diff --git "a/DLI-Lab/Meta-Llama-3.1-8B-Instruct-WM-acctree/trainer_state.json" "b/DLI-Lab/Meta-Llama-3.1-8B-Instruct-WM-acctree/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/DLI-Lab/Meta-Llama-3.1-8B-Instruct-WM-acctree/trainer_state.json"
@@ -0,0 +1,5033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9857954545454546,
+  "eval_steps": 88,
+  "global_step": 704,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002840909090909091,
+      "grad_norm": 2.8675832748413086,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.5046,
+      "step": 1
+    },
+    {
+      "epoch": 0.002840909090909091,
+      "eval_loss": 0.3339672386646271,
+      "eval_runtime": 320.8391,
+      "eval_samples_per_second": 8.4,
+      "eval_steps_per_second": 1.05,
+      "step": 1
+    },
+    {
+      "epoch": 0.005681818181818182,
+      "grad_norm": 3.076528310775757,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5636,
+      "step": 2
+    },
+    {
+      "epoch": 0.008522727272727272,
+      "grad_norm": 2.3580377101898193,
+      "learning_rate": 6e-06,
+      "loss": 0.2059,
+      "step": 3
+    },
+    {
+      "epoch": 0.011363636363636364,
+      "grad_norm": 1.2384682893753052,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.266,
+      "step": 4
+    },
+    {
+      "epoch": 0.014204545454545454,
+      "grad_norm": 0.8427222371101379,
+      "learning_rate": 1e-05,
+      "loss": 0.1559,
+      "step": 5
+    },
+    {
+      "epoch": 0.017045454545454544,
+      "grad_norm": 1.0242825746536255,
+      "learning_rate": 1.2e-05,
+      "loss": 0.3084,
+      "step": 6
+    },
+    {
+      "epoch": 0.019886363636363636,
+      "grad_norm": 0.6649225354194641,
+      "learning_rate": 1.4e-05,
+      "loss": 0.1229,
+      "step": 7
+    },
+    {
+      "epoch": 0.022727272727272728,
+      "grad_norm": 3.461799383163452,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.4072,
+      "step": 8
+    },
+    {
+      "epoch": 0.02556818181818182,
+      "grad_norm": 1.2614651918411255,
+      "learning_rate": 1.8e-05,
+      "loss": 0.1717,
+      "step": 9
+    },
+    {
+      "epoch": 0.028409090909090908,
+      "grad_norm": 1.217345118522644,
+      "learning_rate": 2e-05,
+      "loss": 0.2597,
+      "step": 10
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 0.8754523992538452,
+      "learning_rate": 1.9999897541010772e-05,
+      "loss": 0.1865,
+      "step": 11
+    },
+    {
+      "epoch": 0.03409090909090909,
+      "grad_norm": 0.9129774570465088,
+      "learning_rate": 1.9999590166142656e-05,
+      "loss": 0.1859,
+      "step": 12
+    },
+    {
+      "epoch": 0.036931818181818184,
+      "grad_norm": 1.7369693517684937,
+      "learning_rate": 1.999907788169431e-05,
+      "loss": 0.188,
+      "step": 13
+    },
+    {
+      "epoch": 0.03977272727272727,
+      "grad_norm": 1.3096760511398315,
+      "learning_rate": 1.9998360698163375e-05,
+      "loss": 0.3204,
+      "step": 14
+    },
+    {
+      "epoch": 0.04261363636363636,
+      "grad_norm": 14.911248207092285,
+      "learning_rate": 1.999743863024622e-05,
+      "loss": 0.232,
+      "step": 15
+    },
+    {
+      "epoch": 0.045454545454545456,
+      "grad_norm": 1.7296037673950195,
+      "learning_rate": 1.999631169683768e-05,
+      "loss": 0.2511,
+      "step": 16
+    },
+    {
+      "epoch": 0.048295454545454544,
+      "grad_norm": 1.1196941137313843,
+      "learning_rate": 1.999497992103064e-05,
+      "loss": 0.2549,
+      "step": 17
+    },
+    {
+      "epoch": 0.05113636363636364,
+      "grad_norm": 0.7868685126304626,
+      "learning_rate": 1.9993443330115592e-05,
+      "loss": 0.1295,
+      "step": 18
+    },
+    {
+      "epoch": 0.05397727272727273,
+      "grad_norm": 1.1795912981033325,
+      "learning_rate": 1.999170195558004e-05,
+      "loss": 0.1546,
+      "step": 19
+    },
+    {
+      "epoch": 0.056818181818181816,
+      "grad_norm": 0.7677745223045349,
+      "learning_rate": 1.9989755833107875e-05,
+      "loss": 0.2035,
+      "step": 20
+    },
+    {
+      "epoch": 0.05965909090909091,
+      "grad_norm": 1.0062402486801147,
+      "learning_rate": 1.9987605002578655e-05,
+      "loss": 0.2815,
+      "step": 21
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 0.7316843867301941,
+      "learning_rate": 1.9985249508066754e-05,
+      "loss": 0.1688,
+      "step": 22
+    },
+    {
+      "epoch": 0.06534090909090909,
+      "grad_norm": 0.8864314556121826,
+      "learning_rate": 1.9982689397840497e-05,
+      "loss": 0.2197,
+      "step": 23
+    },
+    {
+      "epoch": 0.06818181818181818,
+      "grad_norm": 0.7916576862335205,
+      "learning_rate": 1.997992472436114e-05,
+      "loss": 0.2421,
+      "step": 24
+    },
+    {
+      "epoch": 0.07102272727272728,
+      "grad_norm": 0.6953737735748291,
+      "learning_rate": 1.9976955544281815e-05,
+      "loss": 0.1548,
+      "step": 25
+    },
+    {
+      "epoch": 0.07386363636363637,
+      "grad_norm": 0.9317605495452881,
+      "learning_rate": 1.9973781918446363e-05,
+      "loss": 0.2112,
+      "step": 26
+    },
+    {
+      "epoch": 0.07670454545454546,
+      "grad_norm": 1.0706557035446167,
+      "learning_rate": 1.9970403911888077e-05,
+      "loss": 0.3094,
+      "step": 27
+    },
+    {
+      "epoch": 0.07954545454545454,
+      "grad_norm": 1.016023874282837,
+      "learning_rate": 1.9966821593828393e-05,
+      "loss": 0.2843,
+      "step": 28
+    },
+    {
+      "epoch": 0.08238636363636363,
+      "grad_norm": 0.7612195611000061,
+      "learning_rate": 1.996303503767544e-05,
+      "loss": 0.1433,
+      "step": 29
+    },
+    {
+      "epoch": 0.08522727272727272,
+      "grad_norm": 0.7575203776359558,
+      "learning_rate": 1.9959044321022563e-05,
+      "loss": 0.1604,
+      "step": 30
+    },
+    {
+      "epoch": 0.08806818181818182,
+      "grad_norm": 0.9231955409049988,
+      "learning_rate": 1.9954849525646727e-05,
+      "loss": 0.2021,
+      "step": 31
+    },
+    {
+      "epoch": 0.09090909090909091,
+      "grad_norm": 0.4868467450141907,
+      "learning_rate": 1.9950450737506825e-05,
+      "loss": 0.0573,
+      "step": 32
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 1.091597080230713,
+      "learning_rate": 1.9945848046741934e-05,
+      "loss": 0.2991,
+      "step": 33
+    },
+    {
+      "epoch": 0.09659090909090909,
+      "grad_norm": 0.8647293448448181,
+      "learning_rate": 1.9941041547669467e-05,
+      "loss": 0.2171,
+      "step": 34
+    },
+    {
+      "epoch": 0.09943181818181818,
+      "grad_norm": 0.48515236377716064,
+      "learning_rate": 1.9936031338783226e-05,
+      "loss": 0.0892,
+      "step": 35
+    },
+    {
+      "epoch": 0.10227272727272728,
+      "grad_norm": 0.8053379654884338,
+      "learning_rate": 1.9930817522751403e-05,
+      "loss": 0.1805,
+      "step": 36
+    },
+    {
+      "epoch": 0.10511363636363637,
+      "grad_norm": 0.648684024810791,
+      "learning_rate": 1.992540020641446e-05,
+      "loss": 0.1416,
+      "step": 37
+    },
+    {
+      "epoch": 0.10795454545454546,
+      "grad_norm": 0.5282648801803589,
+      "learning_rate": 1.991977950078295e-05,
+      "loss": 0.1451,
+      "step": 38
+    },
+    {
+      "epoch": 0.11079545454545454,
+      "grad_norm": 0.5510435104370117,
+      "learning_rate": 1.9913955521035234e-05,
+      "loss": 0.0875,
+      "step": 39
+    },
+    {
+      "epoch": 0.11363636363636363,
+      "grad_norm": 0.7555186152458191,
+      "learning_rate": 1.9907928386515126e-05,
+      "loss": 0.0847,
+      "step": 40
+    },
+    {
+      "epoch": 0.11647727272727272,
+      "grad_norm": 0.4893772006034851,
+      "learning_rate": 1.9901698220729458e-05,
+      "loss": 0.1075,
+      "step": 41
+    },
+    {
+      "epoch": 0.11931818181818182,
+      "grad_norm": 0.9237098097801208,
+      "learning_rate": 1.9895265151345516e-05,
+      "loss": 0.1608,
+      "step": 42
+    },
+    {
+      "epoch": 0.12215909090909091,
+      "grad_norm": 0.7260457277297974,
+      "learning_rate": 1.9888629310188467e-05,
+      "loss": 0.266,
+      "step": 43
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.4444400668144226,
+      "learning_rate": 1.9881790833238615e-05,
+      "loss": 0.1077,
+      "step": 44
+    },
+    {
+      "epoch": 0.1278409090909091,
+      "grad_norm": 1.2695385217666626,
+      "learning_rate": 1.9874749860628658e-05,
+      "loss": 0.136,
+      "step": 45
+    },
+    {
+      "epoch": 0.13068181818181818,
+      "grad_norm": 0.8816536068916321,
+      "learning_rate": 1.9867506536640782e-05,
+      "loss": 0.0873,
+      "step": 46
+    },
+    {
+      "epoch": 0.13352272727272727,
+      "grad_norm": 0.5350353121757507,
+      "learning_rate": 1.9860061009703714e-05,
+      "loss": 0.139,
+      "step": 47
+    },
+    {
+      "epoch": 0.13636363636363635,
+      "grad_norm": 0.49625250697135925,
+      "learning_rate": 1.9852413432389685e-05,
+      "loss": 0.1205,
+      "step": 48
+    },
+    {
+      "epoch": 0.13920454545454544,
+      "grad_norm": 0.5370039343833923,
+      "learning_rate": 1.9844563961411308e-05,
+      "loss": 0.1075,
+      "step": 49
+    },
+    {
+      "epoch": 0.14204545454545456,
+      "grad_norm": 1.234710454940796,
+      "learning_rate": 1.9836512757618355e-05,
+      "loss": 0.2059,
+      "step": 50
+    },
+    {
+      "epoch": 0.14488636363636365,
+      "grad_norm": 0.8026732206344604,
+      "learning_rate": 1.9828259985994465e-05,
+      "loss": 0.2543,
+      "step": 51
+    },
+    {
+      "epoch": 0.14772727272727273,
+      "grad_norm": 0.45312148332595825,
+      "learning_rate": 1.9819805815653768e-05,
+      "loss": 0.1001,
+      "step": 52
+    },
+    {
+      "epoch": 0.15056818181818182,
+      "grad_norm": 0.4792601466178894,
+      "learning_rate": 1.981115041983741e-05,
+      "loss": 0.1408,
+      "step": 53
+    },
+    {
+      "epoch": 0.1534090909090909,
+      "grad_norm": 0.40749409794807434,
+      "learning_rate": 1.9802293975910016e-05,
+      "loss": 0.0694,
+      "step": 54
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 0.5319134593009949,
+      "learning_rate": 1.9793236665356043e-05,
+      "loss": 0.1439,
+      "step": 55
+    },
+    {
+      "epoch": 0.1590909090909091,
+      "grad_norm": 0.5237418413162231,
+      "learning_rate": 1.9783978673776067e-05,
+      "loss": 0.1198,
+      "step": 56
+    },
+    {
+      "epoch": 0.16193181818181818,
+      "grad_norm": 0.5154628157615662,
+      "learning_rate": 1.9774520190882978e-05,
+      "loss": 0.0596,
+      "step": 57
+    },
+    {
+      "epoch": 0.16477272727272727,
+      "grad_norm": 0.6916084885597229,
+      "learning_rate": 1.97648614104981e-05,
+      "loss": 0.1561,
+      "step": 58
+    },
+    {
+      "epoch": 0.16761363636363635,
+      "grad_norm": 0.8141054511070251,
+      "learning_rate": 1.9755002530547206e-05,
+      "loss": 0.2444,
+      "step": 59
+    },
+    {
+      "epoch": 0.17045454545454544,
+      "grad_norm": 0.324054479598999,
+      "learning_rate": 1.974494375305647e-05,
+      "loss": 0.0467,
+      "step": 60
+    },
+    {
+      "epoch": 0.17329545454545456,
+      "grad_norm": 0.49560919404029846,
+      "learning_rate": 1.973468528414833e-05,
+      "loss": 0.1376,
+      "step": 61
+    },
+    {
+      "epoch": 0.17613636363636365,
+      "grad_norm": 0.4895784854888916,
+      "learning_rate": 1.9724227334037255e-05,
+      "loss": 0.1201,
+      "step": 62
+    },
+    {
+      "epoch": 0.17897727272727273,
+      "grad_norm": 0.47013720870018005,
+      "learning_rate": 1.9713570117025444e-05,
+      "loss": 0.1037,
+      "step": 63
+    },
+    {
+      "epoch": 0.18181818181818182,
+      "grad_norm": 0.618483304977417,
+      "learning_rate": 1.9702713851498436e-05,
+      "loss": 0.1531,
+      "step": 64
+    },
+    {
+      "epoch": 0.1846590909090909,
+      "grad_norm": 0.5431604385375977,
+      "learning_rate": 1.9691658759920625e-05,
+      "loss": 0.1495,
+      "step": 65
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 0.6768195033073425,
+      "learning_rate": 1.9680405068830716e-05,
+      "loss": 0.1585,
+      "step": 66
+    },
+    {
+      "epoch": 0.1903409090909091,
+      "grad_norm": 0.5289143919944763,
+      "learning_rate": 1.9668953008837073e-05,
+      "loss": 0.1082,
+      "step": 67
+    },
+    {
+      "epoch": 0.19318181818181818,
+      "grad_norm": 0.38905492424964905,
+      "learning_rate": 1.965730281461299e-05,
+      "loss": 0.0742,
+      "step": 68
+    },
+    {
+      "epoch": 0.19602272727272727,
+      "grad_norm": 0.6071842908859253,
+      "learning_rate": 1.96454547248919e-05,
+      "loss": 0.1822,
+      "step": 69
+    },
+    {
+      "epoch": 0.19886363636363635,
+      "grad_norm": 0.6174128651618958,
+      "learning_rate": 1.9633408982462453e-05,
+      "loss": 0.2036,
+      "step": 70
+    },
+    {
+      "epoch": 0.20170454545454544,
+      "grad_norm": 0.4610503613948822,
+      "learning_rate": 1.962116583416357e-05,
+      "loss": 0.0834,
+      "step": 71
+    },
+    {
+      "epoch": 0.20454545454545456,
+      "grad_norm": 0.5412945747375488,
+      "learning_rate": 1.9608725530879376e-05,
+      "loss": 0.1061,
+      "step": 72
+    },
+    {
+      "epoch": 0.20738636363636365,
+      "grad_norm": 0.2720443606376648,
+      "learning_rate": 1.9596088327534045e-05,
+      "loss": 0.0572,
+      "step": 73
+    },
+    {
+      "epoch": 0.21022727272727273,
+      "grad_norm": 0.3362009525299072,
+      "learning_rate": 1.95832544830866e-05,
+      "loss": 0.0596,
+      "step": 74
+    },
+    {
+      "epoch": 0.21306818181818182,
+      "grad_norm": 0.4517727494239807,
+      "learning_rate": 1.957022426052558e-05,
+      "loss": 0.0983,
+      "step": 75
+    },
+    {
+      "epoch": 0.2159090909090909,
+      "grad_norm": 0.44597986340522766,
+      "learning_rate": 1.9556997926863676e-05,
+      "loss": 0.1082,
+      "step": 76
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.4495086371898651,
+      "learning_rate": 1.9543575753132242e-05,
+      "loss": 0.0877,
+      "step": 77
+    },
+    {
+      "epoch": 0.2215909090909091,
+      "grad_norm": 0.4055721163749695,
+      "learning_rate": 1.9529958014375748e-05,
+      "loss": 0.1046,
+      "step": 78
+    },
+    {
+      "epoch": 0.22443181818181818,
+      "grad_norm": 0.5845265984535217,
+      "learning_rate": 1.9516144989646145e-05,
+      "loss": 0.1601,
+      "step": 79
+    },
+    {
+      "epoch": 0.22727272727272727,
+      "grad_norm": 0.5843460559844971,
+      "learning_rate": 1.9502136961997144e-05,
+      "loss": 0.1922,
+      "step": 80
+    },
+    {
+      "epoch": 0.23011363636363635,
+      "grad_norm": 0.4905230402946472,
+      "learning_rate": 1.9487934218478413e-05,
+      "loss": 0.1149,
+      "step": 81
+    },
+    {
+      "epoch": 0.23295454545454544,
+      "grad_norm": 0.760354220867157,
+      "learning_rate": 1.9473537050129704e-05,
+      "loss": 0.264,
+      "step": 82
+    },
+    {
+      "epoch": 0.23579545454545456,
+      "grad_norm": 0.5880870819091797,
+      "learning_rate": 1.945894575197488e-05,
+      "loss": 0.0957,
+      "step": 83
+    },
+    {
+      "epoch": 0.23863636363636365,
+      "grad_norm": 0.7324073314666748,
+      "learning_rate": 1.9444160623015877e-05,
+      "loss": 0.1335,
+      "step": 84
+    },
+    {
+      "epoch": 0.24147727272727273,
+      "grad_norm": 0.57627272605896,
+      "learning_rate": 1.9429181966226558e-05,
+      "loss": 0.1352,
+      "step": 85
+    },
+    {
+      "epoch": 0.24431818181818182,
+      "grad_norm": 0.41611605882644653,
+      "learning_rate": 1.9414010088546535e-05,
+      "loss": 0.0918,
+      "step": 86
+    },
+    {
+      "epoch": 0.2471590909090909,
+      "grad_norm": 0.5241921544075012,
+      "learning_rate": 1.9398645300874865e-05,
+      "loss": 0.1096,
+      "step": 87
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.45845428109169006,
+      "learning_rate": 1.9383087918063662e-05,
+      "loss": 0.139,
+      "step": 88
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.11835308372974396,
+      "eval_runtime": 324.452,
+      "eval_samples_per_second": 8.306,
+      "eval_steps_per_second": 1.039,
+      "step": 88
+    },
+    {
+      "epoch": 0.2528409090909091,
+      "grad_norm": 0.3612334728240967,
+      "learning_rate": 1.9367338258911674e-05,
+      "loss": 0.0563,
+      "step": 89
+    },
+    {
+      "epoch": 0.2556818181818182,
+      "grad_norm": 0.7781359553337097,
+      "learning_rate": 1.935139664615773e-05,
+      "loss": 0.1679,
+      "step": 90
+    },
+    {
+      "epoch": 0.2585227272727273,
+      "grad_norm": 0.41035768389701843,
+      "learning_rate": 1.933526340647414e-05,
+      "loss": 0.0654,
+      "step": 91
+    },
+    {
+      "epoch": 0.26136363636363635,
+      "grad_norm": 0.422101229429245,
+      "learning_rate": 1.9318938870459984e-05,
+      "loss": 0.0748,
+      "step": 92
+    },
+    {
+      "epoch": 0.26420454545454547,
+      "grad_norm": 0.33451277017593384,
+      "learning_rate": 1.930242337263436e-05,
+      "loss": 0.0502,
+      "step": 93
+    },
+    {
+      "epoch": 0.26704545454545453,
+      "grad_norm": 0.5042566061019897,
+      "learning_rate": 1.9285717251429507e-05,
+      "loss": 0.0972,
+      "step": 94
+    },
+    {
+      "epoch": 0.26988636363636365,
+      "grad_norm": 0.3712761402130127,
+      "learning_rate": 1.9268820849183885e-05,
+      "loss": 0.0804,
+      "step": 95
+    },
+    {
+      "epoch": 0.2727272727272727,
+      "grad_norm": 0.6084650754928589,
+      "learning_rate": 1.9251734512135155e-05,
+      "loss": 0.1383,
+      "step": 96
+    },
+    {
+      "epoch": 0.2755681818181818,
+      "grad_norm": 0.5942140817642212,
+      "learning_rate": 1.9234458590413076e-05,
+      "loss": 0.2123,
+      "step": 97
+    },
+    {
+      "epoch": 0.2784090909090909,
+      "grad_norm": 0.4847292900085449,
+      "learning_rate": 1.921699343803235e-05,
+      "loss": 0.0427,
+      "step": 98
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.6673583388328552,
+      "learning_rate": 1.9199339412885347e-05,
+      "loss": 0.2334,
+      "step": 99
+    },
+    {
+      "epoch": 0.2840909090909091,
+      "grad_norm": 0.5910840034484863,
+      "learning_rate": 1.918149687673478e-05,
+      "loss": 0.1798,
+      "step": 100
+    },
+    {
+      "epoch": 0.2869318181818182,
+      "grad_norm": 0.4884462356567383,
+      "learning_rate": 1.9163466195206292e-05,
+      "loss": 0.0858,
+      "step": 101
+    },
+    {
+      "epoch": 0.2897727272727273,
+      "grad_norm": 0.45178908109664917,
+      "learning_rate": 1.9145247737780963e-05,
+      "loss": 0.0957,
+      "step": 102
+    },
+    {
+      "epoch": 0.29261363636363635,
+      "grad_norm": 0.6220026016235352,
+      "learning_rate": 1.9126841877787745e-05,
+      "loss": 0.215,
+      "step": 103
+    },
+    {
+      "epoch": 0.29545454545454547,
+      "grad_norm": 0.4645046591758728,
+      "learning_rate": 1.9108248992395797e-05,
+      "loss": 0.1185,
+      "step": 104
+    },
+    {
+      "epoch": 0.29829545454545453,
+      "grad_norm": 0.4531736373901367,
+      "learning_rate": 1.9089469462606766e-05,
+      "loss": 0.1177,
+      "step": 105
+    },
+    {
+      "epoch": 0.30113636363636365,
+      "grad_norm": 0.5413585901260376,
+      "learning_rate": 1.9070503673246982e-05,
+      "loss": 0.1,
+      "step": 106
+    },
+    {
+      "epoch": 0.3039772727272727,
+      "grad_norm": 0.3162194490432739,
+      "learning_rate": 1.905135201295957e-05,
+      "loss": 0.046,
+      "step": 107
+    },
+    {
+      "epoch": 0.3068181818181818,
+      "grad_norm": 0.41331952810287476,
+      "learning_rate": 1.9032014874196476e-05,
+      "loss": 0.0757,
+      "step": 108
+    },
+    {
+      "epoch": 0.3096590909090909,
+      "grad_norm": 0.41365161538124084,
+      "learning_rate": 1.901249265321044e-05,
+      "loss": 0.0719,
+      "step": 109
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.762642502784729,
+      "learning_rate": 1.8992785750046866e-05,
+      "loss": 0.3067,
+      "step": 110
+    },
+    {
+      "epoch": 0.3153409090909091,
+      "grad_norm": 0.5370594263076782,
+      "learning_rate": 1.8972894568535632e-05,
+      "loss": 0.1088,
+      "step": 111
+    },
+    {
+      "epoch": 0.3181818181818182,
+      "grad_norm": 0.6596404910087585,
+      "learning_rate": 1.895281951628281e-05,
+      "loss": 0.1886,
+      "step": 112
+    },
+    {
+      "epoch": 0.3210227272727273,
+      "grad_norm": 0.4068351984024048,
+      "learning_rate": 1.8932561004662312e-05,
+      "loss": 0.0762,
+      "step": 113
+    },
+    {
+      "epoch": 0.32386363636363635,
+      "grad_norm": 0.6985118389129639,
+      "learning_rate": 1.891211944880746e-05,
+      "loss": 0.2173,
+      "step": 114
+    },
+    {
+      "epoch": 0.32670454545454547,
+      "grad_norm": 0.24498817324638367,
+      "learning_rate": 1.8891495267602482e-05,
+      "loss": 0.0333,
+      "step": 115
+    },
+    {
+      "epoch": 0.32954545454545453,
+      "grad_norm": 0.6893118023872375,
+      "learning_rate": 1.887068888367394e-05,
+      "loss": 0.2257,
+      "step": 116
+    },
+    {
+      "epoch": 0.33238636363636365,
+      "grad_norm": 0.5168837904930115,
+      "learning_rate": 1.8849700723382038e-05,
+      "loss": 0.1399,
+      "step": 117
+    },
+    {
+      "epoch": 0.3352272727272727,
+      "grad_norm": 0.5254336595535278,
+      "learning_rate": 1.8828531216811912e-05,
+      "loss": 0.0815,
+      "step": 118
+    },
+    {
+      "epoch": 0.3380681818181818,
+      "grad_norm": 0.44607385993003845,
+      "learning_rate": 1.8807180797764822e-05,
+      "loss": 0.0736,
+      "step": 119
+    },
+    {
+      "epoch": 0.3409090909090909,
+      "grad_norm": 0.5060997605323792,
+      "learning_rate": 1.8785649903749236e-05,
+      "loss": 0.1319,
+      "step": 120
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.37058401107788086,
+      "learning_rate": 1.8763938975971873e-05,
+      "loss": 0.0597,
+      "step": 121
+    },
+    {
+      "epoch": 0.3465909090909091,
+      "grad_norm": 0.861648440361023,
+      "learning_rate": 1.8742048459328684e-05,
+      "loss": 0.2988,
+      "step": 122
+    },
+    {
+      "epoch": 0.3494318181818182,
+      "grad_norm": 0.48567265272140503,
+      "learning_rate": 1.8719978802395707e-05,
+      "loss": 0.0934,
+      "step": 123
+    },
+    {
+      "epoch": 0.3522727272727273,
+      "grad_norm": 0.3372555673122406,
+      "learning_rate": 1.8697730457419893e-05,
+      "loss": 0.0523,
+      "step": 124
+    },
+    {
+      "epoch": 0.35511363636363635,
+      "grad_norm": 0.42780211567878723,
+      "learning_rate": 1.8675303880309828e-05,
+      "loss": 0.0615,
+      "step": 125
+    },
+    {
+      "epoch": 0.35795454545454547,
+      "grad_norm": 0.601543128490448,
+      "learning_rate": 1.8652699530626398e-05,
+      "loss": 0.2145,
+      "step": 126
+    },
+    {
+      "epoch": 0.36079545454545453,
+      "grad_norm": 0.42834290862083435,
+      "learning_rate": 1.8629917871573368e-05,
+      "loss": 0.0707,
+      "step": 127
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 0.3425846993923187,
+      "learning_rate": 1.8606959369987885e-05,
+      "loss": 0.06,
+      "step": 128
+    },
+    {
+      "epoch": 0.3664772727272727,
+      "grad_norm": 0.2861498296260834,
+      "learning_rate": 1.8583824496330925e-05,
+      "loss": 0.0591,
+      "step": 129
+    },
+    {
+      "epoch": 0.3693181818181818,
+      "grad_norm": 0.8360418677330017,
+      "learning_rate": 1.8560513724677644e-05,
+      "loss": 0.3178,
+      "step": 130
+    },
+    {
+      "epoch": 0.3721590909090909,
+      "grad_norm": 0.5859141945838928,
+      "learning_rate": 1.853702753270766e-05,
+      "loss": 0.173,
+      "step": 131
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.3947351276874542,
+      "learning_rate": 1.8513366401695277e-05,
+      "loss": 0.0649,
+      "step": 132
+    },
+    {
+      "epoch": 0.3778409090909091,
+      "grad_norm": 0.44475170969963074,
+      "learning_rate": 1.8489530816499597e-05,
+      "loss": 0.0705,
+      "step": 133
+    },
+    {
+      "epoch": 0.3806818181818182,
+      "grad_norm": 0.40042588114738464,
+      "learning_rate": 1.8465521265554623e-05,
+      "loss": 0.053,
+      "step": 134
+    },
+    {
+      "epoch": 0.3835227272727273,
+      "grad_norm": 0.32334399223327637,
+      "learning_rate": 1.8441338240859216e-05,
+      "loss": 0.0741,
+      "step": 135
+    },
+    {
+      "epoch": 0.38636363636363635,
+      "grad_norm": 0.4624631106853485,
+      "learning_rate": 1.841698223796703e-05,
+      "loss": 0.084,
+      "step": 136
+    },
+    {
+      "epoch": 0.38920454545454547,
+      "grad_norm": 0.5654751062393188,
+      "learning_rate": 1.839245375597635e-05,
+      "loss": 0.1694,
+      "step": 137
+    },
+    {
+      "epoch": 0.39204545454545453,
+      "grad_norm": 0.5055428147315979,
+      "learning_rate": 1.8367753297519874e-05,
+      "loss": 0.1569,
+      "step": 138
+    },
+    {
+      "epoch": 0.39488636363636365,
+      "grad_norm": 0.5599436163902283,
+      "learning_rate": 1.8342881368754403e-05,
+      "loss": 0.2114,
+      "step": 139
+    },
+    {
+      "epoch": 0.3977272727272727,
+      "grad_norm": 0.6785524487495422,
+      "learning_rate": 1.8317838479350473e-05,
+      "loss": 0.2652,
+      "step": 140
+    },
+    {
+      "epoch": 0.4005681818181818,
+      "grad_norm": 0.22756998240947723,
+      "learning_rate": 1.829262514248191e-05,
+      "loss": 0.0325,
+      "step": 141
+    },
+    {
+      "epoch": 0.4034090909090909,
+      "grad_norm": 0.3240101635456085,
+      "learning_rate": 1.8267241874815312e-05,
+      "loss": 0.0677,
+      "step": 142
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 0.5141014456748962,
+      "learning_rate": 1.8241689196499475e-05,
+      "loss": 0.0904,
+      "step": 143
+    },
+    {
+      "epoch": 0.4090909090909091,
+      "grad_norm": 0.4461572766304016,
+      "learning_rate": 1.8215967631154718e-05,
+      "loss": 0.068,
+      "step": 144
+    },
+    {
+      "epoch": 0.4119318181818182,
+      "grad_norm": 0.5313737988471985,
+      "learning_rate": 1.8190077705862155e-05,
+      "loss": 0.0917,
+      "step": 145
+    },
+    {
+      "epoch": 0.4147727272727273,
+      "grad_norm": 0.6240700483322144,
+      "learning_rate": 1.8164019951152903e-05,
+      "loss": 0.1616,
+      "step": 146
+    },
+    {
+      "epoch": 0.41761363636363635,
+      "grad_norm": 0.6356779932975769,
+      "learning_rate": 1.8137794900997204e-05,
+      "loss": 0.2113,
+      "step": 147
+    },
+    {
+      "epoch": 0.42045454545454547,
+      "grad_norm": 0.5394648909568787,
+      "learning_rate": 1.811140309279348e-05,
+      "loss": 0.0956,
+      "step": 148
+    },
+    {
+      "epoch": 0.42329545454545453,
+      "grad_norm": 0.5533218383789062,
+      "learning_rate": 1.8084845067357334e-05,
+      "loss": 0.1166,
+      "step": 149
+    },
+    {
+      "epoch": 0.42613636363636365,
+      "grad_norm": 0.5690510869026184,
+      "learning_rate": 1.805812136891046e-05,
+      "loss": 0.1485,
+      "step": 150
+    },
+    {
+      "epoch": 0.4289772727272727,
+      "grad_norm": 0.4042081832885742,
+      "learning_rate": 1.8031232545069468e-05,
+      "loss": 0.0855,
+      "step": 151
+    },
+    {
+      "epoch": 0.4318181818181818,
+      "grad_norm": 0.6914238929748535,
+      "learning_rate": 1.8004179146834712e-05,
+      "loss": 0.1606,
+      "step": 152
+    },
+    {
+      "epoch": 0.4346590909090909,
+      "grad_norm": 0.2679579555988312,
+      "learning_rate": 1.7976961728578965e-05,
+      "loss": 0.0242,
+      "step": 153
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.8089874386787415,
+      "learning_rate": 1.7949580848036045e-05,
+      "loss": 0.147,
+      "step": 154
+    },
+    {
+      "epoch": 0.4403409090909091,
+      "grad_norm": 0.5541207194328308,
+      "learning_rate": 1.7922037066289432e-05,
+      "loss": 0.0898,
+      "step": 155
+    },
+    {
+      "epoch": 0.4431818181818182,
+      "grad_norm": 0.451183557510376,
+      "learning_rate": 1.7894330947760728e-05,
+      "loss": 0.1032,
+      "step": 156
+    },
+    {
+      "epoch": 0.4460227272727273,
+      "grad_norm": 0.4263628423213959,
+      "learning_rate": 1.7866463060198114e-05,
+      "loss": 0.0744,
+      "step": 157
+    },
+    {
+      "epoch": 0.44886363636363635,
+      "grad_norm": 0.3324938714504242,
+      "learning_rate": 1.7838433974664714e-05,
+      "loss": 0.0737,
+      "step": 158
+    },
+    {
+      "epoch": 0.45170454545454547,
+      "grad_norm": 0.4336640238761902,
+      "learning_rate": 1.7810244265526873e-05,
+      "loss": 0.0697,
+      "step": 159
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 0.46159589290618896,
+      "learning_rate": 1.778189451044242e-05,
+      "loss": 0.1031,
+      "step": 160
+    },
+    {
+      "epoch": 0.45738636363636365,
+      "grad_norm": 0.22606071829795837,
+      "learning_rate": 1.77533852903488e-05,
+      "loss": 0.0445,
+      "step": 161
+    },
+    {
+      "epoch": 0.4602272727272727,
+      "grad_norm": 0.468505322933197,
+      "learning_rate": 1.772471718945119e-05,
+      "loss": 0.1115,
+      "step": 162
+    },
+    {
+      "epoch": 0.4630681818181818,
+      "grad_norm": 0.5148159861564636,
+      "learning_rate": 1.7695890795210517e-05,
+      "loss": 0.1074,
+      "step": 163
+    },
+    {
+      "epoch": 0.4659090909090909,
+      "grad_norm": 0.46925628185272217,
+      "learning_rate": 1.7666906698331426e-05,
+      "loss": 0.105,
+      "step": 164
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.9212394952774048,
+      "learning_rate": 1.7637765492750172e-05,
+      "loss": 0.226,
+      "step": 165
+    },
+    {
+      "epoch": 0.4715909090909091,
+      "grad_norm": 0.3673381209373474,
+      "learning_rate": 1.7608467775622448e-05,
+      "loss": 0.118,
+      "step": 166
+    },
+    {
+      "epoch": 0.4744318181818182,
+      "grad_norm": 0.5593389868736267,
+      "learning_rate": 1.757901414731115e-05,
+      "loss": 0.1513,
+      "step": 167
+    },
+    {
+      "epoch": 0.4772727272727273,
+      "grad_norm": 0.3432629704475403,
+      "learning_rate": 1.7549405211374072e-05,
+      "loss": 0.0576,
+      "step": 168
+    },
+    {
+      "epoch": 0.48011363636363635,
+      "grad_norm": 0.325641006231308,
+      "learning_rate": 1.7519641574551548e-05,
+      "loss": 0.061,
+      "step": 169
+    },
+    {
+      "epoch": 0.48295454545454547,
+      "grad_norm": 0.49900880455970764,
+      "learning_rate": 1.7489723846754e-05,
+      "loss": 0.0929,
+      "step": 170
+    },
+    {
+      "epoch": 0.48579545454545453,
+      "grad_norm": 0.4341491162776947,
+      "learning_rate": 1.7459652641049473e-05,
+      "loss": 0.0941,
+      "step": 171
+    },
+    {
+      "epoch": 0.48863636363636365,
+      "grad_norm": 0.48210591077804565,
+      "learning_rate": 1.7429428573651023e-05,
+      "loss": 0.0543,
+      "step": 172
+    },
+    {
+      "epoch": 0.4914772727272727,
+      "grad_norm": 0.5630267262458801,
+      "learning_rate": 1.739905226390413e-05,
+      "loss": 0.1008,
+      "step": 173
+    },
+    {
+      "epoch": 0.4943181818181818,
+      "grad_norm": 0.6051055192947388,
+      "learning_rate": 1.7368524334273998e-05,
+      "loss": 0.1985,
+      "step": 174
+    },
+    {
+      "epoch": 0.4971590909090909,
+      "grad_norm": 0.28148218989372253,
+      "learning_rate": 1.7337845410332784e-05,
+      "loss": 0.0499,
+      "step": 175
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6560457348823547,
+      "learning_rate": 1.73070161207468e-05,
+      "loss": 0.1212,
+      "step": 176
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.0832223892211914,
+      "eval_runtime": 325.0904,
+      "eval_samples_per_second": 8.29,
+      "eval_steps_per_second": 1.037,
+      "step": 176
+    },
+    {
+      "epoch": 0.5028409090909091,
+      "grad_norm": 0.6473656296730042,
+      "learning_rate": 1.727603709726361e-05,
+      "loss": 0.0804,
+      "step": 177
+    },
+    {
+      "epoch": 0.5056818181818182,
+      "grad_norm": 0.34419476985931396,
+      "learning_rate": 1.7244908974699112e-05,
+      "loss": 0.0415,
+      "step": 178
+    },
+    {
+      "epoch": 0.5085227272727273,
+      "grad_norm": 0.3303261697292328,
+      "learning_rate": 1.7213632390924487e-05,
+      "loss": 0.0659,
+      "step": 179
+    },
+    {
+      "epoch": 0.5113636363636364,
+      "grad_norm": 0.32229432463645935,
+      "learning_rate": 1.7182207986853176e-05,
+      "loss": 0.0973,
+      "step": 180
+    },
+    {
+      "epoch": 0.5142045454545454,
+      "grad_norm": 0.44095346331596375,
+      "learning_rate": 1.7150636406427713e-05,
+      "loss": 0.0329,
+      "step": 181
+    },
+    {
+      "epoch": 0.5170454545454546,
+      "grad_norm": 0.43706265091896057,
+      "learning_rate": 1.7118918296606536e-05,
+      "loss": 0.0854,
+      "step": 182
+    },
+    {
+      "epoch": 0.5198863636363636,
+      "grad_norm": 0.32199347019195557,
+      "learning_rate": 1.708705430735075e-05,
+      "loss": 0.0327,
+      "step": 183
+    },
+    {
+      "epoch": 0.5227272727272727,
+      "grad_norm": 0.30175983905792236,
+      "learning_rate": 1.7055045091610772e-05,
+      "loss": 0.0622,
+      "step": 184
+    },
+    {
+      "epoch": 0.5255681818181818,
+      "grad_norm": 0.4883711338043213,
+      "learning_rate": 1.7022891305312987e-05,
+      "loss": 0.0977,
+      "step": 185
+    },
+    {
+      "epoch": 0.5284090909090909,
+      "grad_norm": 0.6222572922706604,
+      "learning_rate": 1.6990593607346276e-05,
+      "loss": 0.1161,
+      "step": 186
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.415580153465271,
+      "learning_rate": 1.695815265954855e-05,
+      "loss": 0.0576,
+      "step": 187
+    },
+    {
+      "epoch": 0.5340909090909091,
+      "grad_norm": 0.3268929421901703,
+      "learning_rate": 1.6925569126693135e-05,
+      "loss": 0.0627,
+      "step": 188
+    },
+    {
+      "epoch": 0.5369318181818182,
+      "grad_norm": 0.41048845648765564,
+      "learning_rate": 1.6892843676475212e-05,
+      "loss": 0.0333,
+      "step": 189
+    },
+    {
+      "epoch": 0.5397727272727273,
+      "grad_norm": 0.414117693901062,
+      "learning_rate": 1.685997697949809e-05,
+      "loss": 0.1126,
+      "step": 190
+    },
+    {
+      "epoch": 0.5426136363636364,
+      "grad_norm": 0.2861213982105255,
+      "learning_rate": 1.6826969709259476e-05,
+      "loss": 0.0293,
+      "step": 191
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 0.55129075050354,
+      "learning_rate": 1.6793822542137683e-05,
+      "loss": 0.136,
+      "step": 192
+    },
+    {
+      "epoch": 0.5482954545454546,
+      "grad_norm": 0.32792291045188904,
+      "learning_rate": 1.6760536157377754e-05,
+      "loss": 0.0898,
+      "step": 193
+    },
+    {
+      "epoch": 0.5511363636363636,
+      "grad_norm": 0.5784844756126404,
+      "learning_rate": 1.672711123707756e-05,
+      "loss": 0.0997,
+      "step": 194
+    },
+    {
+      "epoch": 0.5539772727272727,
+      "grad_norm": 0.9223653674125671,
+      "learning_rate": 1.6693548466173812e-05,
+      "loss": 0.2186,
+      "step": 195
+    },
+    {
+      "epoch": 0.5568181818181818,
+      "grad_norm": 0.44497421383857727,
+      "learning_rate": 1.665984853242802e-05,
+      "loss": 0.1132,
+      "step": 196
+    },
+    {
+      "epoch": 0.5596590909090909,
+      "grad_norm": 0.3333360552787781,
+      "learning_rate": 1.662601212641242e-05,
+      "loss": 0.0383,
+      "step": 197
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.5328719615936279,
+      "learning_rate": 1.6592039941495803e-05,
+      "loss": 0.0994,
+      "step": 198
+    },
+    {
+      "epoch": 0.5653409090909091,
+      "grad_norm": 0.28615421056747437,
+      "learning_rate": 1.6557932673829313e-05,
+      "loss": 0.0417,
+      "step": 199
+    },
+    {
+      "epoch": 0.5681818181818182,
+      "grad_norm": 0.314860463142395,
+      "learning_rate": 1.6523691022332184e-05,
+      "loss": 0.07,
+      "step": 200
+    },
+    {
+      "epoch": 0.5710227272727273,
+      "grad_norm": 0.44072073698043823,
+      "learning_rate": 1.6489315688677416e-05,
+      "loss": 0.0709,
+      "step": 201
+    },
+    {
+      "epoch": 0.5738636363636364,
+      "grad_norm": 0.5856392979621887,
+      "learning_rate": 1.64548073772774e-05,
+      "loss": 0.1405,
+      "step": 202
+    },
+    {
+      "epoch": 0.5767045454545454,
+      "grad_norm": 0.36462002992630005,
+      "learning_rate": 1.6420166795269476e-05,
+      "loss": 0.0441,
+      "step": 203
+    },
+    {
+      "epoch": 0.5795454545454546,
+      "grad_norm": 0.6672521233558655,
+      "learning_rate": 1.6385394652501444e-05,
+      "loss": 0.1754,
+      "step": 204
+    },
+    {
+      "epoch": 0.5823863636363636,
+      "grad_norm": 0.35214805603027344,
+      "learning_rate": 1.6350491661517032e-05,
+      "loss": 0.0583,
+      "step": 205
+    },
+    {
+      "epoch": 0.5852272727272727,
+      "grad_norm": 0.3882180452346802,
+      "learning_rate": 1.6315458537541272e-05,
+      "loss": 0.0775,
+      "step": 206
+    },
+    {
+      "epoch": 0.5880681818181818,
+      "grad_norm": 0.39624354243278503,
+      "learning_rate": 1.628029599846585e-05,
+      "loss": 0.0779,
+      "step": 207
+    },
+    {
+      "epoch": 0.5909090909090909,
+      "grad_norm": 0.5808979272842407,
+      "learning_rate": 1.6245004764834423e-05,
+      "loss": 0.0635,
+      "step": 208
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.23519928753376007,
+      "learning_rate": 1.6209585559827806e-05,
+      "loss": 0.037,
+      "step": 209
+    },
+    {
+      "epoch": 0.5965909090909091,
+      "grad_norm": 0.40159645676612854,
+      "learning_rate": 1.617403910924919e-05,
+      "loss": 0.0582,
+      "step": 210
+    },
+    {
+      "epoch": 0.5994318181818182,
+      "grad_norm": 0.38090643286705017,
+      "learning_rate": 1.613836614150926e-05,
+      "loss": 0.0981,
+      "step": 211
+    },
+    {
+      "epoch": 0.6022727272727273,
+      "grad_norm": 0.44124341011047363,
+      "learning_rate": 1.610256738761125e-05,
+      "loss": 0.0867,
+      "step": 212
+    },
+    {
+      "epoch": 0.6051136363636364,
+      "grad_norm": 0.3115885257720947,
+      "learning_rate": 1.6066643581135994e-05,
+      "loss": 0.0534,
+      "step": 213
+    },
+    {
+      "epoch": 0.6079545454545454,
+      "grad_norm": 0.47533664107322693,
+      "learning_rate": 1.6030595458226872e-05,
+      "loss": 0.099,
+      "step": 214
+    },
+    {
+      "epoch": 0.6107954545454546,
+      "grad_norm": 0.48215967416763306,
+      "learning_rate": 1.599442375757473e-05,
+      "loss": 0.0691,
+      "step": 215
+    },
+    {
+      "epoch": 0.6136363636363636,
+      "grad_norm": 0.46541547775268555,
+      "learning_rate": 1.5958129220402745e-05,
+      "loss": 0.0872,
+      "step": 216
+    },
+    {
+      "epoch": 0.6164772727272727,
+      "grad_norm": 0.5003085136413574,
+      "learning_rate": 1.5921712590451238e-05,
+      "loss": 0.1414,
+      "step": 217
+    },
+    {
+      "epoch": 0.6193181818181818,
+      "grad_norm": 0.28242507576942444,
+      "learning_rate": 1.5885174613962427e-05,
+      "loss": 0.0332,
+      "step": 218
+    },
+    {
+      "epoch": 0.6221590909090909,
+      "grad_norm": 0.574775218963623,
+      "learning_rate": 1.584851603966514e-05,
+      "loss": 0.1244,
+      "step": 219
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.5019505620002747,
+      "learning_rate": 1.581173761875947e-05,
+      "loss": 0.0896,
+      "step": 220
+    },
+    {
+      "epoch": 0.6278409090909091,
+      "grad_norm": 0.34306174516677856,
+      "learning_rate": 1.5774840104901378e-05,
+      "loss": 0.0771,
+      "step": 221
+    },
+    {
+      "epoch": 0.6306818181818182,
+      "grad_norm": 0.2710534334182739,
+      "learning_rate": 1.5737824254187275e-05,
+      "loss": 0.0272,
+      "step": 222
+    },
+    {
+      "epoch": 0.6335227272727273,
+      "grad_norm": 0.43892914056777954,
+      "learning_rate": 1.5700690825138473e-05,
+      "loss": 0.1656,
+      "step": 223
+    },
+    {
+      "epoch": 0.6363636363636364,
+      "grad_norm": 0.36996838450431824,
+      "learning_rate": 1.5663440578685706e-05,
+      "loss": 0.0549,
+      "step": 224
+    },
+    {
+      "epoch": 0.6392045454545454,
+      "grad_norm": 0.33289599418640137,
+      "learning_rate": 1.5626074278153485e-05,
+      "loss": 0.0536,
+      "step": 225
+    },
+    {
+      "epoch": 0.6420454545454546,
+      "grad_norm": 0.25877413153648376,
+      "learning_rate": 1.5588592689244492e-05,
+      "loss": 0.0534,
+      "step": 226
+    },
+    {
+      "epoch": 0.6448863636363636,
+      "grad_norm": 0.402020663022995,
+      "learning_rate": 1.555099658002387e-05,
+      "loss": 0.0455,
+      "step": 227
+    },
+    {
+      "epoch": 0.6477272727272727,
+      "grad_norm": 0.5195022821426392,
+      "learning_rate": 1.5513286720903488e-05,
+      "loss": 0.0739,
+      "step": 228
+    },
+    {
+      "epoch": 0.6505681818181818,
+      "grad_norm": 0.4449024200439453,
+      "learning_rate": 1.5475463884626153e-05,
+      "loss": 0.0849,
+      "step": 229
+    },
+    {
+      "epoch": 0.6534090909090909,
+      "grad_norm": 0.473652184009552,
+      "learning_rate": 1.5437528846249783e-05,
+      "loss": 0.1307,
+      "step": 230
+    },
+    {
+      "epoch": 0.65625,
+      "grad_norm": 0.43150216341018677,
+      "learning_rate": 1.5399482383131516e-05,
+      "loss": 0.0455,
+      "step": 231
+    },
+    {
+      "epoch": 0.6590909090909091,
+      "grad_norm": 0.7418766021728516,
+      "learning_rate": 1.536132527491178e-05,
+      "loss": 0.143,
+      "step": 232
+    },
+    {
+      "epoch": 0.6619318181818182,
+      "grad_norm": 0.3717086911201477,
+      "learning_rate": 1.5323058303498326e-05,
+      "loss": 0.1144,
+      "step": 233
+    },
+    {
+      "epoch": 0.6647727272727273,
+      "grad_norm": 0.33135896921157837,
+      "learning_rate": 1.52846822530502e-05,
+      "loss": 0.0345,
+      "step": 234
+    },
+    {
+      "epoch": 0.6676136363636364,
+      "grad_norm": 0.5481773614883423,
+      "learning_rate": 1.5246197909961659e-05,
+      "loss": 0.1416,
+      "step": 235
+    },
+    {
+      "epoch": 0.6704545454545454,
+      "grad_norm": 0.3348052203655243,
+      "learning_rate": 1.5207606062846092e-05,
+      "loss": 0.0578,
+      "step": 236
+    },
+    {
+      "epoch": 0.6732954545454546,
+      "grad_norm": 0.5042082667350769,
+      "learning_rate": 1.5168907502519823e-05,
+      "loss": 0.059,
+      "step": 237
+    },
+    {
+      "epoch": 0.6761363636363636,
+      "grad_norm": 0.3570152819156647,
+      "learning_rate": 1.5130103021985929e-05,
+      "loss": 0.106,
+      "step": 238
+    },
+    {
+      "epoch": 0.6789772727272727,
+      "grad_norm": 0.5292866826057434,
+      "learning_rate": 1.5091193416417981e-05,
+      "loss": 0.1311,
+      "step": 239
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 0.44494545459747314,
+      "learning_rate": 1.5052179483143752e-05,
+      "loss": 0.0788,
+      "step": 240
+    },
+    {
+      "epoch": 0.6846590909090909,
+      "grad_norm": 0.3993355929851532,
+      "learning_rate": 1.501306202162887e-05,
+      "loss": 0.044,
+      "step": 241
+    },
+    {
+      "epoch": 0.6875,
+      "grad_norm": 0.4046841561794281,
+      "learning_rate": 1.4973841833460457e-05,
+      "loss": 0.066,
+      "step": 242
+    },
+    {
+      "epoch": 0.6903409090909091,
+      "grad_norm": 0.247298926115036,
+      "learning_rate": 1.493451972233067e-05,
+      "loss": 0.0302,
+      "step": 243
+    },
+    {
+      "epoch": 0.6931818181818182,
+      "grad_norm": 0.4500928521156311,
+      "learning_rate": 1.4895096494020274e-05,
+      "loss": 0.0892,
+      "step": 244
+    },
+    {
+      "epoch": 0.6960227272727273,
+      "grad_norm": 0.373241126537323,
+      "learning_rate": 1.4855572956382082e-05,
+      "loss": 0.0691,
+      "step": 245
+    },
+    {
+      "epoch": 0.6988636363636364,
+      "grad_norm": 0.5434209704399109,
+      "learning_rate": 1.4815949919324443e-05,
+      "loss": 0.141,
+      "step": 246
+    },
+    {
+      "epoch": 0.7017045454545454,
+      "grad_norm": 0.43519458174705505,
+      "learning_rate": 1.4776228194794624e-05,
+      "loss": 0.0753,
+      "step": 247
+    },
+    {
+      "epoch": 0.7045454545454546,
+      "grad_norm": 0.23297618329524994,
+      "learning_rate": 1.473640859676217e-05,
+      "loss": 0.028,
+      "step": 248
+    },
+    {
+      "epoch": 0.7073863636363636,
+      "grad_norm": 0.6036296486854553,
+      "learning_rate": 1.469649194120224e-05,
+      "loss": 0.0811,
+      "step": 249
+    },
+    {
+      "epoch": 0.7102272727272727,
+      "grad_norm": 0.4559984505176544,
+      "learning_rate": 1.4656479046078862e-05,
+      "loss": 0.0841,
+      "step": 250
+    },
+    {
+      "epoch": 0.7130681818181818,
+      "grad_norm": 0.5498476624488831,
+      "learning_rate": 1.46163707313282e-05,
+      "loss": 0.1086,
+      "step": 251
+    },
+    {
+      "epoch": 0.7159090909090909,
+      "grad_norm": 0.4110633134841919,
+      "learning_rate": 1.4576167818841732e-05,
+      "loss": 0.076,
+      "step": 252
+    },
+    {
+      "epoch": 0.71875,
+      "grad_norm": 0.35567420721054077,
+      "learning_rate": 1.453587113244941e-05,
+      "loss": 0.0528,
+      "step": 253
+    },
+    {
+      "epoch": 0.7215909090909091,
+      "grad_norm": 0.6413715481758118,
+      "learning_rate": 1.4495481497902789e-05,
+      "loss": 0.1407,
+      "step": 254
+    },
+    {
+      "epoch": 0.7244318181818182,
+      "grad_norm": 0.496463805437088,
+      "learning_rate": 1.4454999742858093e-05,
+      "loss": 0.1195,
+      "step": 255
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.28542259335517883,
+      "learning_rate": 1.441442669685926e-05,
+      "loss": 0.0543,
+      "step": 256
+    },
+    {
+      "epoch": 0.7301136363636364,
+      "grad_norm": 0.32590752840042114,
+      "learning_rate": 1.4373763191320955e-05,
+      "loss": 0.0458,
+      "step": 257
+    },
+    {
+      "epoch": 0.7329545454545454,
+      "grad_norm": 0.46513354778289795,
+      "learning_rate": 1.4333010059511505e-05,
+      "loss": 0.0999,
+      "step": 258
+    },
+    {
+      "epoch": 0.7357954545454546,
+      "grad_norm": 0.5399053692817688,
+      "learning_rate": 1.4292168136535856e-05,
+      "loss": 0.1079,
+      "step": 259
+    },
+    {
+      "epoch": 0.7386363636363636,
+      "grad_norm": 0.27590784430503845,
+      "learning_rate": 1.425123825931843e-05,
+      "loss": 0.0394,
+      "step": 260
+    },
+    {
+      "epoch": 0.7414772727272727,
+      "grad_norm": 0.47675570845603943,
+      "learning_rate": 1.4210221266585997e-05,
+      "loss": 0.1427,
+      "step": 261
+    },
+    {
+      "epoch": 0.7443181818181818,
+      "grad_norm": 0.26074454188346863,
+      "learning_rate": 1.416911799885049e-05,
+      "loss": 0.0376,
+      "step": 262
+    },
+    {
+      "epoch": 0.7471590909090909,
+      "grad_norm": 0.3271232843399048,
+      "learning_rate": 1.4127929298391749e-05,
+      "loss": 0.0375,
+      "step": 263
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5497257113456726,
+      "learning_rate": 1.4086656009240308e-05,
+      "loss": 0.0691,
+      "step": 264
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.06671788543462753,
+      "eval_runtime": 324.3338,
+      "eval_samples_per_second": 8.309,
+      "eval_steps_per_second": 1.039,
+      "step": 264
+    },
+    {
+      "epoch": 0.7528409090909091,
+      "grad_norm": 0.6283138990402222,
+      "learning_rate": 1.4045298977160058e-05,
+      "loss": 0.1483,
+      "step": 265
+    },
+    {
+      "epoch": 0.7556818181818182,
+      "grad_norm": 0.41013386845588684,
+      "learning_rate": 1.4003859049630943e-05,
+      "loss": 0.0693,
+      "step": 266
+    },
+    {
+      "epoch": 0.7585227272727273,
+      "grad_norm": 0.48505860567092896,
+      "learning_rate": 1.3962337075831583e-05,
+      "loss": 0.1313,
+      "step": 267
+    },
+    {
+      "epoch": 0.7613636363636364,
+      "grad_norm": 0.6941801905632019,
+      "learning_rate": 1.3920733906621861e-05,
+      "loss": 0.1285,
+      "step": 268
+    },
+    {
+      "epoch": 0.7642045454545454,
+      "grad_norm": 0.32076820731163025,
+      "learning_rate": 1.3879050394525525e-05,
+      "loss": 0.0492,
+      "step": 269
+    },
+    {
+      "epoch": 0.7670454545454546,
+      "grad_norm": 0.6292155981063843,
+      "learning_rate": 1.3837287393712666e-05,
+      "loss": 0.1172,
+      "step": 270
+    },
+    {
+      "epoch": 0.7698863636363636,
+      "grad_norm": 0.2482394427061081,
+      "learning_rate": 1.3795445759982262e-05,
+      "loss": 0.0429,
+      "step": 271
+    },
+    {
+      "epoch": 0.7727272727272727,
+      "grad_norm": 0.449253648519516,
+      "learning_rate": 1.3753526350744611e-05,
+      "loss": 0.1198,
+      "step": 272
+    },
+    {
+      "epoch": 0.7755681818181818,
+      "grad_norm": 0.25705787539482117,
+      "learning_rate": 1.3711530025003768e-05,
+      "loss": 0.037,
+      "step": 273
+    },
+    {
+      "epoch": 0.7784090909090909,
+      "grad_norm": 0.23594215512275696,
+      "learning_rate": 1.3669457643339956e-05,
+      "loss": 0.032,
+      "step": 274
+    },
+    {
+      "epoch": 0.78125,
+      "grad_norm": 0.2733691930770874,
+      "learning_rate": 1.3627310067891913e-05,
+      "loss": 0.0397,
+      "step": 275
+    },
+    {
+      "epoch": 0.7840909090909091,
+      "grad_norm": 0.4076026678085327,
+      "learning_rate": 1.3585088162339233e-05,
+      "loss": 0.0651,
+      "step": 276
+    },
+    {
+      "epoch": 0.7869318181818182,
+      "grad_norm": 0.2765565812587738,
+      "learning_rate": 1.3542792791884675e-05,
+      "loss": 0.0237,
+      "step": 277
+    },
+    {
+      "epoch": 0.7897727272727273,
+      "grad_norm": 0.43744075298309326,
+      "learning_rate": 1.3500424823236413e-05,
+      "loss": 0.0622,
+      "step": 278
+    },
+    {
+      "epoch": 0.7926136363636364,
+      "grad_norm": 0.40274515748023987,
+      "learning_rate": 1.3457985124590298e-05,
+      "loss": 0.0662,
+      "step": 279
+    },
+    {
+      "epoch": 0.7954545454545454,
+      "grad_norm": 0.22685614228248596,
+      "learning_rate": 1.3415474565612059e-05,
+      "loss": 0.0365,
+      "step": 280
+    },
+    {
+      "epoch": 0.7982954545454546,
+      "grad_norm": 0.3656117618083954,
+      "learning_rate": 1.337289401741947e-05,
+      "loss": 0.0811,
+      "step": 281
+    },
+    {
+      "epoch": 0.8011363636363636,
+      "grad_norm": 0.4993748664855957,
+      "learning_rate": 1.3330244352564527e-05,
+      "loss": 0.1749,
+      "step": 282
+    },
+    {
+      "epoch": 0.8039772727272727,
+      "grad_norm": 0.4087863266468048,
+      "learning_rate": 1.328752644501553e-05,
+      "loss": 0.0426,
+      "step": 283
+    },
+    {
+      "epoch": 0.8068181818181818,
+      "grad_norm": 0.3817732334136963,
+      "learning_rate": 1.324474117013921e-05,
+      "loss": 0.065,
+      "step": 284
+    },
+    {
+      "epoch": 0.8096590909090909,
+      "grad_norm": 0.625643253326416,
+      "learning_rate": 1.320188940468277e-05,
+      "loss": 0.2149,
+      "step": 285
+    },
+    {
+      "epoch": 0.8125,
+      "grad_norm": 0.4166359305381775,
+      "learning_rate": 1.3158972026755927e-05,
+      "loss": 0.0512,
+      "step": 286
+    },
+    {
+      "epoch": 0.8153409090909091,
+      "grad_norm": 0.4035757780075073,
+      "learning_rate": 1.311598991581291e-05,
+      "loss": 0.0812,
+      "step": 287
+    },
+    {
+      "epoch": 0.8181818181818182,
+      "grad_norm": 0.4624742865562439,
+      "learning_rate": 1.3072943952634446e-05,
+      "loss": 0.0593,
+      "step": 288
+    },
+    {
+      "epoch": 0.8210227272727273,
+      "grad_norm": 0.2985156178474426,
+      "learning_rate": 1.3029835019309713e-05,
+      "loss": 0.0708,
+      "step": 289
+    },
+    {
+      "epoch": 0.8238636363636364,
+      "grad_norm": 0.29609620571136475,
+      "learning_rate": 1.2986663999218264e-05,
+      "loss": 0.0337,
+      "step": 290
+    },
+    {
+      "epoch": 0.8267045454545454,
+      "grad_norm": 0.4906909465789795,
+      "learning_rate": 1.2943431777011903e-05,
+      "loss": 0.1062,
+      "step": 291
+    },
+    {
+      "epoch": 0.8295454545454546,
+      "grad_norm": 0.31944286823272705,
+      "learning_rate": 1.2900139238596598e-05,
+      "loss": 0.064,
+      "step": 292
+    },
+    {
+      "epoch": 0.8323863636363636,
+      "grad_norm": 0.25835123658180237,
+      "learning_rate": 1.2856787271114292e-05,
+      "loss": 0.0362,
+      "step": 293
+    },
+    {
+      "epoch": 0.8352272727272727,
+      "grad_norm": 0.29158154129981995,
+      "learning_rate": 1.2813376762924735e-05,
+      "loss": 0.0254,
+      "step": 294
+    },
+    {
+      "epoch": 0.8380681818181818,
+      "grad_norm": 0.47217270731925964,
+      "learning_rate": 1.2769908603587294e-05,
+      "loss": 0.0675,
+      "step": 295
+    },
+    {
+      "epoch": 0.8409090909090909,
+      "grad_norm": 0.3208584785461426,
+      "learning_rate": 1.272638368384269e-05,
+      "loss": 0.059,
+      "step": 296
+    },
+    {
+      "epoch": 0.84375,
+      "grad_norm": 0.27133041620254517,
+      "learning_rate": 1.268280289559479e-05,
+      "loss": 0.0247,
+      "step": 297
+    },
+    {
+      "epoch": 0.8465909090909091,
+      "grad_norm": 0.4569489359855652,
+      "learning_rate": 1.2639167131892294e-05,
+      "loss": 0.0665,
+      "step": 298
+    },
+    {
+      "epoch": 0.8494318181818182,
+      "grad_norm": 0.40543097257614136,
+      "learning_rate": 1.259547728691045e-05,
+      "loss": 0.0712,
+      "step": 299
+    },
+    {
+      "epoch": 0.8522727272727273,
+      "grad_norm": 0.27440789341926575,
+      "learning_rate": 1.2551734255932728e-05,
+      "loss": 0.0373,
+      "step": 300
+    },
+    {
+      "epoch": 0.8551136363636364,
+      "grad_norm": 0.3247217535972595,
+      "learning_rate": 1.2507938935332478e-05,
+      "loss": 0.0293,
+      "step": 301
+    },
+    {
+      "epoch": 0.8579545454545454,
+      "grad_norm": 0.3218747675418854,
+      "learning_rate": 1.2464092222554554e-05,
+      "loss": 0.0455,
+      "step": 302
+    },
+    {
+      "epoch": 0.8607954545454546,
+      "grad_norm": 0.4028306007385254,
+      "learning_rate": 1.2420195016096934e-05,
+      "loss": 0.0585,
+      "step": 303
+    },
+    {
+      "epoch": 0.8636363636363636,
+      "grad_norm": 0.2925693988800049,
+      "learning_rate": 1.2376248215492297e-05,
+      "loss": 0.0471,
+      "step": 304
+    },
+    {
+      "epoch": 0.8664772727272727,
+      "grad_norm": 0.4908764660358429,
+      "learning_rate": 1.2332252721289594e-05,
+      "loss": 0.1723,
+      "step": 305
+    },
+    {
+      "epoch": 0.8693181818181818,
+      "grad_norm": 0.3072865605354309,
+      "learning_rate": 1.2288209435035605e-05,
+      "loss": 0.035,
+      "step": 306
+    },
+    {
+      "epoch": 0.8721590909090909,
+      "grad_norm": 0.22491200268268585,
+      "learning_rate": 1.2244119259256442e-05,
+      "loss": 0.0272,
+      "step": 307
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 0.28174304962158203,
+      "learning_rate": 1.2199983097439079e-05,
+      "loss": 0.0441,
+      "step": 308
+    },
+    {
+      "epoch": 0.8778409090909091,
+      "grad_norm": 0.34012681245803833,
+      "learning_rate": 1.2155801854012816e-05,
+      "loss": 0.0455,
+      "step": 309
+    },
+    {
+      "epoch": 0.8806818181818182,
+      "grad_norm": 0.49420085549354553,
+      "learning_rate": 1.2111576434330767e-05,
+      "loss": 0.0743,
+      "step": 310
+    },
+    {
+      "epoch": 0.8835227272727273,
+      "grad_norm": 0.17976929247379303,
+      "learning_rate": 1.2067307744651288e-05,
+      "loss": 0.0223,
+      "step": 311
+    },
+    {
+      "epoch": 0.8863636363636364,
+      "grad_norm": 0.49375656247138977,
+      "learning_rate": 1.2022996692119424e-05,
+      "loss": 0.1561,
+      "step": 312
+    },
+    {
+      "epoch": 0.8892045454545454,
+      "grad_norm": 0.36805179715156555,
+      "learning_rate": 1.19786441847483e-05,
+      "loss": 0.0672,
+      "step": 313
+    },
+    {
+      "epoch": 0.8920454545454546,
+      "grad_norm": 0.4191720485687256,
+      "learning_rate": 1.1934251131400532e-05,
+      "loss": 0.0792,
+      "step": 314
+    },
+    {
+      "epoch": 0.8948863636363636,
+      "grad_norm": 0.32242196798324585,
+      "learning_rate": 1.18898184417696e-05,
+      "loss": 0.0444,
+      "step": 315
+    },
+    {
+      "epoch": 0.8977272727272727,
+      "grad_norm": 0.32510390877723694,
+      "learning_rate": 1.1845347026361192e-05,
+      "loss": 0.0397,
+      "step": 316
+    },
+    {
+      "epoch": 0.9005681818181818,
+      "grad_norm": 0.2732970118522644,
+      "learning_rate": 1.1800837796474562e-05,
+      "loss": 0.0514,
+      "step": 317
+    },
+    {
+      "epoch": 0.9034090909090909,
+      "grad_norm": 0.3913041055202484,
+      "learning_rate": 1.1756291664183858e-05,
+      "loss": 0.068,
+      "step": 318
+    },
+    {
+      "epoch": 0.90625,
+      "grad_norm": 0.2852975130081177,
+      "learning_rate": 1.1711709542319412e-05,
+      "loss": 0.0248,
+      "step": 319
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.7706276178359985,
+      "learning_rate": 1.1667092344449053e-05,
+      "loss": 0.2667,
+      "step": 320
+    },
+    {
+      "epoch": 0.9119318181818182,
+      "grad_norm": 0.230291947722435,
+      "learning_rate": 1.1622440984859384e-05,
+      "loss": 0.0334,
+      "step": 321
+    },
+    {
+      "epoch": 0.9147727272727273,
+      "grad_norm": 0.35278844833374023,
+      "learning_rate": 1.1577756378537033e-05,
+      "loss": 0.102,
+      "step": 322
+    },
+    {
+      "epoch": 0.9176136363636364,
+      "grad_norm": 0.2541504502296448,
+      "learning_rate": 1.1533039441149927e-05,
+      "loss": 0.0601,
+      "step": 323
+    },
+    {
+      "epoch": 0.9204545454545454,
+      "grad_norm": 0.3052884340286255,
+      "learning_rate": 1.1488291089028501e-05,
+      "loss": 0.0514,
+      "step": 324
+    },
+    {
+      "epoch": 0.9232954545454546,
+      "grad_norm": 0.20483562350273132,
+      "learning_rate": 1.1443512239146941e-05,
+      "loss": 0.0374,
+      "step": 325
+    },
+    {
+      "epoch": 0.9261363636363636,
+      "grad_norm": 0.36739325523376465,
+      "learning_rate": 1.1398703809104392e-05,
+      "loss": 0.0898,
+      "step": 326
+    },
+    {
+      "epoch": 0.9289772727272727,
+      "grad_norm": 0.7544044256210327,
+      "learning_rate": 1.1353866717106137e-05,
+      "loss": 0.3117,
+      "step": 327
+    },
+    {
+      "epoch": 0.9318181818181818,
+      "grad_norm": 0.23758046329021454,
+      "learning_rate": 1.130900188194481e-05,
+      "loss": 0.0341,
+      "step": 328
+    },
+    {
+      "epoch": 0.9346590909090909,
+      "grad_norm": 0.4377850294113159,
+      "learning_rate": 1.1264110222981535e-05,
+      "loss": 0.0541,
+      "step": 329
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 0.15626086294651031,
+      "learning_rate": 1.1219192660127117e-05,
+      "loss": 0.0093,
+      "step": 330
+    },
+    {
+      "epoch": 0.9403409090909091,
+      "grad_norm": 0.23238646984100342,
+      "learning_rate": 1.1174250113823173e-05,
+      "loss": 0.0342,
+      "step": 331
+    },
+    {
+      "epoch": 0.9431818181818182,
+      "grad_norm": 0.3313692808151245,
+      "learning_rate": 1.1129283505023273e-05,
+      "loss": 0.0618,
+      "step": 332
+    },
+    {
+      "epoch": 0.9460227272727273,
+      "grad_norm": 0.45116209983825684,
+      "learning_rate": 1.1084293755174083e-05,
+      "loss": 0.0915,
+      "step": 333
+    },
+    {
+      "epoch": 0.9488636363636364,
+      "grad_norm": 0.29833459854125977,
+      "learning_rate": 1.1039281786196453e-05,
+      "loss": 0.0479,
+      "step": 334
+    },
+    {
+      "epoch": 0.9517045454545454,
+      "grad_norm": 0.26834267377853394,
+      "learning_rate": 1.0994248520466555e-05,
+      "loss": 0.0414,
+      "step": 335
+    },
+    {
+      "epoch": 0.9545454545454546,
+      "grad_norm": 0.2973485291004181,
+      "learning_rate": 1.0949194880796967e-05,
+      "loss": 0.0458,
+      "step": 336
+    },
+    {
+      "epoch": 0.9573863636363636,
+      "grad_norm": 0.5005751252174377,
+      "learning_rate": 1.0904121790417767e-05,
+      "loss": 0.1774,
+      "step": 337
+    },
+    {
+      "epoch": 0.9602272727272727,
+      "grad_norm": 0.3884448707103729,
+      "learning_rate": 1.085903017295761e-05,
+      "loss": 0.0948,
+      "step": 338
+    },
+    {
+      "epoch": 0.9630681818181818,
+      "grad_norm": 0.4331214725971222,
+      "learning_rate": 1.0813920952424806e-05,
+      "loss": 0.0667,
+      "step": 339
+    },
+    {
+      "epoch": 0.9659090909090909,
+      "grad_norm": 0.35598206520080566,
+      "learning_rate": 1.0768795053188379e-05,
+      "loss": 0.0458,
+      "step": 340
+    },
+    {
+      "epoch": 0.96875,
+      "grad_norm": 0.29866865277290344,
+      "learning_rate": 1.0723653399959142e-05,
+      "loss": 0.0408,
+      "step": 341
+    },
+    {
+      "epoch": 0.9715909090909091,
+      "grad_norm": 0.48330047726631165,
+      "learning_rate": 1.067849691777072e-05,
+      "loss": 0.1626,
+      "step": 342
+    },
+    {
+      "epoch": 0.9744318181818182,
+      "grad_norm": 0.4522784352302551,
+      "learning_rate": 1.063332653196062e-05,
+      "loss": 0.06,
+      "step": 343
+    },
+    {
+      "epoch": 0.9772727272727273,
+      "grad_norm": 0.25423404574394226,
+      "learning_rate": 1.0588143168151257e-05,
+      "loss": 0.0335,
+      "step": 344
+    },
+    {
+      "epoch": 0.9801136363636364,
+      "grad_norm": 0.407180517911911,
+      "learning_rate": 1.0542947752230987e-05,
+      "loss": 0.0609,
+      "step": 345
+    },
+    {
+      "epoch": 0.9829545454545454,
+      "grad_norm": 0.5892329812049866,
+      "learning_rate": 1.049774121033514e-05,
+      "loss": 0.149,
+      "step": 346
+    },
+    {
+      "epoch": 0.9857954545454546,
+      "grad_norm": 0.24134470522403717,
+      "learning_rate": 1.0452524468827028e-05,
+      "loss": 0.0419,
+      "step": 347
+    },
+    {
+      "epoch": 0.9886363636363636,
+      "grad_norm": 0.4492047429084778,
+      "learning_rate": 1.0407298454278983e-05,
+      "loss": 0.0709,
+      "step": 348
+    },
+    {
+      "epoch": 0.9914772727272727,
+      "grad_norm": 0.2484259009361267,
+      "learning_rate": 1.0362064093453348e-05,
+      "loss": 0.0315,
+      "step": 349
+    },
+    {
+      "epoch": 0.9943181818181818,
+      "grad_norm": 0.26170334219932556,
+      "learning_rate": 1.0316822313283504e-05,
+      "loss": 0.0428,
+      "step": 350
+    },
+    {
+      "epoch": 0.9971590909090909,
+      "grad_norm": 0.3009549677371979,
+      "learning_rate": 1.0271574040854863e-05,
+      "loss": 0.054,
+      "step": 351
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.25771793723106384,
+      "learning_rate": 1.0226320203385877e-05,
+      "loss": 0.0322,
+      "step": 352
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.055809713900089264,
+      "eval_runtime": 322.0734,
+      "eval_samples_per_second": 8.368,
+      "eval_steps_per_second": 1.046,
+      "step": 352
+    },
+    {
+      "epoch": 1.0028409090909092,
+      "grad_norm": 0.3883383572101593,
+      "learning_rate": 1.0181061728209034e-05,
+      "loss": 0.0703,
+      "step": 353
+    },
+    {
+      "epoch": 1.0056818181818181,
+      "grad_norm": 0.35572555661201477,
+      "learning_rate": 1.0135799542751861e-05,
+      "loss": 0.0373,
+      "step": 354
+    },
+    {
+      "epoch": 1.0085227272727273,
+      "grad_norm": 0.29604148864746094,
+      "learning_rate": 1.0090534574517907e-05,
+      "loss": 0.0499,
+      "step": 355
+    },
+    {
+      "epoch": 1.0113636363636365,
+      "grad_norm": 0.3191431760787964,
+      "learning_rate": 1.0045267751067758e-05,
+      "loss": 0.047,
+      "step": 356
+    },
+    {
+      "epoch": 1.0142045454545454,
+      "grad_norm": 0.2044990211725235,
+      "learning_rate": 1e-05,
+      "loss": 0.0254,
+      "step": 357
+    },
+    {
+      "epoch": 1.0028409090909092,
+      "grad_norm": 0.22580191493034363,
+      "learning_rate": 9.954732248932243e-06,
+      "loss": 0.0359,
+      "step": 358
+    },
+    {
+      "epoch": 1.0056818181818181,
+      "grad_norm": 0.329467236995697,
+      "learning_rate": 9.909465425482093e-06,
+      "loss": 0.0541,
+      "step": 359
+    },
+    {
+      "epoch": 1.0085227272727273,
+      "grad_norm": 0.6182022094726562,
+      "learning_rate": 9.864200457248144e-06,
+      "loss": 0.0741,
+      "step": 360
+    },
+    {
+      "epoch": 1.0113636363636365,
+      "grad_norm": 0.3866288363933563,
+      "learning_rate": 9.81893827179097e-06,
+      "loss": 0.0502,
+      "step": 361
+    },
+    {
+      "epoch": 1.0142045454545454,
+      "grad_norm": 0.3960239291191101,
+      "learning_rate": 9.773679796614124e-06,
+      "loss": 0.0317,
+      "step": 362
+    },
+    {
+      "epoch": 1.0170454545454546,
+      "grad_norm": 0.27909645438194275,
+      "learning_rate": 9.728425959145139e-06,
+      "loss": 0.0365,
+      "step": 363
+    },
+    {
+      "epoch": 1.0198863636363635,
+      "grad_norm": 0.4127347469329834,
+      "learning_rate": 9.683177686716501e-06,
+      "loss": 0.0492,
+      "step": 364
+    },
+    {
+      "epoch": 1.0227272727272727,
+      "grad_norm": 0.44473913311958313,
+      "learning_rate": 9.637935906546655e-06,
+      "loss": 0.0533,
+      "step": 365
+    },
+    {
+      "epoch": 1.0255681818181819,
+      "grad_norm": 0.3825552463531494,
+      "learning_rate": 9.592701545721022e-06,
+      "loss": 0.036,
+      "step": 366
+    },
+    {
+      "epoch": 1.0284090909090908,
+      "grad_norm": 0.2615335285663605,
+      "learning_rate": 9.547475531172973e-06,
+      "loss": 0.0386,
+      "step": 367
+    },
+    {
+      "epoch": 1.03125,
+      "grad_norm": 0.44840914011001587,
+      "learning_rate": 9.502258789664865e-06,
+      "loss": 0.0967,
+      "step": 368
+    },
+    {
+      "epoch": 1.0340909090909092,
+      "grad_norm": 0.5567358136177063,
+      "learning_rate": 9.457052247769018e-06,
+      "loss": 0.0486,
+      "step": 369
+    },
+    {
+      "epoch": 1.0369318181818181,
+      "grad_norm": 0.33771201968193054,
+      "learning_rate": 9.411856831848745e-06,
+      "loss": 0.0439,
+      "step": 370
+    },
+    {
+      "epoch": 1.0397727272727273,
+      "grad_norm": 0.24962462484836578,
+      "learning_rate": 9.366673468039384e-06,
+      "loss": 0.0279,
+      "step": 371
+    },
+    {
+      "epoch": 1.0426136363636365,
+      "grad_norm": 0.44227010011672974,
+      "learning_rate": 9.321503082229281e-06,
+      "loss": 0.0597,
+      "step": 372
+    },
+    {
+      "epoch": 1.0454545454545454,
+      "grad_norm": 0.33881664276123047,
+      "learning_rate": 9.276346600040863e-06,
+      "loss": 0.037,
+      "step": 373
+    },
+    {
+      "epoch": 1.0482954545454546,
+      "grad_norm": 0.2290397733449936,
+      "learning_rate": 9.231204946811624e-06,
+      "loss": 0.0276,
+      "step": 374
+    },
+    {
+      "epoch": 1.0511363636363635,
+      "grad_norm": 0.3605166971683502,
+      "learning_rate": 9.186079047575197e-06,
+      "loss": 0.0375,
+      "step": 375
+    },
+    {
+      "epoch": 1.0539772727272727,
+      "grad_norm": 0.5530329942703247,
+      "learning_rate": 9.140969827042392e-06,
+      "loss": 0.0787,
+      "step": 376
+    },
+    {
+      "epoch": 1.0568181818181819,
+      "grad_norm": 0.14625893533229828,
+      "learning_rate": 9.095878209582237e-06,
+      "loss": 0.014,
+      "step": 377
+    },
+    {
+      "epoch": 1.0596590909090908,
+      "grad_norm": 0.12633948028087616,
+      "learning_rate": 9.050805119203035e-06,
+      "loss": 0.0149,
+      "step": 378
+    },
+    {
+      "epoch": 1.0625,
+      "grad_norm": 0.506729781627655,
+      "learning_rate": 9.00575147953345e-06,
+      "loss": 0.0723,
+      "step": 379
+    },
+    {
+      "epoch": 1.0653409090909092,
+      "grad_norm": 0.3099755644798279,
+      "learning_rate": 8.96071821380355e-06,
+      "loss": 0.0293,
+      "step": 380
+    },
+    {
+      "epoch": 1.0681818181818181,
+      "grad_norm": 0.19945262372493744,
+      "learning_rate": 8.91570624482592e-06,
+      "loss": 0.0231,
+      "step": 381
+    },
+    {
+      "epoch": 1.0710227272727273,
+      "grad_norm": 0.3164744973182678,
+      "learning_rate": 8.87071649497673e-06,
+      "loss": 0.0444,
+      "step": 382
+    },
+    {
+      "epoch": 1.0738636363636365,
+      "grad_norm": 0.3066563308238983,
+      "learning_rate": 8.825749886176832e-06,
+      "loss": 0.0383,
+      "step": 383
+    },
+    {
+      "epoch": 1.0767045454545454,
+      "grad_norm": 0.16921481490135193,
+      "learning_rate": 8.780807339872886e-06,
+      "loss": 0.0194,
+      "step": 384
+    },
+    {
+      "epoch": 1.0795454545454546,
+      "grad_norm": 0.440748929977417,
+      "learning_rate": 8.735889777018466e-06,
+      "loss": 0.0487,
+      "step": 385
+    },
+    {
+      "epoch": 1.0823863636363635,
+      "grad_norm": 0.2782658040523529,
+      "learning_rate": 8.690998118055193e-06,
+      "loss": 0.0333,
+      "step": 386
+    },
+    {
+      "epoch": 1.0852272727272727,
+      "grad_norm": 0.22277198731899261,
+      "learning_rate": 8.646133282893865e-06,
+      "loss": 0.0217,
+      "step": 387
+    },
+    {
+      "epoch": 1.0880681818181819,
+      "grad_norm": 0.21947693824768066,
+      "learning_rate": 8.601296190895611e-06,
+      "loss": 0.0263,
+      "step": 388
+    },
+    {
+      "epoch": 1.0909090909090908,
+      "grad_norm": 0.1716959923505783,
+      "learning_rate": 8.55648776085306e-06,
+      "loss": 0.017,
+      "step": 389
+    },
+    {
+      "epoch": 1.09375,
+      "grad_norm": 0.3042616546154022,
+      "learning_rate": 8.511708910971506e-06,
+      "loss": 0.0392,
+      "step": 390
+    },
+    {
+      "epoch": 1.0965909090909092,
+      "grad_norm": 0.38685595989227295,
+      "learning_rate": 8.466960558850078e-06,
+      "loss": 0.0438,
+      "step": 391
+    },
+    {
+      "epoch": 1.0994318181818181,
+      "grad_norm": 0.5748195052146912,
+      "learning_rate": 8.42224362146297e-06,
+      "loss": 0.0592,
+      "step": 392
+    },
+    {
+      "epoch": 1.1022727272727273,
+      "grad_norm": 0.4325203597545624,
+      "learning_rate": 8.37755901514062e-06,
+      "loss": 0.0682,
+      "step": 393
+    },
+    {
+      "epoch": 1.1051136363636365,
+      "grad_norm": 0.21434345841407776,
+      "learning_rate": 8.332907655550948e-06,
+      "loss": 0.0239,
+      "step": 394
+    },
+    {
+      "epoch": 1.1079545454545454,
+      "grad_norm": 0.243415966629982,
+      "learning_rate": 8.288290457680593e-06,
+      "loss": 0.0207,
+      "step": 395
+    },
+    {
+      "epoch": 1.1107954545454546,
+      "grad_norm": 0.4860462546348572,
+      "learning_rate": 8.243708335816145e-06,
+      "loss": 0.0563,
+      "step": 396
+    },
+    {
+      "epoch": 1.1136363636363635,
+      "grad_norm": 0.37320631742477417,
+      "learning_rate": 8.19916220352544e-06,
+      "loss": 0.0485,
+      "step": 397
+    },
+    {
+      "epoch": 1.1164772727272727,
+      "grad_norm": 0.6365073323249817,
+      "learning_rate": 8.15465297363881e-06,
+      "loss": 0.1069,
+      "step": 398
+    },
+    {
+      "epoch": 1.1193181818181819,
+      "grad_norm": 0.2253241240978241,
+      "learning_rate": 8.110181558230405e-06,
+      "loss": 0.0292,
+      "step": 399
+    },
+    {
+      "epoch": 1.1221590909090908,
+      "grad_norm": 0.32560545206069946,
+      "learning_rate": 8.065748868599471e-06,
+      "loss": 0.0333,
+      "step": 400
+    },
+    {
+      "epoch": 1.125,
+      "grad_norm": 0.1719602346420288,
+      "learning_rate": 8.021355815251703e-06,
+      "loss": 0.0148,
+      "step": 401
+    },
+    {
+      "epoch": 1.1278409090909092,
+      "grad_norm": 0.27399298548698425,
+      "learning_rate": 7.977003307880579e-06,
+      "loss": 0.039,
+      "step": 402
+    },
+    {
+      "epoch": 1.1306818181818181,
+      "grad_norm": 0.12785029411315918,
+      "learning_rate": 7.932692255348711e-06,
+      "loss": 0.0123,
+      "step": 403
+    },
+    {
+      "epoch": 1.1335227272727273,
+      "grad_norm": 0.5757378935813904,
+      "learning_rate": 7.888423565669236e-06,
+      "loss": 0.0443,
+      "step": 404
+    },
+    {
+      "epoch": 1.1363636363636362,
+      "grad_norm": 0.1946515142917633,
+      "learning_rate": 7.844198145987187e-06,
+      "loss": 0.0159,
+      "step": 405
+    },
+    {
+      "epoch": 1.1392045454545454,
+      "grad_norm": 0.1719973236322403,
+      "learning_rate": 7.800016902560924e-06,
+      "loss": 0.0244,
+      "step": 406
+    },
+    {
+      "epoch": 1.1420454545454546,
+      "grad_norm": 0.6020510196685791,
+      "learning_rate": 7.75588074074356e-06,
+      "loss": 0.1468,
+      "step": 407
+    },
+    {
+      "epoch": 1.1448863636363638,
+      "grad_norm": 0.1654803454875946,
+      "learning_rate": 7.7117905649644e-06,
+      "loss": 0.0137,
+      "step": 408
+    },
+    {
+      "epoch": 1.1477272727272727,
+      "grad_norm": 0.20969092845916748,
+      "learning_rate": 7.667747278710407e-06,
+      "loss": 0.0209,
+      "step": 409
+    },
+    {
+      "epoch": 1.1505681818181819,
+      "grad_norm": 0.09986022859811783,
+      "learning_rate": 7.623751784507707e-06,
+      "loss": 0.0082,
+      "step": 410
+    },
+    {
+      "epoch": 1.1534090909090908,
+      "grad_norm": 0.4177515506744385,
+      "learning_rate": 7.579804983903067e-06,
+      "loss": 0.0481,
+      "step": 411
+    },
+    {
+      "epoch": 1.15625,
+      "grad_norm": 0.21432848274707794,
+      "learning_rate": 7.535907777445449e-06,
+      "loss": 0.0222,
+      "step": 412
+    },
+    {
+      "epoch": 1.1590909090909092,
+      "grad_norm": 0.15598714351654053,
+      "learning_rate": 7.492061064667526e-06,
+      "loss": 0.0212,
+      "step": 413
+    },
+    {
+      "epoch": 1.1619318181818181,
+      "grad_norm": 0.17685821652412415,
+      "learning_rate": 7.448265744067275e-06,
+      "loss": 0.0108,
+      "step": 414
+    },
+    {
+      "epoch": 1.1647727272727273,
+      "grad_norm": 0.1826607584953308,
+      "learning_rate": 7.404522713089555e-06,
+      "loss": 0.014,
+      "step": 415
+    },
+    {
+      "epoch": 1.1676136363636362,
+      "grad_norm": 0.4454158842563629,
+      "learning_rate": 7.360832868107708e-06,
+      "loss": 0.0268,
+      "step": 416
+    },
+    {
+      "epoch": 1.1704545454545454,
+      "grad_norm": 0.13606494665145874,
+      "learning_rate": 7.3171971044052135e-06,
+      "loss": 0.014,
+      "step": 417
+    },
+    {
+      "epoch": 1.1732954545454546,
+      "grad_norm": 0.20646145939826965,
+      "learning_rate": 7.273616316157312e-06,
+      "loss": 0.0205,
+      "step": 418
+    },
+    {
+      "epoch": 1.1761363636363638,
+      "grad_norm": 0.2317519336938858,
+      "learning_rate": 7.230091396412711e-06,
+      "loss": 0.0292,
+      "step": 419
+    },
+    {
+      "epoch": 1.1789772727272727,
+      "grad_norm": 0.40039902925491333,
+      "learning_rate": 7.186623237075265e-06,
+      "loss": 0.0699,
+      "step": 420
+    },
+    {
+      "epoch": 1.1818181818181819,
+      "grad_norm": 0.377642959356308,
+      "learning_rate": 7.143212728885714e-06,
+      "loss": 0.0293,
+      "step": 421
+    },
+    {
+      "epoch": 1.1846590909090908,
+      "grad_norm": 0.29831114411354065,
+      "learning_rate": 7.0998607614034035e-06,
+      "loss": 0.0462,
+      "step": 422
+    },
+    {
+      "epoch": 1.1875,
+      "grad_norm": 0.23558694124221802,
+      "learning_rate": 7.056568222988098e-06,
+      "loss": 0.0199,
+      "step": 423
+    },
+    {
+      "epoch": 1.1903409090909092,
+      "grad_norm": 0.5152977108955383,
+      "learning_rate": 7.013336000781739e-06,
+      "loss": 0.0689,
+      "step": 424
+    },
+    {
+      "epoch": 1.1931818181818181,
+      "grad_norm": 0.3257889747619629,
+      "learning_rate": 6.970164980690286e-06,
+      "loss": 0.0272,
+      "step": 425
+    },
+    {
+      "epoch": 1.1960227272727273,
+      "grad_norm": 0.2875789999961853,
+      "learning_rate": 6.927056047365558e-06,
+      "loss": 0.0322,
+      "step": 426
+    },
+    {
+      "epoch": 1.1988636363636362,
+      "grad_norm": 0.3039776086807251,
+      "learning_rate": 6.884010084187094e-06,
+      "loss": 0.0305,
+      "step": 427
+    },
+    {
+      "epoch": 1.2017045454545454,
+      "grad_norm": 0.16907234489917755,
+      "learning_rate": 6.841027973244077e-06,
+      "loss": 0.0104,
+      "step": 428
+    },
+    {
+      "epoch": 1.2045454545454546,
+      "grad_norm": 0.23682954907417297,
+      "learning_rate": 6.798110595317229e-06,
+      "loss": 0.0307,
+      "step": 429
+    },
+    {
+      "epoch": 1.2073863636363638,
+      "grad_norm": 0.6338050961494446,
+      "learning_rate": 6.755258829860791e-06,
+      "loss": 0.1344,
+      "step": 430
+    },
+    {
+      "epoch": 1.2102272727272727,
+      "grad_norm": 0.22749580442905426,
+      "learning_rate": 6.712473554984472e-06,
+      "loss": 0.0176,
+      "step": 431
+    },
+    {
+      "epoch": 1.2130681818181819,
+      "grad_norm": 0.18605318665504456,
+      "learning_rate": 6.669755647435474e-06,
+      "loss": 0.0202,
+      "step": 432
+    },
+    {
+      "epoch": 1.2159090909090908,
+      "grad_norm": 0.36711132526397705,
+      "learning_rate": 6.627105982580529e-06,
+      "loss": 0.0524,
+      "step": 433
+    },
+    {
+      "epoch": 1.21875,
+      "grad_norm": 0.4177369177341461,
+      "learning_rate": 6.584525434387945e-06,
+      "loss": 0.0464,
+      "step": 434
+    },
+    {
+      "epoch": 1.2215909090909092,
+      "grad_norm": 0.285185307264328,
+      "learning_rate": 6.542014875409703e-06,
+      "loss": 0.0278,
+      "step": 435
+    },
+    {
+      "epoch": 1.2244318181818181,
+      "grad_norm": 0.4436541795730591,
+      "learning_rate": 6.499575176763591e-06,
+      "loss": 0.0414,
+      "step": 436
+    },
+    {
+      "epoch": 1.2272727272727273,
+      "grad_norm": 0.33370596170425415,
+      "learning_rate": 6.457207208115328e-06,
+      "loss": 0.0299,
+      "step": 437
+    },
+    {
+      "epoch": 1.2301136363636362,
+      "grad_norm": 0.32635894417762756,
+      "learning_rate": 6.414911837660768e-06,
+      "loss": 0.0343,
+      "step": 438
+    },
+    {
+      "epoch": 1.2329545454545454,
+      "grad_norm": 0.2855762243270874,
+      "learning_rate": 6.372689932108092e-06,
+      "loss": 0.0526,
+      "step": 439
+    },
+    {
+      "epoch": 1.2357954545454546,
+      "grad_norm": 0.194916769862175,
+      "learning_rate": 6.330542356660047e-06,
+      "loss": 0.0186,
+      "step": 440
+    },
+    {
+      "epoch": 1.2357954545454546,
+      "eval_loss": 0.052834562957286835,
+      "eval_runtime": 321.2924,
+      "eval_samples_per_second": 8.388,
+      "eval_steps_per_second": 1.049,
+      "step": 440
+    },
+    {
+      "epoch": 1.2386363636363638,
+      "grad_norm": 0.2691531777381897,
+      "learning_rate": 6.288469974996234e-06,
+      "loss": 0.0363,
+      "step": 441
+    },
+    {
+      "epoch": 1.2414772727272727,
+      "grad_norm": 0.20350024104118347,
+      "learning_rate": 6.246473649255391e-06,
+      "loss": 0.0318,
+      "step": 442
+    },
+    {
+      "epoch": 1.2443181818181819,
+      "grad_norm": 0.36513552069664,
+      "learning_rate": 6.204554240017742e-06,
+      "loss": 0.035,
+      "step": 443
+    },
+    {
+      "epoch": 1.2471590909090908,
+      "grad_norm": 0.21137046813964844,
+      "learning_rate": 6.162712606287335e-06,
+      "loss": 0.0224,
+      "step": 444
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.29429683089256287,
+      "learning_rate": 6.120949605474479e-06,
+      "loss": 0.0308,
+      "step": 445
+    },
+    {
+      "epoch": 1.2528409090909092,
+      "grad_norm": 0.26205527782440186,
+      "learning_rate": 6.079266093378138e-06,
+      "loss": 0.0453,
+      "step": 446
+    },
+    {
+      "epoch": 1.2556818181818181,
+      "grad_norm": 0.26017487049102783,
+      "learning_rate": 6.037662924168419e-06,
+      "loss": 0.031,
+      "step": 447
+    },
+    {
+      "epoch": 1.2585227272727273,
+      "grad_norm": 0.7420312166213989,
+      "learning_rate": 5.9961409503690605e-06,
+      "loss": 0.1057,
+      "step": 448
+    },
+    {
+      "epoch": 1.2613636363636362,
+      "grad_norm": 0.18516167998313904,
+      "learning_rate": 5.954701022839944e-06,
+      "loss": 0.0179,
+      "step": 449
+    },
+    {
+      "epoch": 1.2642045454545454,
+      "grad_norm": 0.2416505217552185,
+      "learning_rate": 5.913343990759695e-06,
+      "loss": 0.0289,
+      "step": 450
+    },
+    {
+      "epoch": 1.2670454545454546,
+      "grad_norm": 0.3561277389526367,
+      "learning_rate": 5.872070701608251e-06,
+      "loss": 0.0559,
+      "step": 451
+    },
+    {
+      "epoch": 1.2698863636363638,
+      "grad_norm": 0.3896782398223877,
+      "learning_rate": 5.830882001149517e-06,
+      "loss": 0.0659,
+      "step": 452
+    },
+    {
+      "epoch": 1.2727272727272727,
+      "grad_norm": 0.26022788882255554,
+      "learning_rate": 5.789778733414004e-06,
+      "loss": 0.0337,
+      "step": 453
+    },
+    {
+      "epoch": 1.2755681818181819,
+      "grad_norm": 0.17623814940452576,
+      "learning_rate": 5.748761740681573e-06,
+      "loss": 0.0171,
+      "step": 454
+    },
+    {
+      "epoch": 1.2784090909090908,
+      "grad_norm": 0.20112119615077972,
+      "learning_rate": 5.707831863464146e-06,
+      "loss": 0.0226,
+      "step": 455
+    },
+    {
+      "epoch": 1.28125,
+      "grad_norm": 0.26868581771850586,
+      "learning_rate": 5.666989940488496e-06,
+      "loss": 0.0281,
+      "step": 456
+    },
+    {
+      "epoch": 1.2840909090909092,
+      "grad_norm": 0.17673923075199127,
+      "learning_rate": 5.626236808679051e-06,
+      "loss": 0.0116,
+      "step": 457
+    },
+    {
+      "epoch": 1.2869318181818181,
+      "grad_norm": 0.42206647992134094,
+      "learning_rate": 5.585573303140741e-06,
+      "loss": 0.0568,
+      "step": 458
+    },
+    {
+      "epoch": 1.2897727272727273,
+      "grad_norm": 0.18107818067073822,
+      "learning_rate": 5.54500025714191e-06,
+      "loss": 0.0224,
+      "step": 459
+    },
+    {
+      "epoch": 1.2926136363636362,
+      "grad_norm": 0.34270885586738586,
+      "learning_rate": 5.504518502097212e-06,
+      "loss": 0.0427,
+      "step": 460
+    },
+    {
+      "epoch": 1.2954545454545454,
+      "grad_norm": 0.3076038956642151,
+      "learning_rate": 5.464128867550593e-06,
+      "loss": 0.0414,
+      "step": 461
+    },
+    {
+      "epoch": 1.2982954545454546,
+      "grad_norm": 0.23357230424880981,
+      "learning_rate": 5.423832181158274e-06,
+      "loss": 0.0226,
+      "step": 462
+    },
+    {
+      "epoch": 1.3011363636363638,
+      "grad_norm": 0.41625964641571045,
+      "learning_rate": 5.383629268671804e-06,
+      "loss": 0.0428,
+      "step": 463
+    },
+    {
+      "epoch": 1.3039772727272727,
+      "grad_norm": 0.34053537249565125,
+      "learning_rate": 5.34352095392114e-06,
+      "loss": 0.0383,
+      "step": 464
+    },
+    {
+      "epoch": 1.3068181818181819,
+      "grad_norm": 0.313376784324646,
+      "learning_rate": 5.3035080587977665e-06,
+      "loss": 0.0429,
+      "step": 465
+    },
+    {
+      "epoch": 1.3096590909090908,
+      "grad_norm": 0.28910282254219055,
+      "learning_rate": 5.263591403237831e-06,
+      "loss": 0.0323,
+      "step": 466
+    },
+    {
+      "epoch": 1.3125,
+      "grad_norm": 0.3461354076862335,
+      "learning_rate": 5.22377180520538e-06,
+      "loss": 0.0589,
+      "step": 467
+    },
+    {
+      "epoch": 1.3153409090909092,
+      "grad_norm": 0.36325037479400635,
+      "learning_rate": 5.184050080675558e-06,
+      "loss": 0.0359,
+      "step": 468
+    },
+    {
+      "epoch": 1.3181818181818181,
+      "grad_norm": 0.20179492235183716,
+      "learning_rate": 5.144427043617919e-06,
+      "loss": 0.0259,
+      "step": 469
+    },
+    {
+      "epoch": 1.3210227272727273,
+      "grad_norm": 0.18621599674224854,
+      "learning_rate": 5.10490350597973e-06,
+      "loss": 0.0156,
+      "step": 470
+    },
+    {
+      "epoch": 1.3238636363636362,
+      "grad_norm": 0.31812962889671326,
+      "learning_rate": 5.06548027766933e-06,
+      "loss": 0.0362,
+      "step": 471
+    },
+    {
+      "epoch": 1.3267045454545454,
+      "grad_norm": 0.19596129655838013,
+      "learning_rate": 5.026158166539548e-06,
+      "loss": 0.0239,
+      "step": 472
+    },
+    {
+      "epoch": 1.3295454545454546,
+      "grad_norm": 0.6086770296096802,
+      "learning_rate": 4.986937978371132e-06,
+      "loss": 0.0502,
+      "step": 473
+    },
+    {
+      "epoch": 1.3323863636363638,
+      "grad_norm": 0.42738351225852966,
+      "learning_rate": 4.9478205168562535e-06,
+      "loss": 0.0484,
+      "step": 474
+    },
+    {
+      "epoch": 1.3352272727272727,
+      "grad_norm": 0.2786358594894409,
+      "learning_rate": 4.908806583582021e-06,
+      "loss": 0.0336,
+      "step": 475
+    },
+    {
+      "epoch": 1.3380681818181819,
+      "grad_norm": 0.202577605843544,
+      "learning_rate": 4.869896978014071e-06,
+      "loss": 0.0286,
+      "step": 476
+    },
+    {
+      "epoch": 1.3409090909090908,
+      "grad_norm": 0.42583322525024414,
+      "learning_rate": 4.83109249748018e-06,
+      "loss": 0.0843,
+      "step": 477
+    },
+    {
+      "epoch": 1.34375,
+      "grad_norm": 0.6603618264198303,
+      "learning_rate": 4.792393937153914e-06,
+      "loss": 0.0674,
+      "step": 478
+    },
+    {
+      "epoch": 1.3465909090909092,
+      "grad_norm": 0.43137821555137634,
+      "learning_rate": 4.753802090038344e-06,
+      "loss": 0.0516,
+      "step": 479
+    },
+    {
+      "epoch": 1.3494318181818181,
+      "grad_norm": 0.5275350213050842,
+      "learning_rate": 4.7153177469498045e-06,
+      "loss": 0.0726,
+      "step": 480
+    },
+    {
+      "epoch": 1.3522727272727273,
+      "grad_norm": 0.35748714208602905,
+      "learning_rate": 4.676941696501674e-06,
+      "loss": 0.0453,
+      "step": 481
+    },
+    {
+      "epoch": 1.3551136363636362,
+      "grad_norm": 0.5304725766181946,
+      "learning_rate": 4.6386747250882226e-06,
+      "loss": 0.0431,
+      "step": 482
+    },
+    {
+      "epoch": 1.3579545454545454,
+      "grad_norm": 0.2420758306980133,
+      "learning_rate": 4.600517616868491e-06,
+      "loss": 0.0284,
+      "step": 483
+    },
+    {
+      "epoch": 1.3607954545454546,
+      "grad_norm": 0.5091615319252014,
+      "learning_rate": 4.562471153750221e-06,
+      "loss": 0.1026,
+      "step": 484
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 0.41949352622032166,
+      "learning_rate": 4.52453611537385e-06,
+      "loss": 0.0516,
+      "step": 485
+    },
+    {
+      "epoch": 1.3664772727272727,
+      "grad_norm": 0.35782483220100403,
+      "learning_rate": 4.486713279096515e-06,
+      "loss": 0.0505,
+      "step": 486
+    },
+    {
+      "epoch": 1.3693181818181819,
+      "grad_norm": 0.4138379395008087,
+      "learning_rate": 4.449003419976133e-06,
+      "loss": 0.0789,
+      "step": 487
+    },
+    {
+      "epoch": 1.3721590909090908,
+      "grad_norm": 0.26523318886756897,
+      "learning_rate": 4.411407310755513e-06,
+      "loss": 0.0311,
+      "step": 488
+    },
+    {
+      "epoch": 1.375,
+      "grad_norm": 0.46949058771133423,
+      "learning_rate": 4.373925721846519e-06,
+      "loss": 0.0293,
+      "step": 489
+    },
+    {
+      "epoch": 1.3778409090909092,
+      "grad_norm": 0.381064772605896,
+      "learning_rate": 4.336559421314298e-06,
+      "loss": 0.0525,
+      "step": 490
+    },
+    {
+      "epoch": 1.3806818181818181,
+      "grad_norm": 0.3074735701084137,
+      "learning_rate": 4.29930917486153e-06,
+      "loss": 0.0216,
+      "step": 491
+    },
+    {
+      "epoch": 1.3835227272727273,
+      "grad_norm": 0.21338820457458496,
+      "learning_rate": 4.262175745812729e-06,
+      "loss": 0.0239,
+      "step": 492
+    },
+    {
+      "epoch": 1.3863636363636362,
+      "grad_norm": 0.31537333130836487,
+      "learning_rate": 4.225159895098623e-06,
+      "loss": 0.0412,
+      "step": 493
+    },
+    {
+      "epoch": 1.3892045454545454,
+      "grad_norm": 0.2787759006023407,
+      "learning_rate": 4.1882623812405345e-06,
+      "loss": 0.0275,
+      "step": 494
+    },
+    {
+      "epoch": 1.3920454545454546,
+      "grad_norm": 0.2170637845993042,
+      "learning_rate": 4.151483960334862e-06,
+      "loss": 0.0244,
+      "step": 495
+    },
+    {
+      "epoch": 1.3948863636363638,
+      "grad_norm": 0.3851534426212311,
+      "learning_rate": 4.1148253860375764e-06,
+      "loss": 0.0426,
+      "step": 496
+    },
+    {
+      "epoch": 1.3977272727272727,
+      "grad_norm": 0.15403486788272858,
+      "learning_rate": 4.078287409548764e-06,
+      "loss": 0.0171,
+      "step": 497
+    },
+    {
+      "epoch": 1.4005681818181819,
+      "grad_norm": 0.3017263114452362,
+      "learning_rate": 4.0418707795972575e-06,
+      "loss": 0.0274,
+      "step": 498
+    },
+    {
+      "epoch": 1.4034090909090908,
+      "grad_norm": 0.3194883465766907,
+      "learning_rate": 4.005576242425272e-06,
+      "loss": 0.0374,
+      "step": 499
+    },
+    {
+      "epoch": 1.40625,
+      "grad_norm": 0.428733766078949,
+      "learning_rate": 3.969404541773132e-06,
+      "loss": 0.0768,
+      "step": 500
+    },
+    {
+      "epoch": 1.4090909090909092,
+      "grad_norm": 0.25702938437461853,
+      "learning_rate": 3.933356418864008e-06,
+      "loss": 0.0256,
+      "step": 501
+    },
+    {
+      "epoch": 1.4119318181818181,
+      "grad_norm": 0.15047182142734528,
+      "learning_rate": 3.897432612388752e-06,
+      "loss": 0.0136,
+      "step": 502
+    },
+    {
+      "epoch": 1.4147727272727273,
+      "grad_norm": 0.14957940578460693,
+      "learning_rate": 3.861633858490746e-06,
+      "loss": 0.0188,
+      "step": 503
+    },
+    {
+      "epoch": 1.4176136363636362,
+      "grad_norm": 0.16923752427101135,
+      "learning_rate": 3.8259608907508105e-06,
+      "loss": 0.0169,
+      "step": 504
+    },
+    {
+      "epoch": 1.4204545454545454,
+      "grad_norm": 0.23884394764900208,
+      "learning_rate": 3.7904144401721976e-06,
+      "loss": 0.0301,
+      "step": 505
+    },
+    {
+      "epoch": 1.4232954545454546,
+      "grad_norm": 0.3131102919578552,
+      "learning_rate": 3.7549952351655795e-06,
+      "loss": 0.0334,
+      "step": 506
+    },
+    {
+      "epoch": 1.4261363636363638,
+      "grad_norm": 0.3991200923919678,
+      "learning_rate": 3.719704001534149e-06,
+      "loss": 0.0712,
+      "step": 507
+    },
+    {
+      "epoch": 1.4289772727272727,
+      "grad_norm": 0.3849276602268219,
+      "learning_rate": 3.6845414624587326e-06,
+      "loss": 0.0476,
+      "step": 508
+    },
+    {
+      "epoch": 1.4318181818181819,
+      "grad_norm": 0.3225853145122528,
+      "learning_rate": 3.649508338482972e-06,
+      "loss": 0.0515,
+      "step": 509
+    },
+    {
+      "epoch": 1.4346590909090908,
+      "grad_norm": 0.3104516565799713,
+      "learning_rate": 3.6146053474985565e-06,
+      "loss": 0.0436,
+      "step": 510
+    },
+    {
+      "epoch": 1.4375,
+      "grad_norm": 0.3936445415019989,
+      "learning_rate": 3.579833204730525e-06,
+      "loss": 0.0487,
+      "step": 511
+    },
+    {
+      "epoch": 1.4403409090909092,
+      "grad_norm": 0.21598631143569946,
+      "learning_rate": 3.5451926227225997e-06,
+      "loss": 0.0216,
+      "step": 512
+    },
+    {
+      "epoch": 1.4431818181818181,
+      "grad_norm": 0.26715341210365295,
+      "learning_rate": 3.5106843113225854e-06,
+      "loss": 0.0269,
+      "step": 513
+    },
+    {
+      "epoch": 1.4460227272727273,
+      "grad_norm": 0.19402551651000977,
+      "learning_rate": 3.4763089776678206e-06,
+      "loss": 0.0212,
+      "step": 514
+    },
+    {
+      "epoch": 1.4488636363636362,
+      "grad_norm": 0.3250589370727539,
+      "learning_rate": 3.4420673261706906e-06,
+      "loss": 0.0473,
+      "step": 515
+    },
+    {
+      "epoch": 1.4517045454545454,
+      "grad_norm": 0.2799818813800812,
+      "learning_rate": 3.4079600585041993e-06,
+      "loss": 0.0313,
+      "step": 516
+    },
+    {
+      "epoch": 1.4545454545454546,
+      "grad_norm": 0.11718513816595078,
+      "learning_rate": 3.3739878735875797e-06,
+      "loss": 0.0127,
+      "step": 517
+    },
+    {
+      "epoch": 1.4573863636363638,
+      "grad_norm": 0.25607946515083313,
+      "learning_rate": 3.3401514675719815e-06,
+      "loss": 0.0307,
+      "step": 518
+    },
+    {
+      "epoch": 1.4602272727272727,
+      "grad_norm": 0.2513953149318695,
+      "learning_rate": 3.3064515338261937e-06,
+      "loss": 0.0334,
+      "step": 519
+    },
+    {
+      "epoch": 1.4630681818181819,
+      "grad_norm": 0.36698806285858154,
+      "learning_rate": 3.2728887629224415e-06,
+      "loss": 0.0558,
+      "step": 520
+    },
+    {
+      "epoch": 1.4659090909090908,
+      "grad_norm": 0.46339672803878784,
+      "learning_rate": 3.239463842622247e-06,
+      "loss": 0.0619,
+      "step": 521
+    },
+    {
+      "epoch": 1.46875,
+      "grad_norm": 0.37225818634033203,
+      "learning_rate": 3.206177457862321e-06,
+      "loss": 0.0443,
+      "step": 522
+    },
+    {
+      "epoch": 1.4715909090909092,
+      "grad_norm": 0.2377389669418335,
+      "learning_rate": 3.173030290740524e-06,
+      "loss": 0.0341,
+      "step": 523
+    },
+    {
+      "epoch": 1.4744318181818181,
+      "grad_norm": 0.40146350860595703,
+      "learning_rate": 3.1400230205019124e-06,
+      "loss": 0.0405,
+      "step": 524
+    },
+    {
+      "epoch": 1.4772727272727273,
+      "grad_norm": 0.3381498157978058,
+      "learning_rate": 3.1071563235247883e-06,
+      "loss": 0.0463,
+      "step": 525
+    },
+    {
+      "epoch": 1.4801136363636362,
+      "grad_norm": 0.19083504378795624,
+      "learning_rate": 3.074430873306865e-06,
+      "loss": 0.0217,
+      "step": 526
+    },
+    {
+      "epoch": 1.4829545454545454,
+      "grad_norm": 0.4946057200431824,
+      "learning_rate": 3.041847340451456e-06,
+      "loss": 0.0499,
+      "step": 527
+    },
+    {
+      "epoch": 1.4857954545454546,
+      "grad_norm": 0.269089937210083,
+      "learning_rate": 3.0094063926537233e-06,
+      "loss": 0.0312,
+      "step": 528
+    },
+    {
+      "epoch": 1.4857954545454546,
+      "eval_loss": 0.04934047535061836,
+      "eval_runtime": 321.3119,
+      "eval_samples_per_second": 8.387,
+      "eval_steps_per_second": 1.049,
+      "step": 528
+    },
+    {
+      "epoch": 1.4886363636363638,
+      "grad_norm": 0.2874927818775177,
+      "learning_rate": 2.9771086946870177e-06,
+      "loss": 0.0212,
+      "step": 529
+    },
+    {
+      "epoch": 1.4914772727272727,
+      "grad_norm": 0.2589636743068695,
+      "learning_rate": 2.944954908389229e-06,
+      "loss": 0.0201,
+      "step": 530
+    },
+    {
+      "epoch": 1.4943181818181819,
+      "grad_norm": 0.3492966294288635,
+      "learning_rate": 2.912945692649255e-06,
+      "loss": 0.0354,
+      "step": 531
+    },
+    {
+      "epoch": 1.4971590909090908,
+      "grad_norm": 0.4711867570877075,
+      "learning_rate": 2.881081703393466e-06,
+      "loss": 0.044,
+      "step": 532
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.2505286931991577,
+      "learning_rate": 2.849363593572293e-06,
+      "loss": 0.0715,
+      "step": 533
+    },
+    {
+      "epoch": 1.5028409090909092,
+      "grad_norm": 0.12618179619312286,
+      "learning_rate": 2.8177920131468273e-06,
+      "loss": 0.0105,
+      "step": 534
+    },
+    {
+      "epoch": 1.5056818181818183,
+      "grad_norm": 0.22411687672138214,
+      "learning_rate": 2.7863676090755175e-06,
+      "loss": 0.0259,
+      "step": 535
+    },
+    {
+      "epoch": 1.5085227272727273,
+      "grad_norm": 0.4143596589565277,
+      "learning_rate": 2.7550910253008933e-06,
+      "loss": 0.0633,
+      "step": 536
+    },
+    {
+      "epoch": 1.5113636363636362,
+      "grad_norm": 0.4833144545555115,
+      "learning_rate": 2.723962902736389e-06,
+      "loss": 0.0649,
+      "step": 537
+    },
+    {
+      "epoch": 1.5142045454545454,
+      "grad_norm": 0.3849513530731201,
+      "learning_rate": 2.6929838792532035e-06,
+      "loss": 0.0527,
+      "step": 538
+    },
+    {
+      "epoch": 1.5170454545454546,
+      "grad_norm": 0.2576873004436493,
+      "learning_rate": 2.662154589667217e-06,
+      "loss": 0.0246,
+      "step": 539
+    },
+    {
+      "epoch": 1.5198863636363638,
+      "grad_norm": 0.23061725497245789,
+      "learning_rate": 2.6314756657260053e-06,
+      "loss": 0.0168,
+      "step": 540
+    },
+    {
+      "epoch": 1.5227272727272727,
+      "grad_norm": 0.18426130712032318,
+      "learning_rate": 2.6009477360958713e-06,
+      "loss": 0.0221,
+      "step": 541
+    },
+    {
+      "epoch": 1.5255681818181817,
+      "grad_norm": 0.21770624816417694,
+      "learning_rate": 2.570571426348978e-06,
+      "loss": 0.0115,
+      "step": 542
+    },
+    {
+      "epoch": 1.5284090909090908,
+      "grad_norm": 0.2922838032245636,
+      "learning_rate": 2.540347358950529e-06,
+      "loss": 0.0354,
+      "step": 543
+    },
+    {
+      "epoch": 1.53125,
+      "grad_norm": 0.15898585319519043,
+      "learning_rate": 2.5102761532460005e-06,
+      "loss": 0.0168,
+      "step": 544
+    },
+    {
+      "epoch": 1.5340909090909092,
+      "grad_norm": 0.3408936560153961,
+      "learning_rate": 2.480358425448457e-06,
+      "loss": 0.0383,
+      "step": 545
+    },
+    {
+      "epoch": 1.5369318181818183,
+      "grad_norm": 0.2640663981437683,
+      "learning_rate": 2.450594788625932e-06,
+      "loss": 0.0391,
+      "step": 546
+    },
+    {
+      "epoch": 1.5397727272727273,
+      "grad_norm": 0.18618500232696533,
+      "learning_rate": 2.420985852688854e-06,
+      "loss": 0.0163,
+      "step": 547
+    },
+    {
+      "epoch": 1.5426136363636362,
+      "grad_norm": 0.195343479514122,
+      "learning_rate": 2.3915322243775564e-06,
+      "loss": 0.0602,
+      "step": 548
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 0.22662031650543213,
+      "learning_rate": 2.362234507249832e-06,
+      "loss": 0.0262,
+      "step": 549
+    },
+    {
+      "epoch": 1.5482954545454546,
+      "grad_norm": 0.36082425713539124,
+      "learning_rate": 2.3330933016685753e-06,
+      "loss": 0.0483,
+      "step": 550
+    },
+    {
+      "epoch": 1.5511363636363638,
+      "grad_norm": 0.2544262111186981,
+      "learning_rate": 2.3041092047894843e-06,
+      "loss": 0.0209,
+      "step": 551
+    },
+    {
+      "epoch": 1.5539772727272727,
+      "grad_norm": 0.16143347322940826,
+      "learning_rate": 2.275282810548811e-06,
+      "loss": 0.0182,
+      "step": 552
+    },
+    {
+      "epoch": 1.5568181818181817,
+      "grad_norm": 0.32790812849998474,
+      "learning_rate": 2.2466147096512037e-06,
+      "loss": 0.0451,
+      "step": 553
+    },
+    {
+      "epoch": 1.5596590909090908,
+      "grad_norm": 0.187930628657341,
+      "learning_rate": 2.2181054895575847e-06,
+      "loss": 0.0259,
+      "step": 554
+    },
+    {
+      "epoch": 1.5625,
+      "grad_norm": 0.2864295542240143,
+      "learning_rate": 2.1897557344731292e-06,
+      "loss": 0.0285,
+      "step": 555
+    },
+    {
+      "epoch": 1.5653409090909092,
+      "grad_norm": 0.20400214195251465,
+      "learning_rate": 2.1615660253352888e-06,
+      "loss": 0.0207,
+      "step": 556
+    },
+    {
+      "epoch": 1.5681818181818183,
+      "grad_norm": 0.21328507363796234,
+      "learning_rate": 2.133536939801888e-06,
+      "loss": 0.0241,
+      "step": 557
+    },
+    {
+      "epoch": 1.5710227272727273,
+      "grad_norm": 0.250006765127182,
+      "learning_rate": 2.105669052239274e-06,
+      "loss": 0.0264,
+      "step": 558
+    },
+    {
+      "epoch": 1.5738636363636362,
+      "grad_norm": 0.28018662333488464,
+      "learning_rate": 2.077962933710572e-06,
+      "loss": 0.0376,
+      "step": 559
+    },
+    {
+      "epoch": 1.5767045454545454,
+      "grad_norm": 0.13916772603988647,
+      "learning_rate": 2.050419151963957e-06,
+      "loss": 0.0119,
+      "step": 560
+    },
+    {
+      "epoch": 1.5795454545454546,
+      "grad_norm": 0.16353723406791687,
+      "learning_rate": 2.0230382714210384e-06,
+      "loss": 0.0175,
+      "step": 561
+    },
+    {
+      "epoch": 1.5823863636363638,
+      "grad_norm": 0.3708636164665222,
+      "learning_rate": 1.9958208531652876e-06,
+      "loss": 0.0556,
+      "step": 562
+    },
+    {
+      "epoch": 1.5852272727272727,
+      "grad_norm": 0.29886114597320557,
+      "learning_rate": 1.9687674549305335e-06,
+      "loss": 0.0489,
+      "step": 563
+    },
+    {
+      "epoch": 1.5880681818181817,
+      "grad_norm": 0.15193718671798706,
+      "learning_rate": 1.9418786310895467e-06,
+      "loss": 0.0141,
+      "step": 564
+    },
+    {
+      "epoch": 1.5909090909090908,
+      "grad_norm": 0.40020623803138733,
+      "learning_rate": 1.9151549326426654e-06,
+      "loss": 0.0554,
+      "step": 565
+    },
+    {
+      "epoch": 1.59375,
+      "grad_norm": 0.1339547485113144,
+      "learning_rate": 1.8885969072065224e-06,
+      "loss": 0.0122,
+      "step": 566
+    },
+    {
+      "epoch": 1.5965909090909092,
+      "grad_norm": 0.14078541100025177,
+      "learning_rate": 1.8622050990027996e-06,
+      "loss": 0.0095,
+      "step": 567
+    },
+    {
+      "epoch": 1.5994318181818183,
+      "grad_norm": 0.20210637152194977,
+      "learning_rate": 1.835980048847098e-06,
+      "loss": 0.0212,
+      "step": 568
+    },
+    {
+      "epoch": 1.6022727272727273,
+      "grad_norm": 0.2009180337190628,
+      "learning_rate": 1.809922294137847e-06,
+      "loss": 0.0211,
+      "step": 569
+    },
+    {
+      "epoch": 1.6051136363636362,
+      "grad_norm": 0.2940455973148346,
+      "learning_rate": 1.7840323688452833e-06,
+      "loss": 0.0357,
+      "step": 570
+    },
+    {
+      "epoch": 1.6079545454545454,
+      "grad_norm": 0.20636039972305298,
+      "learning_rate": 1.7583108035005269e-06,
+      "loss": 0.0217,
+      "step": 571
+    },
+    {
+      "epoch": 1.6107954545454546,
+      "grad_norm": 0.6501495838165283,
+      "learning_rate": 1.7327581251846902e-06,
+      "loss": 0.0743,
+      "step": 572
+    },
+    {
+      "epoch": 1.6136363636363638,
+      "grad_norm": 0.27706483006477356,
+      "learning_rate": 1.7073748575180937e-06,
+      "loss": 0.0234,
+      "step": 573
+    },
+    {
+      "epoch": 1.6164772727272727,
+      "grad_norm": 0.19797846674919128,
+      "learning_rate": 1.6821615206495313e-06,
+      "loss": 0.0181,
+      "step": 574
+    },
+    {
+      "epoch": 1.6193181818181817,
+      "grad_norm": 0.2928190529346466,
+      "learning_rate": 1.6571186312456011e-06,
+      "loss": 0.0491,
+      "step": 575
+    },
+    {
+      "epoch": 1.6221590909090908,
+      "grad_norm": 0.26671117544174194,
+      "learning_rate": 1.6322467024801282e-06,
+      "loss": 0.0264,
+      "step": 576
+    },
+    {
+      "epoch": 1.625,
+      "grad_norm": 0.20558620989322662,
+      "learning_rate": 1.6075462440236512e-06,
+      "loss": 0.0178,
+      "step": 577
+    },
+    {
+      "epoch": 1.6278409090909092,
+      "grad_norm": 0.1802850067615509,
+      "learning_rate": 1.5830177620329712e-06,
+      "loss": 0.0212,
+      "step": 578
+    },
+    {
+      "epoch": 1.6306818181818183,
+      "grad_norm": 0.24599629640579224,
+      "learning_rate": 1.558661759140786e-06,
+      "loss": 0.0298,
+      "step": 579
+    },
+    {
+      "epoch": 1.6335227272727273,
+      "grad_norm": 0.1912572979927063,
+      "learning_rate": 1.5344787344453803e-06,
+      "loss": 0.0221,
+      "step": 580
+    },
+    {
+      "epoch": 1.6363636363636362,
+      "grad_norm": 0.2533075213432312,
+      "learning_rate": 1.510469183500405e-06,
+      "loss": 0.0372,
+      "step": 581
+    },
+    {
+      "epoch": 1.6392045454545454,
+      "grad_norm": 0.3094019293785095,
+      "learning_rate": 1.4866335983047264e-06,
+      "loss": 0.034,
+      "step": 582
+    },
+    {
+      "epoch": 1.6420454545454546,
+      "grad_norm": 0.30636775493621826,
+      "learning_rate": 1.4629724672923384e-06,
+      "loss": 0.0262,
+      "step": 583
+    },
+    {
+      "epoch": 1.6448863636363638,
+      "grad_norm": 0.3447078466415405,
+      "learning_rate": 1.439486275322357e-06,
+      "loss": 0.0446,
+      "step": 584
+    },
+    {
+      "epoch": 1.6477272727272727,
+      "grad_norm": 0.19196145236492157,
+      "learning_rate": 1.4161755036690771e-06,
+      "loss": 0.0185,
+      "step": 585
+    },
+    {
+      "epoch": 1.6505681818181817,
+      "grad_norm": 0.2407604306936264,
+      "learning_rate": 1.393040630012118e-06,
+      "loss": 0.0172,
+      "step": 586
+    },
+    {
+      "epoch": 1.6534090909090908,
+      "grad_norm": 0.3147240877151489,
+      "learning_rate": 1.3700821284266352e-06,
+      "loss": 0.0239,
+      "step": 587
+    },
+    {
+      "epoch": 1.65625,
+      "grad_norm": 0.23018233478069305,
+      "learning_rate": 1.3473004693736037e-06,
+      "loss": 0.026,
+      "step": 588
+    },
+    {
+      "epoch": 1.6590909090909092,
+      "grad_norm": 0.3745492696762085,
+      "learning_rate": 1.324696119690173e-06,
+      "loss": 0.0336,
+      "step": 589
+    },
+    {
+      "epoch": 1.6619318181818183,
+      "grad_norm": 0.1641959398984909,
+      "learning_rate": 1.302269542580109e-06,
+      "loss": 0.0157,
+      "step": 590
+    },
+    {
+      "epoch": 1.6647727272727273,
+      "grad_norm": 0.19555360078811646,
+      "learning_rate": 1.2800211976042942e-06,
+      "loss": 0.0213,
+      "step": 591
+    },
+    {
+      "epoch": 1.6676136363636362,
+      "grad_norm": 0.3400610685348511,
+      "learning_rate": 1.2579515406713194e-06,
+      "loss": 0.0429,
+      "step": 592
+    },
+    {
+      "epoch": 1.6704545454545454,
+      "grad_norm": 0.18559320271015167,
+      "learning_rate": 1.236061024028129e-06,
+      "loss": 0.0164,
+      "step": 593
+    },
+    {
+      "epoch": 1.6732954545454546,
+      "grad_norm": 0.23671524226665497,
+      "learning_rate": 1.214350096250767e-06,
+      "loss": 0.0258,
+      "step": 594
+    },
+    {
+      "epoch": 1.6761363636363638,
+      "grad_norm": 0.1904187798500061,
+      "learning_rate": 1.192819202235178e-06,
+      "loss": 0.0202,
+      "step": 595
+    },
+    {
+      "epoch": 1.6789772727272727,
+      "grad_norm": 0.23071539402008057,
+      "learning_rate": 1.1714687831880867e-06,
+      "loss": 0.0231,
+      "step": 596
+    },
+    {
+      "epoch": 1.6818181818181817,
+      "grad_norm": 0.25730565190315247,
+      "learning_rate": 1.1502992766179666e-06,
+      "loss": 0.0338,
+      "step": 597
+    },
+    {
+      "epoch": 1.6846590909090908,
+      "grad_norm": 0.6486604809761047,
+      "learning_rate": 1.1293111163260639e-06,
+      "loss": 0.0943,
+      "step": 598
+    },
+    {
+      "epoch": 1.6875,
+      "grad_norm": 0.29426586627960205,
+      "learning_rate": 1.1085047323975173e-06,
+      "loss": 0.043,
+      "step": 599
+    },
+    {
+      "epoch": 1.6903409090909092,
+      "grad_norm": 0.25408247113227844,
+      "learning_rate": 1.0878805511925438e-06,
+      "loss": 0.0303,
+      "step": 600
+    },
+    {
+      "epoch": 1.6931818181818183,
+      "grad_norm": 0.372081458568573,
+      "learning_rate": 1.0674389953376928e-06,
+      "loss": 0.033,
+      "step": 601
+    },
+    {
+      "epoch": 1.6960227272727273,
+      "grad_norm": 0.17783237993717194,
+      "learning_rate": 1.0471804837171916e-06,
+      "loss": 0.0124,
+      "step": 602
+    },
+    {
+      "epoch": 1.6988636363636362,
+      "grad_norm": 0.35536065697669983,
+      "learning_rate": 1.027105431464368e-06,
+      "loss": 0.0435,
+      "step": 603
+    },
+    {
+      "epoch": 1.7017045454545454,
+      "grad_norm": 0.2801659405231476,
+      "learning_rate": 1.0072142499531346e-06,
+      "loss": 0.0329,
+      "step": 604
+    },
+    {
+      "epoch": 1.7045454545454546,
+      "grad_norm": 0.1209142804145813,
+      "learning_rate": 9.875073467895635e-07,
+      "loss": 0.011,
+      "step": 605
+    },
+    {
+      "epoch": 1.7073863636363638,
+      "grad_norm": 0.14608953893184662,
+      "learning_rate": 9.679851258035277e-07,
+      "loss": 0.0123,
+      "step": 606
+    },
+    {
+      "epoch": 1.7102272727272727,
+      "grad_norm": 0.17487914860248566,
+      "learning_rate": 9.48647987040433e-07,
+      "loss": 0.0187,
+      "step": 607
+    },
+    {
+      "epoch": 1.7130681818181817,
+      "grad_norm": 0.2767109274864197,
+      "learning_rate": 9.294963267530177e-07,
+      "loss": 0.0359,
+      "step": 608
+    },
+    {
+      "epoch": 1.7159090909090908,
+      "grad_norm": 0.28444820642471313,
+      "learning_rate": 9.105305373932338e-07,
+      "loss": 0.0586,
+      "step": 609
+    },
+    {
+      "epoch": 1.71875,
+      "grad_norm": 0.5057897567749023,
+      "learning_rate": 8.917510076042058e-07,
+      "loss": 0.0459,
+      "step": 610
+    },
+    {
+      "epoch": 1.7215909090909092,
+      "grad_norm": 0.33112823963165283,
+      "learning_rate": 8.731581222122587e-07,
+      "loss": 0.0414,
+      "step": 611
+    },
+    {
+      "epoch": 1.7244318181818183,
+      "grad_norm": 0.22497576475143433,
+      "learning_rate": 8.547522622190385e-07,
+      "loss": 0.0259,
+      "step": 612
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 0.18142753839492798,
+      "learning_rate": 8.365338047937122e-07,
+      "loss": 0.0166,
+      "step": 613
+    },
+    {
+      "epoch": 1.7301136363636362,
+      "grad_norm": 0.18240651488304138,
+      "learning_rate": 8.185031232652252e-07,
+      "loss": 0.0176,
+      "step": 614
+    },
+    {
+      "epoch": 1.7329545454545454,
+      "grad_norm": 0.4496082365512848,
+      "learning_rate": 8.006605871146578e-07,
+      "loss": 0.0361,
+      "step": 615
+    },
+    {
+      "epoch": 1.7357954545454546,
+      "grad_norm": 0.3641889989376068,
+      "learning_rate": 7.830065619676519e-07,
+      "loss": 0.0339,
+      "step": 616
+    },
+    {
+      "epoch": 1.7357954545454546,
+      "eval_loss": 0.046951260417699814,
+      "eval_runtime": 322.0399,
+      "eval_samples_per_second": 8.369,
+      "eval_steps_per_second": 1.046,
+      "step": 616
+    },
+    {
+      "epoch": 1.7386363636363638,
+      "grad_norm": 0.4126671850681305,
+      "learning_rate": 7.65541409586924e-07,
+      "loss": 0.0422,
+      "step": 617
+    },
+    {
+      "epoch": 1.7414772727272727,
+      "grad_norm": 0.3763576149940491,
+      "learning_rate": 7.482654878648465e-07,
+      "loss": 0.0417,
+      "step": 618
+    },
+    {
+      "epoch": 1.7443181818181817,
+      "grad_norm": 0.25049203634262085,
+      "learning_rate": 7.311791508161159e-07,
+      "loss": 0.0264,
+      "step": 619
+    },
+    {
+      "epoch": 1.7471590909090908,
+      "grad_norm": 0.19837282598018646,
+      "learning_rate": 7.142827485704951e-07,
+      "loss": 0.022,
+      "step": 620
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5346717238426208,
+      "learning_rate": 6.975766273656425e-07,
+      "loss": 0.0461,
+      "step": 621
+    },
+    {
+      "epoch": 1.7528409090909092,
+      "grad_norm": 0.5660947561264038,
+      "learning_rate": 6.810611295400171e-07,
+      "loss": 0.0568,
+      "step": 622
+    },
+    {
+      "epoch": 1.7556818181818183,
+      "grad_norm": 0.14554603397846222,
+      "learning_rate": 6.647365935258642e-07,
+      "loss": 0.0158,
+      "step": 623
+    },
+    {
+      "epoch": 1.7585227272727273,
+      "grad_norm": 0.48322027921676636,
+      "learning_rate": 6.48603353842272e-07,
+      "loss": 0.048,
+      "step": 624
+    },
+    {
+      "epoch": 1.7613636363636362,
+      "grad_norm": 0.49817270040512085,
+      "learning_rate": 6.326617410883296e-07,
+      "loss": 0.0823,
+      "step": 625
+    },
+    {
+      "epoch": 1.7642045454545454,
+      "grad_norm": 0.41815218329429626,
+      "learning_rate": 6.169120819363406e-07,
+      "loss": 0.0629,
+      "step": 626
+    },
+    {
+      "epoch": 1.7670454545454546,
+      "grad_norm": 0.1850947141647339,
+      "learning_rate": 6.013546991251373e-07,
+      "loss": 0.0185,
+      "step": 627
+    },
+    {
+      "epoch": 1.7698863636363638,
+      "grad_norm": 0.5910592079162598,
+      "learning_rate": 5.859899114534662e-07,
+      "loss": 0.0857,
+      "step": 628
+    },
+    {
+      "epoch": 1.7727272727272727,
+      "grad_norm": 0.35453927516937256,
+      "learning_rate": 5.708180337734448e-07,
+      "loss": 0.0308,
+      "step": 629
+    },
+    {
+      "epoch": 1.7755681818181817,
+      "grad_norm": 0.19171951711177826,
+      "learning_rate": 5.558393769841286e-07,
+      "loss": 0.0539,
+      "step": 630
+    },
+    {
+      "epoch": 1.7784090909090908,
+      "grad_norm": 0.2592536211013794,
+      "learning_rate": 5.410542480251202e-07,
+      "loss": 0.0207,
+      "step": 631
+    },
+    {
+      "epoch": 1.78125,
+      "grad_norm": 0.2815329432487488,
+      "learning_rate": 5.264629498702966e-07,
+      "loss": 0.0326,
+      "step": 632
+    },
+    {
+      "epoch": 1.7840909090909092,
+      "grad_norm": 0.14800816774368286,
+      "learning_rate": 5.12065781521588e-07,
+      "loss": 0.0153,
+      "step": 633
+    },
+    {
+      "epoch": 1.7869318181818183,
+      "grad_norm": 0.14742711186408997,
+      "learning_rate": 4.978630380028582e-07,
+      "loss": 0.0176,
+      "step": 634
+    },
+    {
+      "epoch": 1.7897727272727273,
+      "grad_norm": 0.5278242826461792,
+      "learning_rate": 4.838550103538575e-07,
+      "loss": 0.0788,
+      "step": 635
+    },
+    {
+      "epoch": 1.7926136363636362,
+      "grad_norm": 0.18585218489170074,
+      "learning_rate": 4.700419856242555e-07,
+      "loss": 0.015,
+      "step": 636
+    },
+    {
+      "epoch": 1.7954545454545454,
+      "grad_norm": 0.16984280943870544,
+      "learning_rate": 4.5642424686776154e-07,
+      "loss": 0.0184,
+      "step": 637
+    },
+    {
+      "epoch": 1.7982954545454546,
+      "grad_norm": 0.18244123458862305,
+      "learning_rate": 4.4300207313632713e-07,
+      "loss": 0.0185,
+      "step": 638
+    },
+    {
+      "epoch": 1.8011363636363638,
+      "grad_norm": 0.2502816915512085,
+      "learning_rate": 4.2977573947442175e-07,
+      "loss": 0.0229,
+      "step": 639
+    },
+    {
+      "epoch": 1.8039772727272727,
+      "grad_norm": 0.22571499645709991,
+      "learning_rate": 4.167455169134027e-07,
+      "loss": 0.0237,
+      "step": 640
+    },
+    {
+      "epoch": 1.8068181818181817,
+      "grad_norm": 0.33113282918930054,
+      "learning_rate": 4.039116724659564e-07,
+      "loss": 0.0334,
+      "step": 641
+    },
+    {
+      "epoch": 1.8096590909090908,
+      "grad_norm": 0.29996100068092346,
+      "learning_rate": 3.9127446912062606e-07,
+      "loss": 0.0295,
+      "step": 642
+    },
+    {
+      "epoch": 1.8125,
+      "grad_norm": 0.26119595766067505,
+      "learning_rate": 3.788341658364314e-07,
+      "loss": 0.0313,
+      "step": 643
+    },
+    {
+      "epoch": 1.8153409090909092,
+      "grad_norm": 0.2088990956544876,
+      "learning_rate": 3.6659101753754975e-07,
+      "loss": 0.0286,
+      "step": 644
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.23757684230804443,
+      "learning_rate": 3.5454527510810355e-07,
+      "loss": 0.0249,
+      "step": 645
+    },
+    {
+      "epoch": 1.8210227272727273,
+      "grad_norm": 0.1616797298192978,
+      "learning_rate": 3.426971853870109e-07,
+      "loss": 0.0129,
+      "step": 646
+    },
+    {
+      "epoch": 1.8238636363636362,
+      "grad_norm": 0.21369072794914246,
+      "learning_rate": 3.3104699116292883e-07,
+      "loss": 0.0254,
+      "step": 647
+    },
+    {
+      "epoch": 1.8267045454545454,
+      "grad_norm": 0.3602796196937561,
+      "learning_rate": 3.1959493116928473e-07,
+      "loss": 0.0458,
+      "step": 648
+    },
+    {
+      "epoch": 1.8295454545454546,
+      "grad_norm": 0.2149425595998764,
+      "learning_rate": 3.0834124007937616e-07,
+      "loss": 0.0231,
+      "step": 649
+    },
+    {
+      "epoch": 1.8323863636363638,
+      "grad_norm": 0.16421711444854736,
+      "learning_rate": 2.972861485015666e-07,
+      "loss": 0.0139,
+      "step": 650
+    },
+    {
+      "epoch": 1.8352272727272727,
+      "grad_norm": 0.45803770422935486,
+      "learning_rate": 2.864298829745571e-07,
+      "loss": 0.0787,
+      "step": 651
+    },
+    {
+      "epoch": 1.8380681818181817,
+      "grad_norm": 0.5590444207191467,
+      "learning_rate": 2.7577266596274577e-07,
+      "loss": 0.0367,
+      "step": 652
+    },
+    {
+      "epoch": 1.8409090909090908,
+      "grad_norm": 0.7343291640281677,
+      "learning_rate": 2.6531471585167e-07,
+      "loss": 0.1223,
+      "step": 653
+    },
+    {
+      "epoch": 1.84375,
+      "grad_norm": 0.23800815641880035,
+      "learning_rate": 2.5505624694353027e-07,
+      "loss": 0.0217,
+      "step": 654
+    },
+    {
+      "epoch": 1.8465909090909092,
+      "grad_norm": 0.2187035083770752,
+      "learning_rate": 2.4499746945279566e-07,
+      "loss": 0.0202,
+      "step": 655
+    },
+    {
+      "epoch": 1.8494318181818183,
+      "grad_norm": 0.4073978364467621,
+      "learning_rate": 2.3513858950190206e-07,
+      "loss": 0.0514,
+      "step": 656
+    },
+    {
+      "epoch": 1.8522727272727273,
+      "grad_norm": 0.17420588433742523,
+      "learning_rate": 2.2547980911702406e-07,
+      "loss": 0.021,
+      "step": 657
+    },
+    {
+      "epoch": 1.8551136363636362,
+      "grad_norm": 0.2199344038963318,
+      "learning_rate": 2.1602132622393745e-07,
+      "loss": 0.0258,
+      "step": 658
+    },
+    {
+      "epoch": 1.8579545454545454,
+      "grad_norm": 0.1626044064760208,
+      "learning_rate": 2.0676333464396125e-07,
+      "loss": 0.0184,
+      "step": 659
+    },
+    {
+      "epoch": 1.8607954545454546,
+      "grad_norm": 0.20894934237003326,
+      "learning_rate": 1.9770602408998642e-07,
+      "loss": 0.0318,
+      "step": 660
+    },
+    {
+      "epoch": 1.8636363636363638,
+      "grad_norm": 0.11425574123859406,
+      "learning_rate": 1.8884958016259114e-07,
+      "loss": 0.009,
+      "step": 661
+    },
+    {
+      "epoch": 1.8664772727272727,
+      "grad_norm": 0.13908714056015015,
+      "learning_rate": 1.8019418434623405e-07,
+      "loss": 0.0117,
+      "step": 662
+    },
+    {
+      "epoch": 1.8693181818181817,
+      "grad_norm": 0.30832770466804504,
+      "learning_rate": 1.7174001400553586e-07,
+      "loss": 0.0376,
+      "step": 663
+    },
+    {
+      "epoch": 1.8721590909090908,
+      "grad_norm": 0.10906349122524261,
+      "learning_rate": 1.6348724238164583e-07,
+      "loss": 0.009,
+      "step": 664
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 0.20455297827720642,
+      "learning_rate": 1.5543603858869216e-07,
+      "loss": 0.0193,
+      "step": 665
+    },
+    {
+      "epoch": 1.8778409090909092,
+      "grad_norm": 0.09518078714609146,
+      "learning_rate": 1.475865676103161e-07,
+      "loss": 0.0081,
+      "step": 666
+    },
+    {
+      "epoch": 1.8806818181818183,
+      "grad_norm": 0.08790381997823715,
+      "learning_rate": 1.3993899029629e-07,
+      "loss": 0.0072,
+      "step": 667
+    },
+    {
+      "epoch": 1.8835227272727273,
+      "grad_norm": 0.27762630581855774,
+      "learning_rate": 1.324934633592201e-07,
+      "loss": 0.0292,
+      "step": 668
+    },
+    {
+      "epoch": 1.8863636363636362,
+      "grad_norm": 0.16567471623420715,
+      "learning_rate": 1.2525013937134122e-07,
+      "loss": 0.016,
+      "step": 669
+    },
+    {
+      "epoch": 1.8892045454545454,
+      "grad_norm": 0.49977800250053406,
+      "learning_rate": 1.1820916676138384e-07,
+      "loss": 0.0452,
+      "step": 670
+    },
+    {
+      "epoch": 1.8920454545454546,
+      "grad_norm": 0.2329680621623993,
+      "learning_rate": 1.1137068981153632e-07,
+      "loss": 0.0233,
+      "step": 671
+    },
+    {
+      "epoch": 1.8948863636363638,
+      "grad_norm": 0.322808176279068,
+      "learning_rate": 1.0473484865448524e-07,
+      "loss": 0.0469,
+      "step": 672
+    },
+    {
+      "epoch": 1.8977272727272727,
+      "grad_norm": 0.2716865837574005,
+      "learning_rate": 9.830177927054429e-08,
+      "loss": 0.0377,
+      "step": 673
+    },
+    {
+      "epoch": 1.9005681818181817,
+      "grad_norm": 0.14299501478672028,
+      "learning_rate": 9.207161348487315e-08,
+      "loss": 0.0178,
+      "step": 674
+    },
+    {
+      "epoch": 1.9034090909090908,
+      "grad_norm": 0.23493549227714539,
+      "learning_rate": 8.604447896476853e-08,
+      "loss": 0.0241,
+      "step": 675
+    },
+    {
+      "epoch": 1.90625,
+      "grad_norm": 0.2541610896587372,
+      "learning_rate": 8.0220499217053e-08,
+      "loss": 0.033,
+      "step": 676
+    },
+    {
+      "epoch": 1.9090909090909092,
+      "grad_norm": 0.207777202129364,
+      "learning_rate": 7.459979358554248e-08,
+      "loss": 0.0328,
+      "step": 677
+    },
+    {
+      "epoch": 1.9119318181818183,
+      "grad_norm": 0.2648699879646301,
+      "learning_rate": 6.918247724859939e-08,
+      "loss": 0.0371,
+      "step": 678
+    },
+    {
+      "epoch": 1.9147727272727273,
+      "grad_norm": 0.4931396245956421,
+      "learning_rate": 6.396866121677558e-08,
+      "loss": 0.0261,
+      "step": 679
+    },
+    {
+      "epoch": 1.9176136363636362,
+      "grad_norm": 0.46385493874549866,
+      "learning_rate": 5.895845233053643e-08,
+      "loss": 0.0771,
+      "step": 680
+    },
+    {
+      "epoch": 1.9204545454545454,
+      "grad_norm": 0.18477115035057068,
+      "learning_rate": 5.415195325806699e-08,
+      "loss": 0.0237,
+      "step": 681
+    },
+    {
+      "epoch": 1.9232954545454546,
+      "grad_norm": 0.23025956749916077,
+      "learning_rate": 4.954926249317815e-08,
+      "loss": 0.0298,
+      "step": 682
+    },
+    {
+      "epoch": 1.9261363636363638,
+      "grad_norm": 0.11648344993591309,
+      "learning_rate": 4.5150474353274906e-08,
+      "loss": 0.0093,
+      "step": 683
+    },
+    {
+      "epoch": 1.9289772727272727,
+      "grad_norm": 0.19737114012241364,
+      "learning_rate": 4.0955678977436796e-08,
+      "loss": 0.0248,
+      "step": 684
+    },
+    {
+      "epoch": 1.9318181818181817,
+      "grad_norm": 0.29464903473854065,
+      "learning_rate": 3.696496232456159e-08,
+      "loss": 0.0268,
+      "step": 685
+    },
+    {
+      "epoch": 1.9346590909090908,
+      "grad_norm": 0.16712042689323425,
+      "learning_rate": 3.3178406171608946e-08,
+      "loss": 0.0178,
+      "step": 686
+    },
+    {
+      "epoch": 1.9375,
+      "grad_norm": 0.1289013773202896,
+      "learning_rate": 2.9596088111922828e-08,
+      "loss": 0.0145,
+      "step": 687
+    },
+    {
+      "epoch": 1.9403409090909092,
+      "grad_norm": 0.2535853981971741,
+      "learning_rate": 2.6218081553638363e-08,
+      "loss": 0.0272,
+      "step": 688
+    },
+    {
+      "epoch": 1.9431818181818183,
+      "grad_norm": 0.18250080943107605,
+      "learning_rate": 2.3044455718185254e-08,
+      "loss": 0.0171,
+      "step": 689
+    },
+    {
+      "epoch": 1.9460227272727273,
+      "grad_norm": 0.4696236848831177,
+      "learning_rate": 2.0075275638862247e-08,
+      "loss": 0.0886,
+      "step": 690
+    },
+    {
+      "epoch": 1.9488636363636362,
+      "grad_norm": 0.22248291969299316,
+      "learning_rate": 1.7310602159505973e-08,
+      "loss": 0.0246,
+      "step": 691
+    },
+    {
+      "epoch": 1.9517045454545454,
+      "grad_norm": 0.21488967537879944,
+      "learning_rate": 1.4750491933247513e-08,
+      "loss": 0.0255,
+      "step": 692
+    },
+    {
+      "epoch": 1.9545454545454546,
+      "grad_norm": 0.19222155213356018,
+      "learning_rate": 1.2394997421347753e-08,
+      "loss": 0.0282,
+      "step": 693
+    },
+    {
+      "epoch": 1.9573863636363638,
+      "grad_norm": 0.2110484093427658,
+      "learning_rate": 1.0244166892124929e-08,
+      "loss": 0.0159,
+      "step": 694
+    },
+    {
+      "epoch": 1.9602272727272727,
+      "grad_norm": 0.3025636374950409,
+      "learning_rate": 8.29804441996207e-09,
+      "loss": 0.0385,
+      "step": 695
+    },
+    {
+      "epoch": 1.9630681818181817,
+      "grad_norm": 0.20144778490066528,
+      "learning_rate": 6.556669884408839e-09,
+      "loss": 0.0288,
+      "step": 696
+    },
+    {
+      "epoch": 1.9659090909090908,
+      "grad_norm": 0.364287406206131,
+      "learning_rate": 5.0200789693588544e-09,
+      "loss": 0.0404,
+      "step": 697
+    },
+    {
+      "epoch": 1.96875,
+      "grad_norm": 0.1691567450761795,
+      "learning_rate": 3.688303162322493e-09,
+      "loss": 0.0192,
+      "step": 698
+    },
+    {
+      "epoch": 1.9715909090909092,
+      "grad_norm": 0.41609564423561096,
+      "learning_rate": 2.5613697537818505e-09,
+      "loss": 0.0607,
+      "step": 699
+    },
+    {
+      "epoch": 1.9744318181818183,
+      "grad_norm": 0.3112366795539856,
+      "learning_rate": 1.6393018366278601e-09,
+      "loss": 0.0486,
+      "step": 700
+    },
+    {
+      "epoch": 1.9772727272727273,
+      "grad_norm": 0.2627716660499573,
+      "learning_rate": 9.221183056895566e-10,
+      "loss": 0.0262,
+      "step": 701
+    },
+    {
+      "epoch": 1.9801136363636362,
+      "grad_norm": 0.3606441020965576,
+      "learning_rate": 4.0983385734660875e-10,
+      "loss": 0.0411,
+      "step": 702
+    },
+    {
+      "epoch": 1.9829545454545454,
+      "grad_norm": 0.25090616941452026,
+      "learning_rate": 1.0245898922844889e-10,
+      "loss": 0.0203,
+      "step": 703
+    },
+    {
+      "epoch": 1.9857954545454546,
+      "grad_norm": 0.23430025577545166,
+      "learning_rate": 0.0,
+      "loss": 0.0275,
+      "step": 704
+    },
+    {
+      "epoch": 1.9857954545454546,
+      "eval_loss": 0.04643603414297104,
+      "eval_runtime": 321.1661,
+      "eval_samples_per_second": 8.391,
+      "eval_steps_per_second": 1.049,
+      "step": 704
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 704,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 176,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.4271718282860954e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}