 57255
    },
    {
      "epoch": 2.906746535357125,
      "grad_norm": 0.3601595461368561,
      "learning_rate": 6.216897642858352e-07,
      "loss": 0.0282,
      "step": 57260
    },
    {
      "epoch": 2.907000355347987,
      "grad_norm": 0.3118194341659546,
      "learning_rate": 6.199976310134187e-07,
      "loss": 0.0333,
      "step": 57265
    },
    {
      "epoch": 2.9072541753388497,
      "grad_norm": 0.28417107462882996,
      "learning_rate": 6.183054977410021e-07,
      "loss": 0.0222,
      "step": 57270
    },
    {
      "epoch": 2.9075079953297123,
      "grad_norm": 0.3415754437446594,
      "learning_rate": 6.166133644685855e-07,
      "loss": 0.0274,
      "step": 57275
    },
    {
      "epoch": 2.9077618153205744,
      "grad_norm": 0.34622934460639954,
      "learning_rate": 6.149212311961691e-07,
      "loss": 0.0242,
      "step": 57280
    },
    {
      "epoch": 2.908015635311437,
      "grad_norm": 0.2505379021167755,
      "learning_rate": 6.132290979237525e-07,
      "loss": 0.0259,
      "step": 57285
    },
    {
      "epoch": 2.9082694553022996,
      "grad_norm": 0.30466291308403015,
      "learning_rate": 6.11536964651336e-07,
      "loss": 0.0341,
      "step": 57290
    },
    {
      "epoch": 2.9085232752931622,
      "grad_norm": 0.33808496594429016,
      "learning_rate": 6.098448313789194e-07,
      "loss": 0.0312,
      "step": 57295
    },
    {
      "epoch": 2.908777095284025,
      "grad_norm": 0.38466644287109375,
      "learning_rate": 6.08152698106503e-07,
      "loss": 0.0384,
      "step": 57300
    },
    {
      "epoch": 2.909030915274887,
      "grad_norm": 0.41694745421409607,
      "learning_rate": 6.064605648340864e-07,
      "loss": 0.0285,
      "step": 57305
    },
    {
      "epoch": 2.9092847352657496,
      "grad_norm": 0.4306418001651764,
      "learning_rate": 6.047684315616698e-07,
      "loss": 0.0294,
      "step": 57310
    },
    {
      "epoch": 2.9095385552566118,
      "grad_norm": 0.35518577694892883,
      "learning_rate": 6.030762982892533e-07,
      "loss": 0.0276,
      "step": 57315
    },
    {
      "epoch": 2.9097923752474744,
      "grad_norm": 0.4232032001018524,
      "learning_rate": 6.013841650168368e-07,
      "loss": 0.0437,
      "step": 57320
    },
    {
      "epoch": 2.910046195238337,
      "grad_norm": 0.2620871067047119,
      "learning_rate": 5.996920317444203e-07,
      "loss": 0.029,
      "step": 57325
    },
    {
      "epoch": 2.9103000152291996,
      "grad_norm": 0.3478947877883911,
      "learning_rate": 5.979998984720037e-07,
      "loss": 0.0334,
      "step": 57330
    },
    {
      "epoch": 2.910553835220062,
      "grad_norm": 0.3235359489917755,
      "learning_rate": 5.963077651995871e-07,
      "loss": 0.0304,
      "step": 57335
    },
    {
      "epoch": 2.9108076552109243,
      "grad_norm": 0.4142983555793762,
      "learning_rate": 5.946156319271707e-07,
      "loss": 0.0341,
      "step": 57340
    },
    {
      "epoch": 2.911061475201787,
      "grad_norm": 0.3530924916267395,
      "learning_rate": 5.929234986547541e-07,
      "loss": 0.0366,
      "step": 57345
    },
    {
      "epoch": 2.9113152951926495,
      "grad_norm": 0.26504990458488464,
      "learning_rate": 5.912313653823376e-07,
      "loss": 0.0246,
      "step": 57350
    },
    {
      "epoch": 2.9115691151835117,
      "grad_norm": 0.16128307580947876,
      "learning_rate": 5.89539232109921e-07,
      "loss": 0.0268,
      "step": 57355
    },
    {
      "epoch": 2.9118229351743743,
      "grad_norm": 0.26564478874206543,
      "learning_rate": 5.878470988375045e-07,
      "loss": 0.0308,
      "step": 57360
    },
    {
      "epoch": 2.912076755165237,
      "grad_norm": 0.3302893042564392,
      "learning_rate": 5.86154965565088e-07,
      "loss": 0.0341,
      "step": 57365
    },
    {
      "epoch": 2.9123305751560995,
      "grad_norm": 0.2747202515602112,
      "learning_rate": 5.844628322926714e-07,
      "loss": 0.0299,
      "step": 57370
    },
    {
      "epoch": 2.9125843951469617,
      "grad_norm": 0.2258002609014511,
      "learning_rate": 5.827706990202549e-07,
      "loss": 0.0292,
      "step": 57375
    },
    {
      "epoch": 2.9128382151378243,
      "grad_norm": 0.3801233172416687,
      "learning_rate": 5.810785657478384e-07,
      "loss": 0.0297,
      "step": 57380
    },
    {
      "epoch": 2.913092035128687,
      "grad_norm": 0.3353663682937622,
      "learning_rate": 5.793864324754218e-07,
      "loss": 0.0289,
      "step": 57385
    },
    {
      "epoch": 2.913345855119549,
      "grad_norm": 0.3122028708457947,
      "learning_rate": 5.776942992030053e-07,
      "loss": 0.0298,
      "step": 57390
    },
    {
      "epoch": 2.9135996751104116,
      "grad_norm": 0.26420414447784424,
      "learning_rate": 5.760021659305887e-07,
      "loss": 0.0303,
      "step": 57395
    },
    {
      "epoch": 2.9138534951012742,
      "grad_norm": 0.5072758197784424,
      "learning_rate": 5.743100326581723e-07,
      "loss": 0.027,
      "step": 57400
    },
    {
      "epoch": 2.914107315092137,
      "grad_norm": 1.083329677581787,
      "learning_rate": 5.726178993857557e-07,
      "loss": 0.0313,
      "step": 57405
    },
    {
      "epoch": 2.914361135082999,
      "grad_norm": 0.2915167510509491,
      "learning_rate": 5.709257661133391e-07,
      "loss": 0.0289,
      "step": 57410
    },
    {
      "epoch": 2.9146149550738616,
      "grad_norm": 0.2800576388835907,
      "learning_rate": 5.692336328409226e-07,
      "loss": 0.0282,
      "step": 57415
    },
    {
      "epoch": 2.914868775064724,
      "grad_norm": 0.32842501997947693,
      "learning_rate": 5.675414995685061e-07,
      "loss": 0.0312,
      "step": 57420
    },
    {
      "epoch": 2.9151225950555864,
      "grad_norm": 0.28682515025138855,
      "learning_rate": 5.658493662960896e-07,
      "loss": 0.0257,
      "step": 57425
    },
    {
      "epoch": 2.915376415046449,
      "grad_norm": 0.28181514143943787,
      "learning_rate": 5.64157233023673e-07,
      "loss": 0.0241,
      "step": 57430
    },
    {
      "epoch": 2.9156302350373116,
      "grad_norm": 0.4202750027179718,
      "learning_rate": 5.624650997512564e-07,
      "loss": 0.0271,
      "step": 57435
    },
    {
      "epoch": 2.915884055028174,
      "grad_norm": 0.29935333132743835,
      "learning_rate": 5.6077296647884e-07,
      "loss": 0.0234,
      "step": 57440
    },
    {
      "epoch": 2.9161378750190368,
      "grad_norm": 0.38342633843421936,
      "learning_rate": 5.590808332064234e-07,
      "loss": 0.0228,
      "step": 57445
    },
    {
      "epoch": 2.916391695009899,
      "grad_norm": 0.2808830440044403,
      "learning_rate": 5.573886999340069e-07,
      "loss": 0.033,
      "step": 57450
    },
    {
      "epoch": 2.9166455150007615,
      "grad_norm": 0.3798730671405792,
      "learning_rate": 5.556965666615903e-07,
      "loss": 0.0295,
      "step": 57455
    },
    {
      "epoch": 2.9168993349916237,
      "grad_norm": 0.29656025767326355,
      "learning_rate": 5.540044333891737e-07,
      "loss": 0.0312,
      "step": 57460
    },
    {
      "epoch": 2.9171531549824863,
      "grad_norm": 0.520729660987854,
      "learning_rate": 5.523123001167573e-07,
      "loss": 0.0337,
      "step": 57465
    },
    {
      "epoch": 2.917406974973349,
      "grad_norm": 0.3630119562149048,
      "learning_rate": 5.506201668443407e-07,
      "loss": 0.0313,
      "step": 57470
    },
    {
      "epoch": 2.9176607949642115,
      "grad_norm": 0.3356974422931671,
      "learning_rate": 5.489280335719242e-07,
      "loss": 0.0287,
      "step": 57475
    },
    {
      "epoch": 2.917914614955074,
      "grad_norm": 0.22973386943340302,
      "learning_rate": 5.472359002995076e-07,
      "loss": 0.0267,
      "step": 57480
    },
    {
      "epoch": 2.9181684349459363,
      "grad_norm": 0.5934128761291504,
      "learning_rate": 5.455437670270911e-07,
      "loss": 0.031,
      "step": 57485
    },
    {
      "epoch": 2.918422254936799,
      "grad_norm": 0.24467769265174866,
      "learning_rate": 5.438516337546746e-07,
      "loss": 0.0236,
      "step": 57490
    },
    {
      "epoch": 2.9186760749276615,
      "grad_norm": 0.2057151347398758,
      "learning_rate": 5.42159500482258e-07,
      "loss": 0.024,
      "step": 57495
    },
    {
      "epoch": 2.9189298949185236,
      "grad_norm": 0.4651114344596863,
      "learning_rate": 5.404673672098415e-07,
      "loss": 0.0253,
      "step": 57500
    },
    {
      "epoch": 2.919183714909386,
      "grad_norm": 0.4996252954006195,
      "learning_rate": 5.38775233937425e-07,
      "loss": 0.0282,
      "step": 57505
    },
    {
      "epoch": 2.919437534900249,
      "grad_norm": 1.3613667488098145,
      "learning_rate": 5.370831006650084e-07,
      "loss": 0.0304,
      "step": 57510
    },
    {
      "epoch": 2.9196913548911114,
      "grad_norm": 0.30651411414146423,
      "learning_rate": 5.353909673925919e-07,
      "loss": 0.0305,
      "step": 57515
    },
    {
      "epoch": 2.9199451748819736,
      "grad_norm": 0.35898858308792114,
      "learning_rate": 5.336988341201753e-07,
      "loss": 0.03,
      "step": 57520
    },
    {
      "epoch": 2.920198994872836,
      "grad_norm": 0.27816376090049744,
      "learning_rate": 5.320067008477589e-07,
      "loss": 0.0307,
      "step": 57525
    },
    {
      "epoch": 2.920452814863699,
      "grad_norm": 0.561227560043335,
      "learning_rate": 5.303145675753423e-07,
      "loss": 0.0262,
      "step": 57530
    },
    {
      "epoch": 2.920706634854561,
      "grad_norm": 0.4042511582374573,
      "learning_rate": 5.286224343029257e-07,
      "loss": 0.0344,
      "step": 57535
    },
    {
      "epoch": 2.9209604548454235,
      "grad_norm": 0.519243061542511,
      "learning_rate": 5.269303010305092e-07,
      "loss": 0.0313,
      "step": 57540
    },
    {
      "epoch": 2.921214274836286,
      "grad_norm": 0.35028374195098877,
      "learning_rate": 5.252381677580927e-07,
      "loss": 0.0333,
      "step": 57545
    },
    {
      "epoch": 2.9214680948271488,
      "grad_norm": 0.4332272410392761,
      "learning_rate": 5.235460344856762e-07,
      "loss": 0.0286,
      "step": 57550
    },
    {
      "epoch": 2.921721914818011,
      "grad_norm": 0.2854234576225281,
      "learning_rate": 5.218539012132596e-07,
      "loss": 0.0321,
      "step": 57555
    },
    {
      "epoch": 2.9219757348088735,
      "grad_norm": 0.3323298394680023,
      "learning_rate": 5.20161767940843e-07,
      "loss": 0.0301,
      "step": 57560
    },
    {
      "epoch": 2.922229554799736,
      "grad_norm": 0.3795822858810425,
      "learning_rate": 5.184696346684266e-07,
      "loss": 0.0266,
      "step": 57565
    },
    {
      "epoch": 2.9224833747905983,
      "grad_norm": 0.3280698359012604,
      "learning_rate": 5.167775013960099e-07,
      "loss": 0.0248,
      "step": 57570
    },
    {
      "epoch": 2.922737194781461,
      "grad_norm": 0.3984909951686859,
      "learning_rate": 5.150853681235935e-07,
      "loss": 0.0293,
      "step": 57575
    },
    {
      "epoch": 2.9229910147723235,
      "grad_norm": 0.28477180004119873,
      "learning_rate": 5.133932348511769e-07,
      "loss": 0.0221,
      "step": 57580
    },
    {
      "epoch": 2.923244834763186,
      "grad_norm": 0.4867945909500122,
      "learning_rate": 5.117011015787604e-07,
      "loss": 0.0366,
      "step": 57585
    },
    {
      "epoch": 2.9234986547540487,
      "grad_norm": 0.5254274010658264,
      "learning_rate": 5.100089683063438e-07,
      "loss": 0.0286,
      "step": 57590
    },
    {
      "epoch": 2.923752474744911,
      "grad_norm": 0.3182218372821808,
      "learning_rate": 5.083168350339273e-07,
      "loss": 0.0236,
      "step": 57595
    },
    {
      "epoch": 2.9240062947357734,
      "grad_norm": 0.26986467838287354,
      "learning_rate": 5.066247017615108e-07,
      "loss": 0.0277,
      "step": 57600
    },
    {
      "epoch": 2.9242601147266356,
      "grad_norm": 0.7450768947601318,
      "learning_rate": 5.049325684890943e-07,
      "loss": 0.0278,
      "step": 57605
    },
    {
      "epoch": 2.924513934717498,
      "grad_norm": 0.3461737334728241,
      "learning_rate": 5.032404352166776e-07,
      "loss": 0.0296,
      "step": 57610
    },
    {
      "epoch": 2.924767754708361,
      "grad_norm": 0.36083826422691345,
      "learning_rate": 5.015483019442612e-07,
      "loss": 0.0275,
      "step": 57615
    },
    {
      "epoch": 2.9250215746992234,
      "grad_norm": 0.5787628293037415,
      "learning_rate": 4.998561686718446e-07,
      "loss": 0.0342,
      "step": 57620
    },
    {
      "epoch": 2.925275394690086,
      "grad_norm": 0.37813305854797363,
      "learning_rate": 4.981640353994282e-07,
      "loss": 0.0266,
      "step": 57625
    },
    {
      "epoch": 2.925529214680948,
      "grad_norm": 0.5129855275154114,
      "learning_rate": 4.964719021270115e-07,
      "loss": 0.0288,
      "step": 57630
    },
    {
      "epoch": 2.925783034671811,
      "grad_norm": 0.3590392470359802,
      "learning_rate": 4.94779768854595e-07,
      "loss": 0.026,
      "step": 57635
    },
    {
      "epoch": 2.9260368546626734,
      "grad_norm": 0.22634489834308624,
      "learning_rate": 4.930876355821785e-07,
      "loss": 0.0237,
      "step": 57640
    },
    {
      "epoch": 2.9262906746535355,
      "grad_norm": 0.4361048638820648,
      "learning_rate": 4.91395502309762e-07,
      "loss": 0.0256,
      "step": 57645
    },
    {
      "epoch": 2.926544494644398,
      "grad_norm": 0.2713603973388672,
      "learning_rate": 4.897033690373454e-07,
      "loss": 0.0246,
      "step": 57650
    },
    {
      "epoch": 2.9267983146352607,
      "grad_norm": 0.4372120797634125,
      "learning_rate": 4.880112357649289e-07,
      "loss": 0.0245,
      "step": 57655
    },
    {
      "epoch": 2.9270521346261233,
      "grad_norm": 0.2845839858055115,
      "learning_rate": 4.863191024925123e-07,
      "loss": 0.0331,
      "step": 57660
    },
    {
      "epoch": 2.9273059546169855,
      "grad_norm": 0.21914979815483093,
      "learning_rate": 4.846269692200959e-07,
      "loss": 0.0255,
      "step": 57665
    },
    {
      "epoch": 2.927559774607848,
      "grad_norm": 0.14833270013332367,
      "learning_rate": 4.829348359476792e-07,
      "loss": 0.0302,
      "step": 57670
    },
    {
      "epoch": 2.9278135945987107,
      "grad_norm": 0.25538668036460876,
      "learning_rate": 4.812427026752628e-07,
      "loss": 0.0234,
      "step": 57675
    },
    {
      "epoch": 2.928067414589573,
      "grad_norm": 0.47089454531669617,
      "learning_rate": 4.795505694028462e-07,
      "loss": 0.025,
      "step": 57680
    },
    {
      "epoch": 2.9283212345804355,
      "grad_norm": 0.3125099539756775,
      "learning_rate": 4.778584361304296e-07,
      "loss": 0.0275,
      "step": 57685
    },
    {
      "epoch": 2.928575054571298,
      "grad_norm": 0.2715403139591217,
      "learning_rate": 4.761663028580131e-07,
      "loss": 0.0343,
      "step": 57690
    },
    {
      "epoch": 2.9288288745621607,
      "grad_norm": 0.3333417475223541,
      "learning_rate": 4.7447416958559657e-07,
      "loss": 0.025,
      "step": 57695
    },
    {
      "epoch": 2.929082694553023,
      "grad_norm": 0.3405666649341583,
      "learning_rate": 4.7278203631318006e-07,
      "loss": 0.0291,
      "step": 57700
    },
    {
      "epoch": 2.9293365145438854,
      "grad_norm": 0.2805671691894531,
      "learning_rate": 4.7108990304076355e-07,
      "loss": 0.0354,
      "step": 57705
    },
    {
      "epoch": 2.929590334534748,
      "grad_norm": 0.43839380145072937,
      "learning_rate": 4.6939776976834704e-07,
      "loss": 0.0271,
      "step": 57710
    },
    {
      "epoch": 2.92984415452561,
      "grad_norm": 0.50478595495224,
      "learning_rate": 4.6770563649593043e-07,
      "loss": 0.0301,
      "step": 57715
    },
    {
      "epoch": 2.930097974516473,
      "grad_norm": 0.5127033591270447,
      "learning_rate": 4.660135032235139e-07,
      "loss": 0.0298,
      "step": 57720
    },
    {
      "epoch": 2.9303517945073354,
      "grad_norm": 0.31215181946754456,
      "learning_rate": 4.643213699510974e-07,
      "loss": 0.0321,
      "step": 57725
    },
    {
      "epoch": 2.930605614498198,
      "grad_norm": 0.35444653034210205,
      "learning_rate": 4.626292366786809e-07,
      "loss": 0.0316,
      "step": 57730
    },
    {
      "epoch": 2.9308594344890606,
      "grad_norm": 0.39015939831733704,
      "learning_rate": 4.609371034062643e-07,
      "loss": 0.0222,
      "step": 57735
    },
    {
      "epoch": 2.9311132544799228,
      "grad_norm": 0.9289717078208923,
      "learning_rate": 4.592449701338478e-07,
      "loss": 0.0307,
      "step": 57740
    },
    {
      "epoch": 2.9313670744707854,
      "grad_norm": 0.33057901263237,
      "learning_rate": 4.5755283686143127e-07,
      "loss": 0.0324,
      "step": 57745
    },
    {
      "epoch": 2.9316208944616475,
      "grad_norm": 0.31482622027397156,
      "learning_rate": 4.5586070358901476e-07,
      "loss": 0.0265,
      "step": 57750
    },
    {
      "epoch": 2.93187471445251,
      "grad_norm": 0.5130469799041748,
      "learning_rate": 4.5416857031659815e-07,
      "loss": 0.0276,
      "step": 57755
    },
    {
      "epoch": 2.9321285344433727,
      "grad_norm": 0.5404413938522339,
      "learning_rate": 4.5247643704418164e-07,
      "loss": 0.0329,
      "step": 57760
    },
    {
      "epoch": 2.9323823544342353,
      "grad_norm": 0.5473409295082092,
      "learning_rate": 4.5078430377176513e-07,
      "loss": 0.0295,
      "step": 57765
    },
    {
      "epoch": 2.932636174425098,
      "grad_norm": 0.30557894706726074,
      "learning_rate": 4.490921704993486e-07,
      "loss": 0.037,
      "step": 57770
    },
    {
      "epoch": 2.93288999441596,
      "grad_norm": 0.28683364391326904,
      "learning_rate": 4.47400037226932e-07,
      "loss": 0.0271,
      "step": 57775
    },
    {
      "epoch": 2.9331438144068227,
      "grad_norm": 0.3118693232536316,
      "learning_rate": 4.457079039545155e-07,
      "loss": 0.0264,
      "step": 57780
    },
    {
      "epoch": 2.9333976343976853,
      "grad_norm": 0.23888196051120758,
      "learning_rate": 4.44015770682099e-07,
      "loss": 0.0351,
      "step": 57785
    },
    {
      "epoch": 2.9336514543885475,
      "grad_norm": 0.36562538146972656,
      "learning_rate": 4.4232363740968243e-07,
      "loss": 0.036,
      "step": 57790
    },
    {
      "epoch": 2.93390527437941,
      "grad_norm": 0.26900336146354675,
      "learning_rate": 4.4063150413726587e-07,
      "loss": 0.0271,
      "step": 57795
    },
    {
      "epoch": 2.9341590943702727,
      "grad_norm": 0.23322747647762299,
      "learning_rate": 4.3893937086484936e-07,
      "loss": 0.0243,
      "step": 57800
    },
    {
      "epoch": 2.9344129143611353,
      "grad_norm": 0.3980046510696411,
      "learning_rate": 4.3724723759243285e-07,
      "loss": 0.0262,
      "step": 57805
    },
    {
      "epoch": 2.9346667343519974,
      "grad_norm": 0.3880888521671295,
      "learning_rate": 4.355551043200163e-07,
      "loss": 0.027,
      "step": 57810
    },
    {
      "epoch": 2.93492055434286,
      "grad_norm": 0.2841983139514923,
      "learning_rate": 4.3386297104759973e-07,
      "loss": 0.0331,
      "step": 57815
    },
    {
      "epoch": 2.9351743743337226,
      "grad_norm": 0.5500996112823486,
      "learning_rate": 4.321708377751832e-07,
      "loss": 0.0257,
      "step": 57820
    },
    {
      "epoch": 2.935428194324585,
      "grad_norm": 0.3994247615337372,
      "learning_rate": 4.3047870450276666e-07,
      "loss": 0.0337,
      "step": 57825
    },
    {
      "epoch": 2.9356820143154474,
      "grad_norm": 0.23893968760967255,
      "learning_rate": 4.2878657123035015e-07,
      "loss": 0.0373,
      "step": 57830
    },
    {
      "epoch": 2.93593583430631,
      "grad_norm": 0.39422041177749634,
      "learning_rate": 4.270944379579336e-07,
      "loss": 0.0282,
      "step": 57835
    },
    {
      "epoch": 2.9361896542971726,
      "grad_norm": 0.30572760105133057,
      "learning_rate": 4.254023046855171e-07,
      "loss": 0.0367,
      "step": 57840
    },
    {
      "epoch": 2.9364434742880348,
      "grad_norm": 0.23431634902954102,
      "learning_rate": 4.237101714131005e-07,
      "loss": 0.0294,
      "step": 57845
    },
    {
      "epoch": 2.9366972942788974,
      "grad_norm": 0.46416381001472473,
      "learning_rate": 4.22018038140684e-07,
      "loss": 0.0334,
      "step": 57850
    },
    {
      "epoch": 2.93695111426976,
      "grad_norm": 0.29117438197135925,
      "learning_rate": 4.2032590486826745e-07,
      "loss": 0.0255,
      "step": 57855
    },
    {
      "epoch": 2.937204934260622,
      "grad_norm": 0.2204287201166153,
      "learning_rate": 4.1863377159585094e-07,
      "loss": 0.0277,
      "step": 57860
    },
    {
      "epoch": 2.9374587542514847,
      "grad_norm": 0.2894386649131775,
      "learning_rate": 4.169416383234344e-07,
      "loss": 0.0247,
      "step": 57865
    },
    {
      "epoch": 2.9377125742423473,
      "grad_norm": 0.2967325747013092,
      "learning_rate": 4.1524950505101787e-07,
      "loss": 0.0255,
      "step": 57870
    },
    {
      "epoch": 2.93796639423321,
      "grad_norm": 0.26310840249061584,
      "learning_rate": 4.135573717786013e-07,
      "loss": 0.0335,
      "step": 57875
    },
    {
      "epoch": 2.9382202142240725,
      "grad_norm": 0.3313550353050232,
      "learning_rate": 4.1186523850618475e-07,
      "loss": 0.035,
      "step": 57880
    },
    {
      "epoch": 2.9384740342149347,
      "grad_norm": 0.3628803491592407,
      "learning_rate": 4.1017310523376824e-07,
      "loss": 0.0307,
      "step": 57885
    },
    {
      "epoch": 2.9387278542057973,
      "grad_norm": 0.2926108241081238,
      "learning_rate": 4.0848097196135173e-07,
      "loss": 0.0278,
      "step": 57890
    },
    {
      "epoch": 2.9389816741966595,
      "grad_norm": 0.2612711787223816,
      "learning_rate": 4.0678883868893517e-07,
      "loss": 0.0291,
      "step": 57895
    },
    {
      "epoch": 2.939235494187522,
      "grad_norm": 0.35889166593551636,
      "learning_rate": 4.050967054165186e-07,
      "loss": 0.0278,
      "step": 57900
    },
    {
      "epoch": 2.9394893141783847,
      "grad_norm": 0.34467950463294983,
      "learning_rate": 4.034045721441021e-07,
      "loss": 0.0264,
      "step": 57905
    },
    {
      "epoch": 2.9397431341692473,
      "grad_norm": 0.2898823320865631,
      "learning_rate": 4.017124388716856e-07,
      "loss": 0.0303,
      "step": 57910
    },
    {
      "epoch": 2.93999695416011,
      "grad_norm": 0.4356616735458374,
      "learning_rate": 4.0002030559926903e-07,
      "loss": 0.0296,
      "step": 57915
    },
    {
      "epoch": 2.940250774150972,
      "grad_norm": 0.46433550119400024,
      "learning_rate": 3.9832817232685247e-07,
      "loss": 0.0294,
      "step": 57920
    },
    {
      "epoch": 2.9405045941418346,
      "grad_norm": 0.4361477196216583,
      "learning_rate": 3.9663603905443596e-07,
      "loss": 0.033,
      "step": 57925
    },
    {
      "epoch": 2.9407584141326972,
      "grad_norm": 0.2655782103538513,
      "learning_rate": 3.9494390578201946e-07,
      "loss": 0.0339,
      "step": 57930
    },
    {
      "epoch": 2.9410122341235594,
      "grad_norm": 0.28838610649108887,
      "learning_rate": 3.9325177250960284e-07,
      "loss": 0.0277,
      "step": 57935
    },
    {
      "epoch": 2.941266054114422,
      "grad_norm": 0.32494908571243286,
      "learning_rate": 3.9155963923718633e-07,
      "loss": 0.0348,
      "step": 57940
    },
    {
      "epoch": 2.9415198741052846,
      "grad_norm": 0.23130418360233307,
      "learning_rate": 3.898675059647698e-07,
      "loss": 0.0311,
      "step": 57945
    },
    {
      "epoch": 2.941773694096147,
      "grad_norm": 0.3082735538482666,
      "learning_rate": 3.881753726923533e-07,
      "loss": 0.0302,
      "step": 57950
    },
    {
      "epoch": 2.9420275140870094,
      "grad_norm": 0.2921549379825592,
      "learning_rate": 3.864832394199367e-07,
      "loss": 0.0262,
      "step": 57955
    },
    {
      "epoch": 2.942281334077872,
      "grad_norm": 0.2690448760986328,
      "learning_rate": 3.847911061475202e-07,
      "loss": 0.0296,
      "step": 57960
    },
    {
      "epoch": 2.9425351540687346,
      "grad_norm": 0.27467915415763855,
      "learning_rate": 3.830989728751037e-07,
      "loss": 0.0311,
      "step": 57965
    },
    {
      "epoch": 2.9427889740595967,
      "grad_norm": 0.3122628927230835,
      "learning_rate": 3.814068396026872e-07,
      "loss": 0.0314,
      "step": 57970
    },
    {
      "epoch": 2.9430427940504593,
      "grad_norm": 0.20584361255168915,
      "learning_rate": 3.7971470633027056e-07,
      "loss": 0.0256,
      "step": 57975
    },
    {
      "epoch": 2.943296614041322,
      "grad_norm": 0.3449113667011261,
      "learning_rate": 3.7802257305785405e-07,
      "loss": 0.0316,
      "step": 57980
    },
    {
      "epoch": 2.9435504340321845,
      "grad_norm": 0.24136188626289368,
      "learning_rate": 3.7633043978543755e-07,
      "loss": 0.0254,
      "step": 57985
    },
    {
      "epoch": 2.943804254023047,
      "grad_norm": 0.314791738986969,
      "learning_rate": 3.7463830651302104e-07,
      "loss": 0.0296,
      "step": 57990
    },
    {
      "epoch": 2.9440580740139093,
      "grad_norm": 0.41983452439308167,
      "learning_rate": 3.729461732406044e-07,
      "loss": 0.0303,
      "step": 57995
    },
    {
      "epoch": 2.944311894004772,
      "grad_norm": 0.41986700892448425,
      "learning_rate": 3.712540399681879e-07,
      "loss": 0.0397,
      "step": 58000
    },
    {
      "epoch": 2.944565713995634,
      "grad_norm": 0.39581939578056335,
      "learning_rate": 3.695619066957714e-07,
      "loss": 0.0319,
      "step": 58005
    },
    {
      "epoch": 2.9448195339864967,
      "grad_norm": 0.33107373118400574,
      "learning_rate": 3.678697734233549e-07,
      "loss": 0.0375,
      "step": 58010
    },
    {
      "epoch": 2.9450733539773593,
      "grad_norm": 0.25613731145858765,
      "learning_rate": 3.661776401509383e-07,
      "loss": 0.0309,
      "step": 58015
    },
    {
      "epoch": 2.945327173968222,
      "grad_norm": 0.42751434445381165,
      "learning_rate": 3.644855068785218e-07,
      "loss": 0.0291,
      "step": 58020
    },
    {
      "epoch": 2.9455809939590845,
      "grad_norm": 0.29339128732681274,
      "learning_rate": 3.6279337360610527e-07,
      "loss": 0.0324,
      "step": 58025
    },
    {
      "epoch": 2.9458348139499466,
      "grad_norm": 0.3385933041572571,
      "learning_rate": 3.6110124033368876e-07,
      "loss": 0.0229,
      "step": 58030
    },
    {
      "epoch": 2.946088633940809,
      "grad_norm": 0.37300586700439453,
      "learning_rate": 3.5940910706127214e-07,
      "loss": 0.0266,
      "step": 58035
    },
    {
      "epoch": 2.946342453931672,
      "grad_norm": 0.33182457089424133,
      "learning_rate": 3.5771697378885564e-07,
      "loss": 0.0213,
      "step": 58040
    },
    {
      "epoch": 2.946596273922534,
      "grad_norm": 0.4372585713863373,
      "learning_rate": 3.5602484051643913e-07,
      "loss": 0.0266,
      "step": 58045
    },
    {
      "epoch": 2.9468500939133966,
      "grad_norm": 0.3945082426071167,
      "learning_rate": 3.543327072440226e-07,
      "loss": 0.0316,
      "step": 58050
    },
    {
      "epoch": 2.947103913904259,
      "grad_norm": 0.3530140221118927,
      "learning_rate": 3.52640573971606e-07,
      "loss": 0.0248,
      "step": 58055
    },
    {
      "epoch": 2.947357733895122,
      "grad_norm": 0.2474001795053482,
      "learning_rate": 3.509484406991895e-07,
      "loss": 0.0307,
      "step": 58060
    },
    {
      "epoch": 2.947611553885984,
      "grad_norm": 0.37006187438964844,
      "learning_rate": 3.49256307426773e-07,
      "loss": 0.028,
      "step": 58065
    },
    {
      "epoch": 2.9478653738768466,
      "grad_norm": 0.2380700260400772,
      "learning_rate": 3.475641741543564e-07,
      "loss": 0.0254,
      "step": 58070
    },
    {
      "epoch": 2.948119193867709,
      "grad_norm": 0.2604024112224579,
      "learning_rate": 3.4587204088193986e-07,
      "loss": 0.0362,
      "step": 58075
    },
    {
      "epoch": 2.9483730138585713,
      "grad_norm": 0.32506248354911804,
      "learning_rate": 3.4417990760952336e-07,
      "loss": 0.0325,
      "step": 58080
    },
    {
      "epoch": 2.948626833849434,
      "grad_norm": 0.22234822809696198,
      "learning_rate": 3.4248777433710685e-07,
      "loss": 0.0306,
      "step": 58085
    },
    {
      "epoch": 2.9488806538402965,
      "grad_norm": 0.7750864028930664,
      "learning_rate": 3.407956410646903e-07,
      "loss": 0.0376,
      "step": 58090
    },
    {
      "epoch": 2.949134473831159,
      "grad_norm": 0.3774649202823639,
      "learning_rate": 3.391035077922737e-07,
      "loss": 0.0311,
      "step": 58095
    },
    {
      "epoch": 2.9493882938220213,
      "grad_norm": 0.358844518661499,
      "learning_rate": 3.374113745198572e-07,
      "loss": 0.0335,
      "step": 58100
    },
    {
      "epoch": 2.949642113812884,
      "grad_norm": 0.4665594696998596,
      "learning_rate": 3.3571924124744066e-07,
      "loss": 0.0314,
      "step": 58105
    },
    {
      "epoch": 2.9498959338037465,
      "grad_norm": 0.28644874691963196,
      "learning_rate": 3.3402710797502415e-07,
      "loss": 0.0279,
      "step": 58110
    },
    {
      "epoch": 2.9501497537946086,
      "grad_norm": 0.31177669763565063,
      "learning_rate": 3.323349747026076e-07,
      "loss": 0.0291,
      "step": 58115
    },
    {
      "epoch": 2.9504035737854712,
      "grad_norm": 0.26134926080703735,
      "learning_rate": 3.306428414301911e-07,
      "loss": 0.0305,
      "step": 58120
    },
    {
      "epoch": 2.950657393776334,
      "grad_norm": 0.3544405996799469,
      "learning_rate": 3.289507081577745e-07,
      "loss": 0.0317,
      "step": 58125
    },
    {
      "epoch": 2.9509112137671965,
      "grad_norm": 0.39834409952163696,
      "learning_rate": 3.27258574885358e-07,
      "loss": 0.0304,
      "step": 58130
    },
    {
      "epoch": 2.951165033758059,
      "grad_norm": 0.4057234823703766,
      "learning_rate": 3.2556644161294145e-07,
      "loss": 0.0343,
      "step": 58135
    },
    {
      "epoch": 2.951418853748921,
      "grad_norm": 0.24034041166305542,
      "learning_rate": 3.2387430834052494e-07,
      "loss": 0.0242,
      "step": 58140
    },
    {
      "epoch": 2.951672673739784,
      "grad_norm": 0.2940845191478729,
      "learning_rate": 3.221821750681084e-07,
      "loss": 0.0236,
      "step": 58145
    },
    {
      "epoch": 2.951926493730646,
      "grad_norm": 0.3044266998767853,
      "learning_rate": 3.2049004179569187e-07,
      "loss": 0.0257,
      "step": 58150
    },
    {
      "epoch": 2.9521803137215086,
      "grad_norm": 0.27591419219970703,
      "learning_rate": 3.187979085232753e-07,
      "loss": 0.0271,
      "step": 58155
    },
    {
      "epoch": 2.952434133712371,
      "grad_norm": 0.277483731508255,
      "learning_rate": 3.1710577525085875e-07,
      "loss": 0.0346,
      "step": 58160
    },
    {
      "epoch": 2.952687953703234,
      "grad_norm": 0.45967957377433777,
      "learning_rate": 3.1541364197844224e-07,
      "loss": 0.0296,
      "step": 58165
    },
    {
      "epoch": 2.9529417736940964,
      "grad_norm": 0.32595378160476685,
      "learning_rate": 3.1372150870602573e-07,
      "loss": 0.0318,
      "step": 58170
    },
    {
      "epoch": 2.9531955936849585,
      "grad_norm": 0.25108686089515686,
      "learning_rate": 3.1202937543360917e-07,
      "loss": 0.0279,
      "step": 58175
    },
    {
      "epoch": 2.953449413675821,
      "grad_norm": 0.3554666340351105,
      "learning_rate": 3.103372421611926e-07,
      "loss": 0.0232,
      "step": 58180
    },
    {
      "epoch": 2.9537032336666837,
      "grad_norm": 0.324294775724411,
      "learning_rate": 3.086451088887761e-07,
      "loss": 0.0262,
      "step": 58185
    },
    {
      "epoch": 2.953957053657546,
      "grad_norm": 0.27495795488357544,
      "learning_rate": 3.0695297561635954e-07,
      "loss": 0.0265,
      "step": 58190
    },
    {
      "epoch": 2.9542108736484085,
      "grad_norm": 0.3297526240348816,
      "learning_rate": 3.0526084234394303e-07,
      "loss": 0.0254,
      "step": 58195
    },
    {
      "epoch": 2.954464693639271,
      "grad_norm": 0.37720391154289246,
      "learning_rate": 3.0356870907152647e-07,
      "loss": 0.0295,
      "step": 58200
    },
    {
      "epoch": 2.9547185136301337,
      "grad_norm": 0.3667004108428955,
      "learning_rate": 3.0187657579910996e-07,
      "loss": 0.0364,
      "step": 58205
    },
    {
      "epoch": 2.954972333620996,
      "grad_norm": 0.6564368605613708,
      "learning_rate": 3.001844425266934e-07,
      "loss": 0.0348,
      "step": 58210
    },
    {
      "epoch": 2.9552261536118585,
      "grad_norm": 0.31273505091667175,
      "learning_rate": 2.984923092542769e-07,
      "loss": 0.0272,
      "step": 58215
    },
    {
      "epoch": 2.955479973602721,
      "grad_norm": 0.35449856519699097,
      "learning_rate": 2.9680017598186033e-07,
      "loss": 0.0266,
      "step": 58220
    },
    {
      "epoch": 2.9557337935935832,
      "grad_norm": 0.3688802719116211,
      "learning_rate": 2.951080427094438e-07,
      "loss": 0.0275,
      "step": 58225
    },
    {
      "epoch": 2.955987613584446,
      "grad_norm": 0.525647759437561,
      "learning_rate": 2.9341590943702726e-07,
      "loss": 0.0292,
      "step": 58230
    },
    {
      "epoch": 2.9562414335753084,
      "grad_norm": 0.3015739619731903,
      "learning_rate": 2.9172377616461075e-07,
      "loss": 0.0299,
      "step": 58235
    },
    {
      "epoch": 2.956495253566171,
      "grad_norm": 0.26125138998031616,
      "learning_rate": 2.900316428921942e-07,
      "loss": 0.0329,
      "step": 58240
    },
    {
      "epoch": 2.956749073557033,
      "grad_norm": 0.2997949421405792,
      "learning_rate": 2.883395096197777e-07,
      "loss": 0.0279,
      "step": 58245
    },
    {
      "epoch": 2.957002893547896,
      "grad_norm": 0.3520687222480774,
      "learning_rate": 2.8664737634736117e-07,
      "loss": 0.0301,
      "step": 58250
    },
    {
      "epoch": 2.9572567135387584,
      "grad_norm": 0.3813325762748718,
      "learning_rate": 2.849552430749446e-07,
      "loss": 0.0309,
      "step": 58255
    },
    {
      "epoch": 2.9575105335296206,
      "grad_norm": 0.3722776770591736,
      "learning_rate": 2.832631098025281e-07,
      "loss": 0.0282,
      "step": 58260
    },
    {
      "epoch": 2.957764353520483,
      "grad_norm": 0.3159470558166504,
      "learning_rate": 2.8157097653011154e-07,
      "loss": 0.0273,
      "step": 58265
    },
    {
      "epoch": 2.9580181735113458,
      "grad_norm": 0.21048100292682648,
      "learning_rate": 2.7987884325769503e-07,
      "loss": 0.0231,
      "step": 58270
    },
    {
      "epoch": 2.9582719935022084,
      "grad_norm": 0.2904769480228424,
      "learning_rate": 2.7818670998527847e-07,
      "loss": 0.0284,
      "step": 58275
    },
    {
      "epoch": 2.958525813493071,
      "grad_norm": 0.2318405956029892,
      "learning_rate": 2.7649457671286196e-07,
      "loss": 0.0293,
      "step": 58280
    },
    {
      "epoch": 2.958779633483933,
      "grad_norm": 0.323593407869339,
      "learning_rate": 2.748024434404454e-07,
      "loss": 0.0225,
      "step": 58285
    },
    {
      "epoch": 2.9590334534747957,
      "grad_norm": 0.31280264258384705,
      "learning_rate": 2.731103101680289e-07,
      "loss": 0.0293,
      "step": 58290
    },
    {
      "epoch": 2.959287273465658,
      "grad_norm": 0.3686768412590027,
      "learning_rate": 2.7141817689561233e-07,
      "loss": 0.0303,
      "step": 58295
    },
    {
      "epoch": 2.9595410934565205,
      "grad_norm": 0.31249767541885376,
      "learning_rate": 2.697260436231958e-07,
      "loss": 0.0353,
      "step": 58300
    },
    {
      "epoch": 2.959794913447383,
      "grad_norm": 0.33410879969596863,
      "learning_rate": 2.6803391035077926e-07,
      "loss": 0.0337,
      "step": 58305
    },
    {
      "epoch": 2.9600487334382457,
      "grad_norm": 0.4810965359210968,
      "learning_rate": 2.6634177707836275e-07,
      "loss": 0.0336,
      "step": 58310
    },
    {
      "epoch": 2.9603025534291083,
      "grad_norm": 0.24375031888484955,
      "learning_rate": 2.646496438059462e-07,
      "loss": 0.0263,
      "step": 58315
    },
    {
      "epoch": 2.9605563734199705,
      "grad_norm": 0.314066082239151,
      "learning_rate": 2.629575105335297e-07,
      "loss": 0.0351,
      "step": 58320
    },
    {
      "epoch": 2.960810193410833,
      "grad_norm": 0.6520958542823792,
      "learning_rate": 2.612653772611131e-07,
      "loss": 0.0339,
      "step": 58325
    },
    {
      "epoch": 2.9610640134016957,
      "grad_norm": 0.4829593300819397,
      "learning_rate": 2.5957324398869656e-07,
      "loss": 0.0292,
      "step": 58330
    },
    {
      "epoch": 2.961317833392558,
      "grad_norm": 0.39782652258872986,
      "learning_rate": 2.5788111071628005e-07,
      "loss": 0.0302,
      "step": 58335
    },
    {
      "epoch": 2.9615716533834204,
      "grad_norm": 0.3647368550300598,
      "learning_rate": 2.561889774438635e-07,
      "loss": 0.0288,
      "step": 58340
    },
    {
      "epoch": 2.961825473374283,
      "grad_norm": 0.3723721504211426,
      "learning_rate": 2.54496844171447e-07,
      "loss": 0.0271,
      "step": 58345
    },
    {
      "epoch": 2.9620792933651456,
      "grad_norm": 0.5521222352981567,
      "learning_rate": 2.528047108990304e-07,
      "loss": 0.0264,
      "step": 58350
    },
    {
      "epoch": 2.962333113356008,
      "grad_norm": 0.2993130385875702,
      "learning_rate": 2.511125776266139e-07,
      "loss": 0.0284,
      "step": 58355
    },
    {
      "epoch": 2.9625869333468704,
      "grad_norm": 0.827579915523529,
      "learning_rate": 2.4942044435419735e-07,
      "loss": 0.0343,
      "step": 58360
    },
    {
      "epoch": 2.962840753337733,
      "grad_norm": 0.3333107829093933,
      "learning_rate": 2.4772831108178084e-07,
      "loss": 0.0364,
      "step": 58365
    },
    {
      "epoch": 2.963094573328595,
      "grad_norm": 0.3528420925140381,
      "learning_rate": 2.460361778093643e-07,
      "loss": 0.0308,
      "step": 58370
    },
    {
      "epoch": 2.9633483933194578,
      "grad_norm": 0.2768530249595642,
      "learning_rate": 2.443440445369477e-07,
      "loss": 0.0272,
      "step": 58375
    },
    {
      "epoch": 2.9636022133103204,
      "grad_norm": 0.2978229820728302,
      "learning_rate": 2.426519112645312e-07,
      "loss": 0.0282,
      "step": 58380
    },
    {
      "epoch": 2.963856033301183,
      "grad_norm": 0.24835215508937836,
      "learning_rate": 2.4095977799211465e-07,
      "loss": 0.026,
      "step": 58385
    },
    {
      "epoch": 2.964109853292045,
      "grad_norm": 0.2698346674442291,
      "learning_rate": 2.3926764471969814e-07,
      "loss": 0.0254,
      "step": 58390
    },
    {
      "epoch": 2.9643636732829077,
      "grad_norm": 0.28580522537231445,
      "learning_rate": 2.375755114472816e-07,
      "loss": 0.0268,
      "step": 58395
    },
    {
      "epoch": 2.9646174932737703,
      "grad_norm": 0.31052786111831665,
      "learning_rate": 2.3588337817486507e-07,
      "loss": 0.0238,
      "step": 58400
    },
    {
      "epoch": 2.9648713132646325,
      "grad_norm": 0.35599517822265625,
      "learning_rate": 2.3419124490244854e-07,
      "loss": 0.032,
      "step": 58405
    },
    {
      "epoch": 2.965125133255495,
      "grad_norm": 0.3290466070175171,
      "learning_rate": 2.32499111630032e-07,
      "loss": 0.0326,
      "step": 58410
    },
    {
      "epoch": 2.9653789532463577,
      "grad_norm": 0.46071192622184753,
      "learning_rate": 2.3080697835761547e-07,
      "loss": 0.0305,
      "step": 58415
    },
    {
      "epoch": 2.9656327732372203,
      "grad_norm": 0.20090284943580627,
      "learning_rate": 2.2911484508519893e-07,
      "loss": 0.0259,
      "step": 58420
    },
    {
      "epoch": 2.965886593228083,
      "grad_norm": 0.47280457615852356,
      "learning_rate": 2.274227118127824e-07,
      "loss": 0.0251,
      "step": 58425
    },
    {
      "epoch": 2.966140413218945,
      "grad_norm": 0.3381558656692505,
      "learning_rate": 2.2573057854036586e-07,
      "loss": 0.0239,
      "step": 58430
    },
    {
      "epoch": 2.9663942332098077,
      "grad_norm": 0.27177587151527405,
      "learning_rate": 2.2403844526794933e-07,
      "loss": 0.0262,
      "step": 58435
    },
    {
      "epoch": 2.96664805320067,
      "grad_norm": 0.2976479232311249,
      "learning_rate": 2.223463119955328e-07,
      "loss": 0.0264,
      "step": 58440
    },
    {
      "epoch": 2.9669018731915324,
      "grad_norm": 0.3702794909477234,
      "learning_rate": 2.2065417872311623e-07,
      "loss": 0.0255,
      "step": 58445
    },
    {
      "epoch": 2.967155693182395,
      "grad_norm": 0.31776705384254456,
      "learning_rate": 2.1896204545069972e-07,
      "loss": 0.0272,
      "step": 58450
    },
    {
      "epoch": 2.9674095131732576,
      "grad_norm": 0.5179955959320068,
      "learning_rate": 2.1726991217828316e-07,
      "loss": 0.025,
      "step": 58455
    },
    {
      "epoch": 2.9676633331641202,
      "grad_norm": 0.2706942558288574,
      "learning_rate": 2.1557777890586665e-07,
      "loss": 0.0315,
      "step": 58460
    },
    {
      "epoch": 2.9679171531549824,
      "grad_norm": 0.3058590590953827,
      "learning_rate": 2.138856456334501e-07,
      "loss": 0.03,
      "step": 58465
    },
    {
      "epoch": 2.968170973145845,
      "grad_norm": 0.33392003178596497,
      "learning_rate": 2.1219351236103358e-07,
      "loss": 0.0252,
      "step": 58470
    },
    {
      "epoch": 2.9684247931367076,
      "grad_norm": 1.4577997922897339,
      "learning_rate": 2.1050137908861702e-07,
      "loss": 0.0249,
      "step": 58475
    },
    {
      "epoch": 2.9686786131275698,
      "grad_norm": 0.37016749382019043,
      "learning_rate": 2.0880924581620051e-07,
      "loss": 0.0295,
      "step": 58480
    },
    {
      "epoch": 2.9689324331184324,
      "grad_norm": 0.31658273935317993,
      "learning_rate": 2.0711711254378395e-07,
      "loss": 0.031,
      "step": 58485
    },
    {
      "epoch": 2.969186253109295,
      "grad_norm": 0.5395975112915039,
      "learning_rate": 2.0542497927136744e-07,
      "loss": 0.0305,
      "step": 58490
    },
    {
      "epoch": 2.9694400731001576,
      "grad_norm": 0.5022832155227661,
      "learning_rate": 2.0373284599895088e-07,
      "loss": 0.0328,
      "step": 58495
    },
    {
      "epoch": 2.9696938930910197,
      "grad_norm": 0.46155476570129395,
      "learning_rate": 2.0204071272653437e-07,
      "loss": 0.031,
      "step": 58500
    },
    {
      "epoch": 2.9699477130818823,
      "grad_norm": 0.26691576838493347,
      "learning_rate": 2.003485794541178e-07,
      "loss": 0.0263,
      "step": 58505
    },
    {
      "epoch": 2.970201533072745,
      "grad_norm": 0.2853245437145233,
      "learning_rate": 1.986564461817013e-07,
      "loss": 0.0292,
      "step": 58510
    },
    {
      "epoch": 2.970455353063607,
      "grad_norm": 0.3737592101097107,
      "learning_rate": 1.9696431290928474e-07,
      "loss": 0.0229,
      "step": 58515
    },
    {
      "epoch": 2.9707091730544697,
      "grad_norm": 0.30825868248939514,
      "learning_rate": 1.9527217963686823e-07,
      "loss": 0.0284,
      "step": 58520
    },
    {
      "epoch": 2.9709629930453323,
      "grad_norm": 0.2496730536222458,
      "learning_rate": 1.9358004636445167e-07,
      "loss": 0.0246,
      "step": 58525
    },
    {
      "epoch": 2.971216813036195,
      "grad_norm": 0.3173964023590088,
      "learning_rate": 1.9188791309203517e-07,
      "loss": 0.0276,
      "step": 58530
    },
    {
      "epoch": 2.971470633027057,
      "grad_norm": 0.44984614849090576,
      "learning_rate": 1.901957798196186e-07,
      "loss": 0.0288,
      "step": 58535
    },
    {
      "epoch": 2.9717244530179197,
      "grad_norm": 0.3062257468700409,
      "learning_rate": 1.8850364654720207e-07,
      "loss": 0.021,
      "step": 58540
    },
    {
      "epoch": 2.9719782730087823,
      "grad_norm": 0.36753204464912415,
      "learning_rate": 1.8681151327478553e-07,
      "loss": 0.0302,
      "step": 58545
    },
    {
      "epoch": 2.9722320929996444,
      "grad_norm": 0.2601029574871063,
      "learning_rate": 1.85119380002369e-07,
      "loss": 0.0235,
      "step": 58550
    },
    {
      "epoch": 2.972485912990507,
      "grad_norm": 0.2703913152217865,
      "learning_rate": 1.8342724672995246e-07,
      "loss": 0.0238,
      "step": 58555
    },
    {
      "epoch": 2.9727397329813696,
      "grad_norm": 0.3573007881641388,
      "learning_rate": 1.8173511345753593e-07,
      "loss": 0.0346,
      "step": 58560
    },
    {
      "epoch": 2.9729935529722322,
      "grad_norm": 0.35567259788513184,
      "learning_rate": 1.800429801851194e-07,
      "loss": 0.0295,
      "step": 58565
    },
    {
      "epoch": 2.973247372963095,
      "grad_norm": 0.3184675872325897,
      "learning_rate": 1.7835084691270286e-07,
      "loss": 0.0256,
      "step": 58570
    },
    {
      "epoch": 2.973501192953957,
      "grad_norm": 0.27316588163375854,
      "learning_rate": 1.7665871364028632e-07,
      "loss": 0.0245,
      "step": 58575
    },
    {
      "epoch": 2.9737550129448196,
      "grad_norm": 0.27470746636390686,
      "learning_rate": 1.749665803678698e-07,
      "loss": 0.0225,
      "step": 58580
    },
    {
      "epoch": 2.9740088329356817,
      "grad_norm": 0.23476144671440125,
      "learning_rate": 1.7327444709545323e-07,
      "loss": 0.0242,
      "step": 58585
    },
    {
      "epoch": 2.9742626529265443,
      "grad_norm": 0.38638877868652344,
      "learning_rate": 1.7158231382303672e-07,
      "loss": 0.0277,
      "step": 58590
    },
    {
      "epoch": 2.974516472917407,
      "grad_norm": 0.4438152611255646,
      "learning_rate": 1.6989018055062016e-07,
      "loss": 0.0358,
      "step": 58595
    },
    {
      "epoch": 2.9747702929082696,
      "grad_norm": 0.26275134086608887,
      "learning_rate": 1.6819804727820365e-07,
      "loss": 0.0334,
      "step": 58600
    },
    {
      "epoch": 2.975024112899132,
      "grad_norm": 0.5719013214111328,
      "learning_rate": 1.665059140057871e-07,
      "loss": 0.0332,
      "step": 58605
    },
    {
      "epoch": 2.9752779328899943,
      "grad_norm": 0.3588959574699402,
      "learning_rate": 1.6481378073337058e-07,
      "loss": 0.0316,
      "step": 58610
    },
    {
      "epoch": 2.975531752880857,
      "grad_norm": 0.34206733107566833,
      "learning_rate": 1.6312164746095402e-07,
      "loss": 0.0337,
      "step": 58615
    },
    {
      "epoch": 2.9757855728717195,
      "grad_norm": 0.3827134966850281,
      "learning_rate": 1.614295141885375e-07,
      "loss": 0.0313,
      "step": 58620
    },
    {
      "epoch": 2.9760393928625817,
      "grad_norm": 0.32642507553100586,
      "learning_rate": 1.5973738091612095e-07,
      "loss": 0.0275,
      "step": 58625
    },
    {
      "epoch": 2.9762932128534443,
      "grad_norm": 0.3430155813694,
      "learning_rate": 1.5804524764370444e-07,
      "loss": 0.0273,
      "step": 58630
    },
    {
      "epoch": 2.976547032844307,
      "grad_norm": 0.37102141976356506,
      "learning_rate": 1.5635311437128788e-07,
      "loss": 0.0285,
      "step": 58635
    },
    {
      "epoch": 2.9768008528351695,
      "grad_norm": 0.2242543250322342,
      "learning_rate": 1.5466098109887137e-07,
      "loss": 0.0283,
      "step": 58640
    },
    {
      "epoch": 2.9770546728260316,
      "grad_norm": 0.3857623040676117,
      "learning_rate": 1.5296884782645484e-07,
      "loss": 0.0274,
      "step": 58645
    },
    {
      "epoch": 2.9773084928168942,
      "grad_norm": 0.18669942021369934,
      "learning_rate": 1.512767145540383e-07,
      "loss": 0.0263,
      "step": 58650
    },
    {
      "epoch": 2.977562312807757,
      "grad_norm": 0.40870609879493713,
      "learning_rate": 1.4958458128162177e-07,
      "loss": 0.0365,
      "step": 58655
    },
    {
      "epoch": 2.977816132798619,
      "grad_norm": 0.28093990683555603,
      "learning_rate": 1.4789244800920523e-07,
      "loss": 0.0243,
      "step": 58660
    },
    {
      "epoch": 2.9780699527894816,
      "grad_norm": 0.34547853469848633,
      "learning_rate": 1.462003147367887e-07,
      "loss": 0.0299,
      "step": 58665
    },
    {
      "epoch": 2.978323772780344,
      "grad_norm": 0.23397915065288544,
      "learning_rate": 1.4450818146437216e-07,
      "loss": 0.038,
      "step": 58670
    },
    {
      "epoch": 2.978577592771207,
      "grad_norm": 0.2273659110069275,
      "learning_rate": 1.428160481919556e-07,
      "loss": 0.0255,
      "step": 58675
    },
    {
      "epoch": 2.978831412762069,
      "grad_norm": 0.2079605758190155,
      "learning_rate": 1.4112391491953907e-07,
      "loss": 0.0347,
      "step": 58680
    },
    {
      "epoch": 2.9790852327529316,
      "grad_norm": 0.31091585755348206,
      "learning_rate": 1.3943178164712253e-07,
      "loss": 0.0299,
      "step": 58685
    },
    {
      "epoch": 2.979339052743794,
      "grad_norm": 0.28483375906944275,
      "learning_rate": 1.37739648374706e-07,
      "loss": 0.0327,
      "step": 58690
    },
    {
      "epoch": 2.9795928727346563,
      "grad_norm": 0.3354802429676056,
      "learning_rate": 1.3604751510228946e-07,
      "loss": 0.0228,
      "step": 58695
    },
    {
      "epoch": 2.979846692725519,
      "grad_norm": 0.18329808115959167,
      "learning_rate": 1.3435538182987293e-07,
      "loss": 0.024,
      "step": 58700
    },
    {
      "epoch": 2.9801005127163815,
      "grad_norm": 0.3315074145793915,
      "learning_rate": 1.326632485574564e-07,
      "loss": 0.0259,
      "step": 58705
    },
    {
      "epoch": 2.980354332707244,
      "grad_norm": 0.32475075125694275,
      "learning_rate": 1.3097111528503986e-07,
      "loss": 0.0256,
      "step": 58710
    },
    {
      "epoch": 2.9806081526981067,
      "grad_norm": 0.33447596430778503,
      "learning_rate": 1.2927898201262332e-07,
      "loss": 0.0247,
      "step": 58715
    },
    {
      "epoch": 2.980861972688969,
      "grad_norm": 0.32187217473983765,
      "learning_rate": 1.275868487402068e-07,
      "loss": 0.0233,
      "step": 58720
    },
    {
      "epoch": 2.9811157926798315,
      "grad_norm": 0.34107112884521484,
      "learning_rate": 1.2589471546779025e-07,
      "loss": 0.0335,
      "step": 58725
    },
    {
      "epoch": 2.9813696126706937,
      "grad_norm": 0.4276149272918701,
      "learning_rate": 1.2420258219537372e-07,
      "loss": 0.0261,
      "step": 58730
    },
    {
      "epoch": 2.9816234326615563,
      "grad_norm": 0.3580925166606903,
      "learning_rate": 1.2251044892295718e-07,
      "loss": 0.0309,
      "step": 58735
    },
    {
      "epoch": 2.981877252652419,
      "grad_norm": 0.2032964825630188,
      "learning_rate": 1.2081831565054065e-07,
      "loss": 0.0283,
      "step": 58740
    },
    {
      "epoch": 2.9821310726432815,
      "grad_norm": 0.27535682916641235,
      "learning_rate": 1.1912618237812411e-07,
      "loss": 0.0278,
      "step": 58745
    },
    {
      "epoch": 2.982384892634144,
      "grad_norm": 0.7258585691452026,
      "learning_rate": 1.1743404910570758e-07,
      "loss": 0.0306,
      "step": 58750
    },
    {
      "epoch": 2.9826387126250062,
      "grad_norm": 0.2572552561759949,
      "learning_rate": 1.1574191583329104e-07,
      "loss": 0.0224,
      "step": 58755
    },
    {
      "epoch": 2.982892532615869,
      "grad_norm": 0.42224279046058655,
      "learning_rate": 1.1404978256087451e-07,
      "loss": 0.0292,
      "step": 58760
    },
    {
      "epoch": 2.9831463526067314,
      "grad_norm": 0.33688056468963623,
      "learning_rate": 1.1235764928845797e-07,
      "loss": 0.0328,
      "step": 58765
    },
    {
      "epoch": 2.9834001725975936,
      "grad_norm": 0.25573021173477173,
      "learning_rate": 1.1066551601604144e-07,
      "loss": 0.0264,
      "step": 58770
    },
    {
      "epoch": 2.983653992588456,
      "grad_norm": 0.2507616877555847,
      "learning_rate": 1.0897338274362489e-07,
      "loss": 0.0344,
      "step": 58775
    },
    {
      "epoch": 2.983907812579319,
      "grad_norm": 0.3209986388683319,
      "learning_rate": 1.0728124947120836e-07,
      "loss": 0.0228,
      "step": 58780
    },
    {
      "epoch": 2.9841616325701814,
      "grad_norm": 0.6421720385551453,
      "learning_rate": 1.0558911619879182e-07,
      "loss": 0.022,
      "step": 58785
    },
    {
      "epoch": 2.9844154525610436,
      "grad_norm": 0.3483797311782837,
      "learning_rate": 1.0389698292637529e-07,
      "loss": 0.0292,
      "step": 58790
    },
    {
      "epoch": 2.984669272551906,
      "grad_norm": 0.2420889437198639,
      "learning_rate": 1.0220484965395875e-07,
      "loss": 0.035,
      "step": 58795
    },
    {
      "epoch": 2.9849230925427688,
      "grad_norm": 0.4202343821525574,
      "learning_rate": 1.0051271638154222e-07,
      "loss": 0.033,
      "step": 58800
    },
    {
      "epoch": 2.985176912533631,
      "grad_norm": 0.247958704829216,
      "learning_rate": 9.882058310912568e-08,
      "loss": 0.0231,
      "step": 58805
    },
    {
      "epoch": 2.9854307325244935,
      "grad_norm": 0.3831901252269745,
      "learning_rate": 9.712844983670915e-08,
      "loss": 0.0388,
      "step": 58810
    },
    {
      "epoch": 2.985684552515356,
      "grad_norm": 0.34248149394989014,
      "learning_rate": 9.543631656429261e-08,
      "loss": 0.028,
      "step": 58815
    },
    {
      "epoch": 2.9859383725062187,
      "grad_norm": 0.2834693491458893,
      "learning_rate": 9.374418329187608e-08,
      "loss": 0.0308,
      "step": 58820
    },
    {
      "epoch": 2.9861921924970813,
      "grad_norm": 0.3265041410923004,
      "learning_rate": 9.205205001945954e-08,
      "loss": 0.0333,
      "step": 58825
    },
    {
      "epoch": 2.9864460124879435,
      "grad_norm": 0.34004950523376465,
      "learning_rate": 9.035991674704301e-08,
      "loss": 0.0277,
      "step": 58830
    },
    {
      "epoch": 2.986699832478806,
      "grad_norm": 0.32642513513565063,
      "learning_rate": 8.866778347462647e-08,
      "loss": 0.0252,
      "step": 58835
    },
    {
      "epoch": 2.9869536524696683,
      "grad_norm": 0.30227944254875183,
      "learning_rate": 8.697565020220994e-08,
      "loss": 0.0248,
      "step": 58840
    },
    {
      "epoch": 2.987207472460531,
      "grad_norm": 0.8182564377784729,
      "learning_rate": 8.528351692979339e-08,
      "loss": 0.029,
      "step": 58845
    },
    {
      "epoch": 2.9874612924513935,
      "grad_norm": 0.32346653938293457,
      "learning_rate": 8.359138365737685e-08,
      "loss": 0.0269,
      "step": 58850
    },
    {
      "epoch": 2.987715112442256,
      "grad_norm": 0.2588391900062561,
      "learning_rate": 8.189925038496032e-08,
      "loss": 0.0238,
      "step": 58855
    },
    {
      "epoch": 2.9879689324331187,
      "grad_norm": 0.44902580976486206,
      "learning_rate": 8.020711711254378e-08,
      "loss": 0.0275,
      "step": 58860
    },
    {
      "epoch": 2.988222752423981,
      "grad_norm": 0.3256726861000061,
      "learning_rate": 7.851498384012725e-08,
      "loss": 0.0243,
      "step": 58865
    },
    {
      "epoch": 2.9884765724148434,
      "grad_norm": 0.2934799790382385,
      "learning_rate": 7.682285056771071e-08,
      "loss": 0.0308,
      "step": 58870
    },
    {
      "epoch": 2.988730392405706,
      "grad_norm": 0.8592537045478821,
      "learning_rate": 7.513071729529418e-08,
      "loss": 0.0336,
      "step": 58875
    },
    {
      "epoch": 2.988984212396568,
      "grad_norm": 0.25373250246047974,
      "learning_rate": 7.343858402287764e-08,
      "loss": 0.029,
      "step": 58880
    },
    {
      "epoch": 2.989238032387431,
      "grad_norm": 0.2816363573074341,
      "learning_rate": 7.174645075046111e-08,
      "loss": 0.0299,
      "step": 58885
    },
    {
      "epoch": 2.9894918523782934,
      "grad_norm": 0.26894432306289673,
      "learning_rate": 7.005431747804458e-08,
      "loss": 0.0323,
      "step": 58890
    },
    {
      "epoch": 2.989745672369156,
      "grad_norm": 0.2629813551902771,
      "learning_rate": 6.836218420562804e-08,
      "loss": 0.0294,
      "step": 58895
    },
    {
      "epoch": 2.989999492360018,
      "grad_norm": 0.24025852978229523,
      "learning_rate": 6.66700509332115e-08,
      "loss": 0.0305,
      "step": 58900
    },
    {
      "epoch": 2.9902533123508808,
      "grad_norm": 0.26809507608413696,
      "learning_rate": 6.497791766079497e-08,
      "loss": 0.0284,
      "step": 58905
    },
    {
      "epoch": 2.9905071323417434,
      "grad_norm": 0.32450565695762634,
      "learning_rate": 6.328578438837844e-08,
      "loss": 0.0265,
      "step": 58910
    },
    {
      "epoch": 2.9907609523326055,
      "grad_norm": 0.3137778043746948,
      "learning_rate": 6.15936511159619e-08,
      "loss": 0.0296,
      "step": 58915
    },
    {
      "epoch": 2.991014772323468,
      "grad_norm": 0.472679078578949,
      "learning_rate": 5.990151784354537e-08,
      "loss": 0.0265,
      "step": 58920
    },
    {
      "epoch": 2.9912685923143307,
      "grad_norm": 0.2648654580116272,
      "learning_rate": 5.820938457112883e-08,
      "loss": 0.0259,
      "step": 58925
    },
    {
      "epoch": 2.9915224123051933,
      "grad_norm": 0.25962552428245544,
      "learning_rate": 5.6517251298712296e-08,
      "loss": 0.0246,
      "step": 58930
    },
    {
      "epoch": 2.9917762322960555,
      "grad_norm": 0.32846131920814514,
      "learning_rate": 5.482511802629576e-08,
      "loss": 0.0249,
      "step": 58935
    },
    {
      "epoch": 2.992030052286918,
      "grad_norm": 0.3609420657157898,
      "learning_rate": 5.313298475387922e-08,
      "loss": 0.026,
      "step": 58940
    },
    {
      "epoch": 2.9922838722777807,
      "grad_norm": 0.33980563282966614,
      "learning_rate": 5.1440851481462685e-08,
      "loss": 0.0307,
      "step": 58945
    },
    {
      "epoch": 2.992537692268643,
      "grad_norm": 0.2794458270072937,
      "learning_rate": 4.974871820904615e-08,
      "loss": 0.0314,
      "step": 58950
    },
    {
      "epoch": 2.9927915122595055,
      "grad_norm": 0.536361038684845,
      "learning_rate": 4.8056584936629615e-08,
      "loss": 0.0329,
      "step": 58955
    },
    {
      "epoch": 2.993045332250368,
      "grad_norm": 0.5886624455451965,
      "learning_rate": 4.636445166421308e-08,
      "loss": 0.0385,
      "step": 58960
    },
    {
      "epoch": 2.9932991522412307,
      "grad_norm": 0.3537057638168335,
      "learning_rate": 4.4672318391796546e-08,
      "loss": 0.0313,
      "step": 58965
    },
    {
      "epoch": 2.9935529722320933,
      "grad_norm": 0.2326638102531433,
      "learning_rate": 4.298018511938001e-08,
      "loss": 0.024,
      "step": 58970
    },
    {
      "epoch": 2.9938067922229554,
      "grad_norm": 0.47342896461486816,
      "learning_rate": 4.128805184696347e-08,
      "loss": 0.0358,
      "step": 58975
    },
    {
      "epoch": 2.994060612213818,
      "grad_norm": 0.27559787034988403,
      "learning_rate": 3.9595918574546934e-08,
      "loss": 0.0317,
      "step": 58980
    },
    {
      "epoch": 2.99431443220468,
      "grad_norm": 0.34952622652053833,
      "learning_rate": 3.79037853021304e-08,
      "loss": 0.0322,
      "step": 58985
    },
    {
      "epoch": 2.994568252195543,
      "grad_norm": 0.5940717458724976,
      "learning_rate": 3.6211652029713865e-08,
      "loss": 0.0301,
      "step": 58990
    },
    {
      "epoch": 2.9948220721864054,
      "grad_norm": 0.31109818816185,
      "learning_rate": 3.451951875729733e-08,
      "loss": 0.0338,
      "step": 58995
    },
    {
      "epoch": 2.995075892177268,
      "grad_norm": 0.37035757303237915,
      "learning_rate": 3.2827385484880795e-08,
      "loss": 0.03,
      "step": 59000
    },
    {
      "epoch": 2.9953297121681306,
      "grad_norm": 0.4075457751750946,
      "learning_rate": 3.113525221246426e-08,
      "loss": 0.032,
      "step": 59005
    },
    {
      "epoch": 2.9955835321589928,
      "grad_norm": 0.2673121392726898,
      "learning_rate": 2.9443118940047722e-08,
      "loss": 0.0286,
      "step": 59010
    },
    {
      "epoch": 2.9958373521498554,
      "grad_norm": 0.2799220383167267,
      "learning_rate": 2.7750985667631187e-08,
      "loss": 0.0297,
      "step": 59015
    },
    {
      "epoch": 2.996091172140718,
      "grad_norm": 0.232395201921463,
      "learning_rate": 2.605885239521465e-08,
      "loss": 0.0282,
      "step": 59020
    },
    {
      "epoch": 2.99634499213158,
      "grad_norm": 0.34497490525245667,
      "learning_rate": 2.4366719122798114e-08,
      "loss": 0.0311,
      "step": 59025
    },
    {
      "epoch": 2.9965988121224427,
      "grad_norm": 0.39901643991470337,
      "learning_rate": 2.267458585038158e-08,
      "loss": 0.0312,
      "step": 59030
    },
    {
      "epoch": 2.9968526321133053,
      "grad_norm": 0.31049707531929016,
      "learning_rate": 2.098245257796504e-08,
      "loss": 0.0291,
      "step": 59035
    },
    {
      "epoch": 2.997106452104168,
      "grad_norm": 0.374683141708374,
      "learning_rate": 1.9290319305548506e-08,
      "loss": 0.0324,
      "step": 59040
    },
    {
      "epoch": 2.99736027209503,
      "grad_norm": 0.3533037602901459,
      "learning_rate": 1.759818603313197e-08,
      "loss": 0.0333,
      "step": 59045
    },
    {
      "epoch": 2.9976140920858927,
      "grad_norm": 0.35654324293136597,
      "learning_rate": 1.5906052760715436e-08,
      "loss": 0.0313,
      "step": 59050
    },
    {
      "epoch": 2.9978679120767553,
      "grad_norm": 0.25922712683677673,
      "learning_rate": 1.42139194882989e-08,
      "loss": 0.0312,
      "step": 59055
    },
    {
      "epoch": 2.9981217320676175,
      "grad_norm": 0.28118982911109924,
      "learning_rate": 1.2521786215882363e-08,
      "loss": 0.0212,
      "step": 59060
    },
    {
      "epoch": 2.99837555205848,
      "grad_norm": 0.23331542313098907,
      "learning_rate": 1.0829652943465827e-08,
      "loss": 0.0289,
      "step": 59065
    },
    {
      "epoch": 2.9986293720493427,
      "grad_norm": 0.3286423683166504,
      "learning_rate": 9.137519671049294e-09,
      "loss": 0.0252,
      "step": 59070
    },
    {
      "epoch": 2.9988831920402053,
      "grad_norm": 0.38054725527763367,
      "learning_rate": 7.445386398632757e-09,
      "loss": 0.0272,
      "step": 59075
    },
    {
      "epoch": 2.9991370120310674,
      "grad_norm": 0.23584502935409546,
      "learning_rate": 5.753253126216221e-09,
      "loss": 0.0302,
      "step": 59080
    },
    {
      "epoch": 2.99939083202193,
      "grad_norm": 0.40845179557800293,
      "learning_rate": 4.061119853799686e-09,
      "loss": 0.0338,
      "step": 59085
    },
    {
      "epoch": 2.9996446520127926,
      "grad_norm": 0.2870352864265442,
      "learning_rate": 2.36898658138315e-09,
      "loss": 0.0369,
      "step": 59090
    },
    {
      "epoch": 2.999898472003655,
      "grad_norm": 0.2551342248916626,
      "learning_rate": 6.768533089666142e-10,
      "loss": 0.0264,
      "step": 59095
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.197622150182724,
      "eval_runtime": 1775.8365,
      "eval_samples_per_second": 70.389,
      "eval_steps_per_second": 2.2,
      "step": 59097
    }
  ],
  "logging_steps": 5,
  "max_steps": 59097,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1337388851021742e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}