diff --git "a/checkpoint-1128/trainer_state.json" "b/checkpoint-1128/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1128/trainer_state.json"
@@ -0,0 +1,7961 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 226,
+  "global_step": 1128,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0017730496453900709,
+      "grad_norm": 6.88893985748291,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 4.048,
+      "step": 1
+    },
+    {
+      "epoch": 0.0035460992907801418,
+      "grad_norm": 6.882226467132568,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 4.2161,
+      "step": 2
+    },
+    {
+      "epoch": 0.005319148936170213,
+      "grad_norm": 4.995457649230957,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 3.9501,
+      "step": 3
+    },
+    {
+      "epoch": 0.0070921985815602835,
+      "grad_norm": 3.743339776992798,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 3.7819,
+      "step": 4
+    },
+    {
+      "epoch": 0.008865248226950355,
+      "grad_norm": 3.3828792572021484,
+      "learning_rate": 0.00015,
+      "loss": 3.2351,
+      "step": 5
+    },
+    {
+      "epoch": 0.010638297872340425,
+      "grad_norm": 3.3940675258636475,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 2.7343,
+      "step": 6
+    },
+    {
+      "epoch": 0.012411347517730497,
+      "grad_norm": 3.3435909748077393,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 2.0898,
+      "step": 7
+    },
+    {
+      "epoch": 0.014184397163120567,
+      "grad_norm": 4.3948235511779785,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 1.6608,
+      "step": 8
+    },
+    {
+      "epoch": 0.015957446808510637,
+      "grad_norm": 4.144540309906006,
+      "learning_rate": 0.00027,
+      "loss": 0.9104,
+      "step": 9
+    },
+    {
+      "epoch": 0.01773049645390071,
+      "grad_norm": 5.76711368560791,
+      "learning_rate": 0.0003,
+      "loss": 0.6078,
+      "step": 10
+    },
+    {
+      "epoch": 0.01950354609929078,
+      "grad_norm": 11.037919998168945,
+      "learning_rate": 0.00029973166368515205,
+      "loss": 0.439,
+      "step": 11
+    },
+    {
+      "epoch": 0.02127659574468085,
+      "grad_norm": 5.379396915435791,
+      "learning_rate": 0.00029946332737030407,
+      "loss": 0.5497,
+      "step": 12
+    },
+    {
+      "epoch": 0.02304964539007092,
+      "grad_norm": 6.821712970733643,
+      "learning_rate": 0.00029919499105545615,
+      "loss": 0.6396,
+      "step": 13
+    },
+    {
+      "epoch": 0.024822695035460994,
+      "grad_norm": 3.520746946334839,
+      "learning_rate": 0.00029892665474060817,
+      "loss": 0.4007,
+      "step": 14
+    },
+    {
+      "epoch": 0.026595744680851064,
+      "grad_norm": 1.7975941896438599,
+      "learning_rate": 0.00029865831842576025,
+      "loss": 0.2384,
+      "step": 15
+    },
+    {
+      "epoch": 0.028368794326241134,
+      "grad_norm": 6.044783592224121,
+      "learning_rate": 0.0002983899821109123,
+      "loss": 0.5119,
+      "step": 16
+    },
+    {
+      "epoch": 0.030141843971631204,
+      "grad_norm": 0.9617128968238831,
+      "learning_rate": 0.0002981216457960644,
+      "loss": 0.2032,
+      "step": 17
+    },
+    {
+      "epoch": 0.031914893617021274,
+      "grad_norm": 1.6461282968521118,
+      "learning_rate": 0.0002978533094812164,
+      "loss": 0.3118,
+      "step": 18
+    },
+    {
+      "epoch": 0.03368794326241135,
+      "grad_norm": 1.2195343971252441,
+      "learning_rate": 0.0002975849731663685,
+      "loss": 0.2522,
+      "step": 19
+    },
+    {
+      "epoch": 0.03546099290780142,
+      "grad_norm": 1.5621528625488281,
+      "learning_rate": 0.00029731663685152057,
+      "loss": 0.3118,
+      "step": 20
+    },
+    {
+      "epoch": 0.03723404255319149,
+      "grad_norm": 1.3373510837554932,
+      "learning_rate": 0.0002970483005366726,
+      "loss": 0.2101,
+      "step": 21
+    },
+    {
+      "epoch": 0.03900709219858156,
+      "grad_norm": 0.9622485041618347,
+      "learning_rate": 0.00029677996422182467,
+      "loss": 0.1797,
+      "step": 22
+    },
+    {
+      "epoch": 0.040780141843971635,
+      "grad_norm": 1.7572095394134521,
+      "learning_rate": 0.0002965116279069767,
+      "loss": 0.1338,
+      "step": 23
+    },
+    {
+      "epoch": 0.0425531914893617,
+      "grad_norm": 1.871632695198059,
+      "learning_rate": 0.00029624329159212877,
+      "loss": 0.1457,
+      "step": 24
+    },
+    {
+      "epoch": 0.044326241134751775,
+      "grad_norm": 2.204594612121582,
+      "learning_rate": 0.00029597495527728084,
+      "loss": 0.1621,
+      "step": 25
+    },
+    {
+      "epoch": 0.04609929078014184,
+      "grad_norm": 1.5381090641021729,
+      "learning_rate": 0.0002957066189624329,
+      "loss": 0.1703,
+      "step": 26
+    },
+    {
+      "epoch": 0.047872340425531915,
+      "grad_norm": 0.8811854124069214,
+      "learning_rate": 0.00029543828264758494,
+      "loss": 0.0787,
+      "step": 27
+    },
+    {
+      "epoch": 0.04964539007092199,
+      "grad_norm": 0.8814285397529602,
+      "learning_rate": 0.000295169946332737,
+      "loss": 0.2091,
+      "step": 28
+    },
+    {
+      "epoch": 0.051418439716312055,
+      "grad_norm": 1.371406078338623,
+      "learning_rate": 0.0002949016100178891,
+      "loss": 0.3093,
+      "step": 29
+    },
+    {
+      "epoch": 0.05319148936170213,
+      "grad_norm": 1.1055055856704712,
+      "learning_rate": 0.0002946332737030411,
+      "loss": 0.1644,
+      "step": 30
+    },
+    {
+      "epoch": 0.0549645390070922,
+      "grad_norm": 1.2226752042770386,
+      "learning_rate": 0.0002943649373881932,
+      "loss": 0.1597,
+      "step": 31
+    },
+    {
+      "epoch": 0.05673758865248227,
+      "grad_norm": 1.1774011850357056,
+      "learning_rate": 0.0002940966010733452,
+      "loss": 0.243,
+      "step": 32
+    },
+    {
+      "epoch": 0.05851063829787234,
+      "grad_norm": 1.8663568496704102,
+      "learning_rate": 0.0002938282647584973,
+      "loss": 0.3355,
+      "step": 33
+    },
+    {
+      "epoch": 0.06028368794326241,
+      "grad_norm": 0.7782514095306396,
+      "learning_rate": 0.0002935599284436493,
+      "loss": 0.1426,
+      "step": 34
+    },
+    {
+      "epoch": 0.06205673758865248,
+      "grad_norm": 0.6109825968742371,
+      "learning_rate": 0.0002932915921288014,
+      "loss": 0.1334,
+      "step": 35
+    },
+    {
+      "epoch": 0.06382978723404255,
+      "grad_norm": 1.297591209411621,
+      "learning_rate": 0.00029302325581395347,
+      "loss": 0.3717,
+      "step": 36
+    },
+    {
+      "epoch": 0.06560283687943262,
+      "grad_norm": 0.8587173223495483,
+      "learning_rate": 0.00029275491949910554,
+      "loss": 0.1469,
+      "step": 37
+    },
+    {
+      "epoch": 0.0673758865248227,
+      "grad_norm": 0.8923099637031555,
+      "learning_rate": 0.00029248658318425756,
+      "loss": 0.1838,
+      "step": 38
+    },
+    {
+      "epoch": 0.06914893617021277,
+      "grad_norm": 0.8410468697547913,
+      "learning_rate": 0.00029221824686940964,
+      "loss": 0.2471,
+      "step": 39
+    },
+    {
+      "epoch": 0.07092198581560284,
+      "grad_norm": 0.5034096240997314,
+      "learning_rate": 0.0002919499105545617,
+      "loss": 0.0616,
+      "step": 40
+    },
+    {
+      "epoch": 0.0726950354609929,
+      "grad_norm": 0.31462836265563965,
+      "learning_rate": 0.00029168157423971374,
+      "loss": 0.0519,
+      "step": 41
+    },
+    {
+      "epoch": 0.07446808510638298,
+      "grad_norm": 0.5840567350387573,
+      "learning_rate": 0.0002914132379248658,
+      "loss": 0.0814,
+      "step": 42
+    },
+    {
+      "epoch": 0.07624113475177305,
+      "grad_norm": 0.5441888570785522,
+      "learning_rate": 0.00029114490161001784,
+      "loss": 0.0731,
+      "step": 43
+    },
+    {
+      "epoch": 0.07801418439716312,
+      "grad_norm": 1.0477969646453857,
+      "learning_rate": 0.0002908765652951699,
+      "loss": 0.1842,
+      "step": 44
+    },
+    {
+      "epoch": 0.0797872340425532,
+      "grad_norm": 1.046530842781067,
+      "learning_rate": 0.000290608228980322,
+      "loss": 0.2727,
+      "step": 45
+    },
+    {
+      "epoch": 0.08156028368794327,
+      "grad_norm": 0.721427857875824,
+      "learning_rate": 0.00029033989266547406,
+      "loss": 0.1343,
+      "step": 46
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "grad_norm": 2.0451300144195557,
+      "learning_rate": 0.0002900715563506261,
+      "loss": 0.076,
+      "step": 47
+    },
+    {
+      "epoch": 0.0851063829787234,
+      "grad_norm": 0.3739748001098633,
+      "learning_rate": 0.00028980322003577816,
+      "loss": 0.0419,
+      "step": 48
+    },
+    {
+      "epoch": 0.08687943262411348,
+      "grad_norm": 0.5523413419723511,
+      "learning_rate": 0.00028953488372093024,
+      "loss": 0.0563,
+      "step": 49
+    },
+    {
+      "epoch": 0.08865248226950355,
+      "grad_norm": 0.5457972288131714,
+      "learning_rate": 0.00028926654740608226,
+      "loss": 0.0256,
+      "step": 50
+    },
+    {
+      "epoch": 0.09042553191489362,
+      "grad_norm": 2.8798046112060547,
+      "learning_rate": 0.00028899821109123434,
+      "loss": 0.8653,
+      "step": 51
+    },
+    {
+      "epoch": 0.09219858156028368,
+      "grad_norm": 2.86541748046875,
+      "learning_rate": 0.00028872987477638636,
+      "loss": 0.5141,
+      "step": 52
+    },
+    {
+      "epoch": 0.09397163120567376,
+      "grad_norm": 10.113655090332031,
+      "learning_rate": 0.00028846153846153843,
+      "loss": 0.278,
+      "step": 53
+    },
+    {
+      "epoch": 0.09574468085106383,
+      "grad_norm": 3.4064385890960693,
+      "learning_rate": 0.00028819320214669046,
+      "loss": 0.3353,
+      "step": 54
+    },
+    {
+      "epoch": 0.0975177304964539,
+      "grad_norm": 4.866039752960205,
+      "learning_rate": 0.00028792486583184253,
+      "loss": 0.4487,
+      "step": 55
+    },
+    {
+      "epoch": 0.09929078014184398,
+      "grad_norm": 5.0878143310546875,
+      "learning_rate": 0.0002876565295169946,
+      "loss": 0.2099,
+      "step": 56
+    },
+    {
+      "epoch": 0.10106382978723404,
+      "grad_norm": 5.835223197937012,
+      "learning_rate": 0.0002873881932021467,
+      "loss": 0.4639,
+      "step": 57
+    },
+    {
+      "epoch": 0.10283687943262411,
+      "grad_norm": 3.5590853691101074,
+      "learning_rate": 0.00028711985688729876,
+      "loss": 0.3146,
+      "step": 58
+    },
+    {
+      "epoch": 0.10460992907801418,
+      "grad_norm": 5.281091213226318,
+      "learning_rate": 0.0002868515205724508,
+      "loss": 0.4222,
+      "step": 59
+    },
+    {
+      "epoch": 0.10638297872340426,
+      "grad_norm": 6.650637149810791,
+      "learning_rate": 0.00028658318425760286,
+      "loss": 0.4737,
+      "step": 60
+    },
+    {
+      "epoch": 0.10815602836879433,
+      "grad_norm": 1.2466630935668945,
+      "learning_rate": 0.0002863148479427549,
+      "loss": 0.2634,
+      "step": 61
+    },
+    {
+      "epoch": 0.1099290780141844,
+      "grad_norm": 0.73129802942276,
+      "learning_rate": 0.00028604651162790696,
+      "loss": 0.1581,
+      "step": 62
+    },
+    {
+      "epoch": 0.11170212765957446,
+      "grad_norm": 0.8522558212280273,
+      "learning_rate": 0.000285778175313059,
+      "loss": 0.2586,
+      "step": 63
+    },
+    {
+      "epoch": 0.11347517730496454,
+      "grad_norm": 0.7632489204406738,
+      "learning_rate": 0.00028550983899821105,
+      "loss": 0.2528,
+      "step": 64
+    },
+    {
+      "epoch": 0.11524822695035461,
+      "grad_norm": 0.4096511900424957,
+      "learning_rate": 0.00028524150268336313,
+      "loss": 0.0851,
+      "step": 65
+    },
+    {
+      "epoch": 0.11702127659574468,
+      "grad_norm": 0.9045661687850952,
+      "learning_rate": 0.0002849731663685152,
+      "loss": 0.2253,
+      "step": 66
+    },
+    {
+      "epoch": 0.11879432624113476,
+      "grad_norm": 0.6839063763618469,
+      "learning_rate": 0.00028470483005366723,
+      "loss": 0.1416,
+      "step": 67
+    },
+    {
+      "epoch": 0.12056737588652482,
+      "grad_norm": 2.112931728363037,
+      "learning_rate": 0.0002844364937388193,
+      "loss": 0.2554,
+      "step": 68
+    },
+    {
+      "epoch": 0.12234042553191489,
+      "grad_norm": 1.187727928161621,
+      "learning_rate": 0.0002841681574239714,
+      "loss": 0.2891,
+      "step": 69
+    },
+    {
+      "epoch": 0.12411347517730496,
+      "grad_norm": 0.7449979782104492,
+      "learning_rate": 0.0002838998211091234,
+      "loss": 0.1743,
+      "step": 70
+    },
+    {
+      "epoch": 0.12588652482269502,
+      "grad_norm": 0.5655006766319275,
+      "learning_rate": 0.0002836314847942755,
+      "loss": 0.1247,
+      "step": 71
+    },
+    {
+      "epoch": 0.1276595744680851,
+      "grad_norm": 0.7953659296035767,
+      "learning_rate": 0.0002833631484794275,
+      "loss": 0.1277,
+      "step": 72
+    },
+    {
+      "epoch": 0.12943262411347517,
+      "grad_norm": 0.7307471632957458,
+      "learning_rate": 0.0002830948121645796,
+      "loss": 0.2575,
+      "step": 73
+    },
+    {
+      "epoch": 0.13120567375886524,
+      "grad_norm": 0.815560519695282,
+      "learning_rate": 0.0002828264758497316,
+      "loss": 0.2547,
+      "step": 74
+    },
+    {
+      "epoch": 0.13297872340425532,
+      "grad_norm": 0.7381014227867126,
+      "learning_rate": 0.0002825581395348837,
+      "loss": 0.1005,
+      "step": 75
+    },
+    {
+      "epoch": 0.1347517730496454,
+      "grad_norm": 1.0876243114471436,
+      "learning_rate": 0.00028228980322003575,
+      "loss": 0.2536,
+      "step": 76
+    },
+    {
+      "epoch": 0.13652482269503546,
+      "grad_norm": 0.6369747519493103,
+      "learning_rate": 0.00028202146690518783,
+      "loss": 0.1696,
+      "step": 77
+    },
+    {
+      "epoch": 0.13829787234042554,
+      "grad_norm": 0.5377818942070007,
+      "learning_rate": 0.0002817531305903399,
+      "loss": 0.1528,
+      "step": 78
+    },
+    {
+      "epoch": 0.1400709219858156,
+      "grad_norm": 0.5050985813140869,
+      "learning_rate": 0.0002814847942754919,
+      "loss": 0.1376,
+      "step": 79
+    },
+    {
+      "epoch": 0.14184397163120568,
+      "grad_norm": 0.46020108461380005,
+      "learning_rate": 0.000281216457960644,
+      "loss": 0.1164,
+      "step": 80
+    },
+    {
+      "epoch": 0.14361702127659576,
+      "grad_norm": 0.4778999090194702,
+      "learning_rate": 0.000280948121645796,
+      "loss": 0.1078,
+      "step": 81
+    },
+    {
+      "epoch": 0.1453900709219858,
+      "grad_norm": 0.7942756414413452,
+      "learning_rate": 0.0002806797853309481,
+      "loss": 0.2139,
+      "step": 82
+    },
+    {
+      "epoch": 0.14716312056737588,
+      "grad_norm": 0.5343471169471741,
+      "learning_rate": 0.0002804114490161001,
+      "loss": 0.1331,
+      "step": 83
+    },
+    {
+      "epoch": 0.14893617021276595,
+      "grad_norm": 0.8038661479949951,
+      "learning_rate": 0.0002801431127012522,
+      "loss": 0.1801,
+      "step": 84
+    },
+    {
+      "epoch": 0.15070921985815602,
+      "grad_norm": 0.3083706498146057,
+      "learning_rate": 0.0002798747763864043,
+      "loss": 0.0437,
+      "step": 85
+    },
+    {
+      "epoch": 0.1524822695035461,
+      "grad_norm": 0.6344548463821411,
+      "learning_rate": 0.00027960644007155635,
+      "loss": 0.1676,
+      "step": 86
+    },
+    {
+      "epoch": 0.15425531914893617,
+      "grad_norm": 0.5747407674789429,
+      "learning_rate": 0.00027933810375670837,
+      "loss": 0.1079,
+      "step": 87
+    },
+    {
+      "epoch": 0.15602836879432624,
+      "grad_norm": 0.5270193815231323,
+      "learning_rate": 0.00027906976744186045,
+      "loss": 0.1263,
+      "step": 88
+    },
+    {
+      "epoch": 0.15780141843971632,
+      "grad_norm": 0.4451313614845276,
+      "learning_rate": 0.0002788014311270125,
+      "loss": 0.0511,
+      "step": 89
+    },
+    {
+      "epoch": 0.1595744680851064,
+      "grad_norm": 0.691502034664154,
+      "learning_rate": 0.00027853309481216455,
+      "loss": 0.1637,
+      "step": 90
+    },
+    {
+      "epoch": 0.16134751773049646,
+      "grad_norm": 0.5651169419288635,
+      "learning_rate": 0.0002782647584973166,
+      "loss": 0.1371,
+      "step": 91
+    },
+    {
+      "epoch": 0.16312056737588654,
+      "grad_norm": 0.5306984782218933,
+      "learning_rate": 0.00027799642218246864,
+      "loss": 0.1535,
+      "step": 92
+    },
+    {
+      "epoch": 0.16489361702127658,
+      "grad_norm": 0.5752344727516174,
+      "learning_rate": 0.0002777280858676207,
+      "loss": 0.1622,
+      "step": 93
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 0.3410884737968445,
+      "learning_rate": 0.0002774597495527728,
+      "loss": 0.0683,
+      "step": 94
+    },
+    {
+      "epoch": 0.16843971631205673,
+      "grad_norm": 0.6347271203994751,
+      "learning_rate": 0.0002771914132379248,
+      "loss": 0.1577,
+      "step": 95
+    },
+    {
+      "epoch": 0.1702127659574468,
+      "grad_norm": 0.36072492599487305,
+      "learning_rate": 0.0002769230769230769,
+      "loss": 0.0367,
+      "step": 96
+    },
+    {
+      "epoch": 0.17198581560283688,
+      "grad_norm": 0.29759401082992554,
+      "learning_rate": 0.00027665474060822897,
+      "loss": 0.0218,
+      "step": 97
+    },
+    {
+      "epoch": 0.17375886524822695,
+      "grad_norm": 0.37029051780700684,
+      "learning_rate": 0.00027638640429338105,
+      "loss": 0.0263,
+      "step": 98
+    },
+    {
+      "epoch": 0.17553191489361702,
+      "grad_norm": 0.3705526292324066,
+      "learning_rate": 0.00027611806797853307,
+      "loss": 0.0219,
+      "step": 99
+    },
+    {
+      "epoch": 0.1773049645390071,
+      "grad_norm": 0.31463027000427246,
+      "learning_rate": 0.00027584973166368514,
+      "loss": 0.023,
+      "step": 100
+    },
+    {
+      "epoch": 0.17907801418439717,
+      "grad_norm": 1.7321122884750366,
+      "learning_rate": 0.00027558139534883717,
+      "loss": 0.6776,
+      "step": 101
+    },
+    {
+      "epoch": 0.18085106382978725,
+      "grad_norm": 1.03788423538208,
+      "learning_rate": 0.00027531305903398924,
+      "loss": 0.4406,
+      "step": 102
+    },
+    {
+      "epoch": 0.18262411347517732,
+      "grad_norm": 1.13809335231781,
+      "learning_rate": 0.00027504472271914126,
+      "loss": 0.4302,
+      "step": 103
+    },
+    {
+      "epoch": 0.18439716312056736,
+      "grad_norm": 0.6666789650917053,
+      "learning_rate": 0.00027477638640429334,
+      "loss": 0.2029,
+      "step": 104
+    },
+    {
+      "epoch": 0.18617021276595744,
+      "grad_norm": 0.9535943865776062,
+      "learning_rate": 0.0002745080500894454,
+      "loss": 0.2808,
+      "step": 105
+    },
+    {
+      "epoch": 0.1879432624113475,
+      "grad_norm": 1.2909356355667114,
+      "learning_rate": 0.0002742397137745975,
+      "loss": 0.3593,
+      "step": 106
+    },
+    {
+      "epoch": 0.18971631205673758,
+      "grad_norm": 0.9249064326286316,
+      "learning_rate": 0.0002739713774597495,
+      "loss": 0.355,
+      "step": 107
+    },
+    {
+      "epoch": 0.19148936170212766,
+      "grad_norm": 0.5897863507270813,
+      "learning_rate": 0.0002737030411449016,
+      "loss": 0.2349,
+      "step": 108
+    },
+    {
+      "epoch": 0.19326241134751773,
+      "grad_norm": 1.060029149055481,
+      "learning_rate": 0.00027343470483005367,
+      "loss": 0.2517,
+      "step": 109
+    },
+    {
+      "epoch": 0.1950354609929078,
+      "grad_norm": 0.6262145638465881,
+      "learning_rate": 0.0002731663685152057,
+      "loss": 0.2119,
+      "step": 110
+    },
+    {
+      "epoch": 0.19680851063829788,
+      "grad_norm": 0.7881852984428406,
+      "learning_rate": 0.00027289803220035777,
+      "loss": 0.2792,
+      "step": 111
+    },
+    {
+      "epoch": 0.19858156028368795,
+      "grad_norm": 0.7693052887916565,
+      "learning_rate": 0.0002726296958855098,
+      "loss": 0.2824,
+      "step": 112
+    },
+    {
+      "epoch": 0.20035460992907803,
+      "grad_norm": 0.7235965132713318,
+      "learning_rate": 0.00027236135957066186,
+      "loss": 0.0918,
+      "step": 113
+    },
+    {
+      "epoch": 0.20212765957446807,
+      "grad_norm": 1.0118364095687866,
+      "learning_rate": 0.00027209302325581394,
+      "loss": 0.211,
+      "step": 114
+    },
+    {
+      "epoch": 0.20390070921985815,
+      "grad_norm": 0.9211074709892273,
+      "learning_rate": 0.00027182468694096596,
+      "loss": 0.2766,
+      "step": 115
+    },
+    {
+      "epoch": 0.20567375886524822,
+      "grad_norm": 1.3270055055618286,
+      "learning_rate": 0.00027155635062611804,
+      "loss": 0.1172,
+      "step": 116
+    },
+    {
+      "epoch": 0.2074468085106383,
+      "grad_norm": 0.6904440522193909,
+      "learning_rate": 0.0002712880143112701,
+      "loss": 0.2208,
+      "step": 117
+    },
+    {
+      "epoch": 0.20921985815602837,
+      "grad_norm": 0.44743916392326355,
+      "learning_rate": 0.0002710196779964222,
+      "loss": 0.1265,
+      "step": 118
+    },
+    {
+      "epoch": 0.21099290780141844,
+      "grad_norm": 1.5493639707565308,
+      "learning_rate": 0.0002707513416815742,
+      "loss": 0.2085,
+      "step": 119
+    },
+    {
+      "epoch": 0.2127659574468085,
+      "grad_norm": 0.7296533584594727,
+      "learning_rate": 0.0002704830053667263,
+      "loss": 0.253,
+      "step": 120
+    },
+    {
+      "epoch": 0.21453900709219859,
+      "grad_norm": 1.300660490989685,
+      "learning_rate": 0.0002702146690518783,
+      "loss": 0.2164,
+      "step": 121
+    },
+    {
+      "epoch": 0.21631205673758866,
+      "grad_norm": 0.9226039052009583,
+      "learning_rate": 0.0002699463327370304,
+      "loss": 0.2533,
+      "step": 122
+    },
+    {
+      "epoch": 0.21808510638297873,
+      "grad_norm": 0.9466855525970459,
+      "learning_rate": 0.0002696779964221824,
+      "loss": 0.2528,
+      "step": 123
+    },
+    {
+      "epoch": 0.2198581560283688,
+      "grad_norm": 0.6484300494194031,
+      "learning_rate": 0.0002694096601073345,
+      "loss": 0.1598,
+      "step": 124
+    },
+    {
+      "epoch": 0.22163120567375885,
+      "grad_norm": 0.7165495157241821,
+      "learning_rate": 0.00026914132379248656,
+      "loss": 0.2448,
+      "step": 125
+    },
+    {
+      "epoch": 0.22340425531914893,
+      "grad_norm": 1.370329737663269,
+      "learning_rate": 0.00026887298747763864,
+      "loss": 0.3169,
+      "step": 126
+    },
+    {
+      "epoch": 0.225177304964539,
+      "grad_norm": 0.41645482182502747,
+      "learning_rate": 0.00026860465116279066,
+      "loss": 0.1352,
+      "step": 127
+    },
+    {
+      "epoch": 0.22695035460992907,
+      "grad_norm": 0.4441884458065033,
+      "learning_rate": 0.00026833631484794273,
+      "loss": 0.1376,
+      "step": 128
+    },
+    {
+      "epoch": 0.22872340425531915,
+      "grad_norm": 0.3925491273403168,
+      "learning_rate": 0.0002680679785330948,
+      "loss": 0.1194,
+      "step": 129
+    },
+    {
+      "epoch": 0.23049645390070922,
+      "grad_norm": 0.3546803295612335,
+      "learning_rate": 0.00026779964221824683,
+      "loss": 0.0936,
+      "step": 130
+    },
+    {
+      "epoch": 0.2322695035460993,
+      "grad_norm": 0.683130145072937,
+      "learning_rate": 0.0002675313059033989,
+      "loss": 0.2044,
+      "step": 131
+    },
+    {
+      "epoch": 0.23404255319148937,
+      "grad_norm": 0.666155219078064,
+      "learning_rate": 0.00026726296958855093,
+      "loss": 0.2361,
+      "step": 132
+    },
+    {
+      "epoch": 0.23581560283687944,
+      "grad_norm": 0.5066004991531372,
+      "learning_rate": 0.000266994633273703,
+      "loss": 0.1449,
+      "step": 133
+    },
+    {
+      "epoch": 0.2375886524822695,
+      "grad_norm": 0.745506763458252,
+      "learning_rate": 0.0002667262969588551,
+      "loss": 0.1987,
+      "step": 134
+    },
+    {
+      "epoch": 0.2393617021276596,
+      "grad_norm": 0.9249691963195801,
+      "learning_rate": 0.0002664579606440071,
+      "loss": 0.2485,
+      "step": 135
+    },
+    {
+      "epoch": 0.24113475177304963,
+      "grad_norm": 0.5891656875610352,
+      "learning_rate": 0.0002661896243291592,
+      "loss": 0.2172,
+      "step": 136
+    },
+    {
+      "epoch": 0.2429078014184397,
+      "grad_norm": 0.6771588325500488,
+      "learning_rate": 0.00026592128801431126,
+      "loss": 0.2293,
+      "step": 137
+    },
+    {
+      "epoch": 0.24468085106382978,
+      "grad_norm": 1.0185132026672363,
+      "learning_rate": 0.00026565295169946333,
+      "loss": 0.2819,
+      "step": 138
+    },
+    {
+      "epoch": 0.24645390070921985,
+      "grad_norm": 0.357791930437088,
+      "learning_rate": 0.00026538461538461536,
+      "loss": 0.0876,
+      "step": 139
+    },
+    {
+      "epoch": 0.24822695035460993,
+      "grad_norm": 0.5244898200035095,
+      "learning_rate": 0.00026511627906976743,
+      "loss": 0.0989,
+      "step": 140
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.553326427936554,
+      "learning_rate": 0.00026484794275491945,
+      "loss": 0.1503,
+      "step": 141
+    },
+    {
+      "epoch": 0.25177304964539005,
+      "grad_norm": 0.7104494571685791,
+      "learning_rate": 0.00026457960644007153,
+      "loss": 0.2487,
+      "step": 142
+    },
+    {
+      "epoch": 0.25354609929078015,
+      "grad_norm": 0.5205265879631042,
+      "learning_rate": 0.0002643112701252236,
+      "loss": 0.1185,
+      "step": 143
+    },
+    {
+      "epoch": 0.2553191489361702,
+      "grad_norm": 0.41840335726737976,
+      "learning_rate": 0.00026404293381037563,
+      "loss": 0.0841,
+      "step": 144
+    },
+    {
+      "epoch": 0.2570921985815603,
+      "grad_norm": 0.572222888469696,
+      "learning_rate": 0.0002637745974955277,
+      "loss": 0.1331,
+      "step": 145
+    },
+    {
+      "epoch": 0.25886524822695034,
+      "grad_norm": 0.8182592391967773,
+      "learning_rate": 0.0002635062611806798,
+      "loss": 0.2042,
+      "step": 146
+    },
+    {
+      "epoch": 0.26063829787234044,
+      "grad_norm": 0.5115887522697449,
+      "learning_rate": 0.0002632379248658318,
+      "loss": 0.1148,
+      "step": 147
+    },
+    {
+      "epoch": 0.2624113475177305,
+      "grad_norm": 0.5078725218772888,
+      "learning_rate": 0.0002629695885509839,
+      "loss": 0.0762,
+      "step": 148
+    },
+    {
+      "epoch": 0.2641843971631206,
+      "grad_norm": 0.22413299977779388,
+      "learning_rate": 0.00026270125223613595,
+      "loss": 0.0355,
+      "step": 149
+    },
+    {
+      "epoch": 0.26595744680851063,
+      "grad_norm": 0.25390034914016724,
+      "learning_rate": 0.000262432915921288,
+      "loss": 0.0319,
+      "step": 150
+    },
+    {
+      "epoch": 0.26773049645390073,
+      "grad_norm": 3.294741153717041,
+      "learning_rate": 0.00026216457960644005,
+      "loss": 0.4876,
+      "step": 151
+    },
+    {
+      "epoch": 0.2695035460992908,
+      "grad_norm": 1.5393906831741333,
+      "learning_rate": 0.0002618962432915921,
+      "loss": 0.2547,
+      "step": 152
+    },
+    {
+      "epoch": 0.2712765957446808,
+      "grad_norm": 1.7025604248046875,
+      "learning_rate": 0.00026162790697674415,
+      "loss": 0.4747,
+      "step": 153
+    },
+    {
+      "epoch": 0.2730496453900709,
+      "grad_norm": 0.8344413638114929,
+      "learning_rate": 0.0002613595706618962,
+      "loss": 0.2431,
+      "step": 154
+    },
+    {
+      "epoch": 0.274822695035461,
+      "grad_norm": 0.9473798871040344,
+      "learning_rate": 0.00026109123434704825,
+      "loss": 0.2135,
+      "step": 155
+    },
+    {
+      "epoch": 0.2765957446808511,
+      "grad_norm": 1.0257476568222046,
+      "learning_rate": 0.0002608228980322003,
+      "loss": 0.1833,
+      "step": 156
+    },
+    {
+      "epoch": 0.2783687943262411,
+      "grad_norm": 1.2221112251281738,
+      "learning_rate": 0.0002605545617173524,
+      "loss": 0.3174,
+      "step": 157
+    },
+    {
+      "epoch": 0.2801418439716312,
+      "grad_norm": 0.5619825124740601,
+      "learning_rate": 0.0002602862254025045,
+      "loss": 0.1492,
+      "step": 158
+    },
+    {
+      "epoch": 0.28191489361702127,
+      "grad_norm": 1.1845368146896362,
+      "learning_rate": 0.0002600178890876565,
+      "loss": 0.4336,
+      "step": 159
+    },
+    {
+      "epoch": 0.28368794326241137,
+      "grad_norm": 1.2730036973953247,
+      "learning_rate": 0.0002597495527728086,
+      "loss": 0.3466,
+      "step": 160
+    },
+    {
+      "epoch": 0.2854609929078014,
+      "grad_norm": 0.4464946985244751,
+      "learning_rate": 0.0002594812164579606,
+      "loss": 0.1239,
+      "step": 161
+    },
+    {
+      "epoch": 0.2872340425531915,
+      "grad_norm": 0.5267930030822754,
+      "learning_rate": 0.00025921288014311267,
+      "loss": 0.1372,
+      "step": 162
+    },
+    {
+      "epoch": 0.28900709219858156,
+      "grad_norm": 0.4249666929244995,
+      "learning_rate": 0.00025894454382826475,
+      "loss": 0.1402,
+      "step": 163
+    },
+    {
+      "epoch": 0.2907801418439716,
+      "grad_norm": 0.7619920372962952,
+      "learning_rate": 0.00025867620751341677,
+      "loss": 0.2309,
+      "step": 164
+    },
+    {
+      "epoch": 0.2925531914893617,
+      "grad_norm": 0.9128990769386292,
+      "learning_rate": 0.00025840787119856885,
+      "loss": 0.2832,
+      "step": 165
+    },
+    {
+      "epoch": 0.29432624113475175,
+      "grad_norm": 0.4988587200641632,
+      "learning_rate": 0.0002581395348837209,
+      "loss": 0.1517,
+      "step": 166
+    },
+    {
+      "epoch": 0.29609929078014185,
+      "grad_norm": 0.7568597793579102,
+      "learning_rate": 0.000257871198568873,
+      "loss": 0.2449,
+      "step": 167
+    },
+    {
+      "epoch": 0.2978723404255319,
+      "grad_norm": 0.7556333541870117,
+      "learning_rate": 0.000257602862254025,
+      "loss": 0.2591,
+      "step": 168
+    },
+    {
+      "epoch": 0.299645390070922,
+      "grad_norm": 0.9023774266242981,
+      "learning_rate": 0.0002573345259391771,
+      "loss": 0.239,
+      "step": 169
+    },
+    {
+      "epoch": 0.30141843971631205,
+      "grad_norm": 0.490642786026001,
+      "learning_rate": 0.0002570661896243291,
+      "loss": 0.1508,
+      "step": 170
+    },
+    {
+      "epoch": 0.30319148936170215,
+      "grad_norm": 0.6215077042579651,
+      "learning_rate": 0.0002567978533094812,
+      "loss": 0.2277,
+      "step": 171
+    },
+    {
+      "epoch": 0.3049645390070922,
+      "grad_norm": 0.4409044086933136,
+      "learning_rate": 0.00025652951699463327,
+      "loss": 0.1354,
+      "step": 172
+    },
+    {
+      "epoch": 0.3067375886524823,
+      "grad_norm": 0.420749694108963,
+      "learning_rate": 0.0002562611806797853,
+      "loss": 0.068,
+      "step": 173
+    },
+    {
+      "epoch": 0.30851063829787234,
+      "grad_norm": 0.40215224027633667,
+      "learning_rate": 0.00025599284436493737,
+      "loss": 0.0707,
+      "step": 174
+    },
+    {
+      "epoch": 0.3102836879432624,
+      "grad_norm": 0.7107430696487427,
+      "learning_rate": 0.0002557245080500894,
+      "loss": 0.1413,
+      "step": 175
+    },
+    {
+      "epoch": 0.3120567375886525,
+      "grad_norm": 0.9405664801597595,
+      "learning_rate": 0.00025545617173524147,
+      "loss": 0.1555,
+      "step": 176
+    },
+    {
+      "epoch": 0.31382978723404253,
+      "grad_norm": 0.9159305095672607,
+      "learning_rate": 0.00025518783542039354,
+      "loss": 0.1827,
+      "step": 177
+    },
+    {
+      "epoch": 0.31560283687943264,
+      "grad_norm": 1.2095725536346436,
+      "learning_rate": 0.0002549194991055456,
+      "loss": 0.2713,
+      "step": 178
+    },
+    {
+      "epoch": 0.3173758865248227,
+      "grad_norm": 0.5971267223358154,
+      "learning_rate": 0.00025465116279069764,
+      "loss": 0.2094,
+      "step": 179
+    },
+    {
+      "epoch": 0.3191489361702128,
+      "grad_norm": 0.5078379511833191,
+      "learning_rate": 0.0002543828264758497,
+      "loss": 0.1102,
+      "step": 180
+    },
+    {
+      "epoch": 0.32092198581560283,
+      "grad_norm": 0.5540158748626709,
+      "learning_rate": 0.0002541144901610018,
+      "loss": 0.1408,
+      "step": 181
+    },
+    {
+      "epoch": 0.32269503546099293,
+      "grad_norm": 0.3938644826412201,
+      "learning_rate": 0.0002538461538461538,
+      "loss": 0.1044,
+      "step": 182
+    },
+    {
+      "epoch": 0.324468085106383,
+      "grad_norm": 0.43856045603752136,
+      "learning_rate": 0.0002535778175313059,
+      "loss": 0.1069,
+      "step": 183
+    },
+    {
+      "epoch": 0.3262411347517731,
+      "grad_norm": 0.6097897887229919,
+      "learning_rate": 0.0002533094812164579,
+      "loss": 0.1447,
+      "step": 184
+    },
+    {
+      "epoch": 0.3280141843971631,
+      "grad_norm": 0.6639567613601685,
+      "learning_rate": 0.00025304114490161,
+      "loss": 0.1622,
+      "step": 185
+    },
+    {
+      "epoch": 0.32978723404255317,
+      "grad_norm": 0.4464520812034607,
+      "learning_rate": 0.00025277280858676207,
+      "loss": 0.1113,
+      "step": 186
+    },
+    {
+      "epoch": 0.33156028368794327,
+      "grad_norm": 0.45327168703079224,
+      "learning_rate": 0.00025250447227191414,
+      "loss": 0.1121,
+      "step": 187
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.4880326986312866,
+      "learning_rate": 0.00025223613595706616,
+      "loss": 0.1055,
+      "step": 188
+    },
+    {
+      "epoch": 0.3351063829787234,
+      "grad_norm": 0.44586634635925293,
+      "learning_rate": 0.00025196779964221824,
+      "loss": 0.1544,
+      "step": 189
+    },
+    {
+      "epoch": 0.33687943262411346,
+      "grad_norm": 0.7135107517242432,
+      "learning_rate": 0.00025169946332737026,
+      "loss": 0.1047,
+      "step": 190
+    },
+    {
+      "epoch": 0.33865248226950356,
+      "grad_norm": 0.5775635838508606,
+      "learning_rate": 0.00025143112701252234,
+      "loss": 0.175,
+      "step": 191
+    },
+    {
+      "epoch": 0.3404255319148936,
+      "grad_norm": 0.6252586841583252,
+      "learning_rate": 0.0002511627906976744,
+      "loss": 0.1478,
+      "step": 192
+    },
+    {
+      "epoch": 0.3421985815602837,
+      "grad_norm": 2.061342477798462,
+      "learning_rate": 0.00025089445438282644,
+      "loss": 0.1931,
+      "step": 193
+    },
+    {
+      "epoch": 0.34397163120567376,
+      "grad_norm": 0.47072601318359375,
+      "learning_rate": 0.0002506261180679785,
+      "loss": 0.0928,
+      "step": 194
+    },
+    {
+      "epoch": 0.34574468085106386,
+      "grad_norm": 0.5164952278137207,
+      "learning_rate": 0.00025035778175313053,
+      "loss": 0.1287,
+      "step": 195
+    },
+    {
+      "epoch": 0.3475177304964539,
+      "grad_norm": 0.44344618916511536,
+      "learning_rate": 0.0002500894454382826,
+      "loss": 0.0924,
+      "step": 196
+    },
+    {
+      "epoch": 0.34929078014184395,
+      "grad_norm": 0.4380563795566559,
+      "learning_rate": 0.0002498211091234347,
+      "loss": 0.1393,
+      "step": 197
+    },
+    {
+      "epoch": 0.35106382978723405,
+      "grad_norm": 0.3074440360069275,
+      "learning_rate": 0.00024955277280858676,
+      "loss": 0.04,
+      "step": 198
+    },
+    {
+      "epoch": 0.3528368794326241,
+      "grad_norm": 0.25228482484817505,
+      "learning_rate": 0.0002492844364937388,
+      "loss": 0.037,
+      "step": 199
+    },
+    {
+      "epoch": 0.3546099290780142,
+      "grad_norm": 0.4954518973827362,
+      "learning_rate": 0.00024901610017889086,
+      "loss": 0.1563,
+      "step": 200
+    },
+    {
+      "epoch": 0.35638297872340424,
+      "grad_norm": 9.530560493469238,
+      "learning_rate": 0.00024874776386404294,
+      "loss": 0.6406,
+      "step": 201
+    },
+    {
+      "epoch": 0.35815602836879434,
+      "grad_norm": 1.2426095008850098,
+      "learning_rate": 0.00024847942754919496,
+      "loss": 0.5336,
+      "step": 202
+    },
+    {
+      "epoch": 0.3599290780141844,
+      "grad_norm": 1.4315325021743774,
+      "learning_rate": 0.00024821109123434703,
+      "loss": 0.4691,
+      "step": 203
+    },
+    {
+      "epoch": 0.3617021276595745,
+      "grad_norm": 1.3048845529556274,
+      "learning_rate": 0.00024794275491949906,
+      "loss": 0.4098,
+      "step": 204
+    },
+    {
+      "epoch": 0.36347517730496454,
+      "grad_norm": 1.1803832054138184,
+      "learning_rate": 0.00024767441860465113,
+      "loss": 0.3023,
+      "step": 205
+    },
+    {
+      "epoch": 0.36524822695035464,
+      "grad_norm": 0.895564079284668,
+      "learning_rate": 0.0002474060822898032,
+      "loss": 0.2474,
+      "step": 206
+    },
+    {
+      "epoch": 0.3670212765957447,
+      "grad_norm": 0.7290531992912292,
+      "learning_rate": 0.0002471377459749553,
+      "loss": 0.2365,
+      "step": 207
+    },
+    {
+      "epoch": 0.36879432624113473,
+      "grad_norm": 0.5765687823295593,
+      "learning_rate": 0.0002468694096601073,
+      "loss": 0.2233,
+      "step": 208
+    },
+    {
+      "epoch": 0.37056737588652483,
+      "grad_norm": 1.518298625946045,
+      "learning_rate": 0.0002466010733452594,
+      "loss": 0.3787,
+      "step": 209
+    },
+    {
+      "epoch": 0.3723404255319149,
+      "grad_norm": 0.9093344211578369,
+      "learning_rate": 0.00024633273703041146,
+      "loss": 0.2682,
+      "step": 210
+    },
+    {
+      "epoch": 0.374113475177305,
+      "grad_norm": 0.9133920669555664,
+      "learning_rate": 0.0002460644007155635,
+      "loss": 0.2236,
+      "step": 211
+    },
+    {
+      "epoch": 0.375886524822695,
+      "grad_norm": 0.9125071167945862,
+      "learning_rate": 0.00024579606440071556,
+      "loss": 0.344,
+      "step": 212
+    },
+    {
+      "epoch": 0.3776595744680851,
+      "grad_norm": 0.9353806972503662,
+      "learning_rate": 0.0002455277280858676,
+      "loss": 0.3213,
+      "step": 213
+    },
+    {
+      "epoch": 0.37943262411347517,
+      "grad_norm": 0.32626235485076904,
+      "learning_rate": 0.00024525939177101966,
+      "loss": 0.089,
+      "step": 214
+    },
+    {
+      "epoch": 0.38120567375886527,
+      "grad_norm": 0.722865104675293,
+      "learning_rate": 0.0002449910554561717,
+      "loss": 0.2475,
+      "step": 215
+    },
+    {
+      "epoch": 0.3829787234042553,
+      "grad_norm": 1.2499586343765259,
+      "learning_rate": 0.00024472271914132375,
+      "loss": 0.1726,
+      "step": 216
+    },
+    {
+      "epoch": 0.38475177304964536,
+      "grad_norm": 0.8918619155883789,
+      "learning_rate": 0.00024445438282647583,
+      "loss": 0.1832,
+      "step": 217
+    },
+    {
+      "epoch": 0.38652482269503546,
+      "grad_norm": 0.9703326225280762,
+      "learning_rate": 0.0002441860465116279,
+      "loss": 0.3934,
+      "step": 218
+    },
+    {
+      "epoch": 0.3882978723404255,
+      "grad_norm": 0.9967647194862366,
+      "learning_rate": 0.00024391771019677995,
+      "loss": 0.2148,
+      "step": 219
+    },
+    {
+      "epoch": 0.3900709219858156,
+      "grad_norm": 0.6736385226249695,
+      "learning_rate": 0.000243649373881932,
+      "loss": 0.2019,
+      "step": 220
+    },
+    {
+      "epoch": 0.39184397163120566,
+      "grad_norm": 0.6814044117927551,
+      "learning_rate": 0.00024338103756708408,
+      "loss": 0.3198,
+      "step": 221
+    },
+    {
+      "epoch": 0.39361702127659576,
+      "grad_norm": 0.6858790516853333,
+      "learning_rate": 0.0002431127012522361,
+      "loss": 0.2017,
+      "step": 222
+    },
+    {
+      "epoch": 0.3953900709219858,
+      "grad_norm": 0.4301723837852478,
+      "learning_rate": 0.00024284436493738818,
+      "loss": 0.1283,
+      "step": 223
+    },
+    {
+      "epoch": 0.3971631205673759,
+      "grad_norm": 0.47749027609825134,
+      "learning_rate": 0.00024257602862254023,
+      "loss": 0.1513,
+      "step": 224
+    },
+    {
+      "epoch": 0.39893617021276595,
+      "grad_norm": 0.5345844626426697,
+      "learning_rate": 0.0002423076923076923,
+      "loss": 0.0872,
+      "step": 225
+    },
+    {
+      "epoch": 0.40070921985815605,
+      "grad_norm": 0.5137667059898376,
+      "learning_rate": 0.00024203935599284433,
+      "loss": 0.0726,
+      "step": 226
+    },
+    {
+      "epoch": 0.40070921985815605,
+      "eval_loss": 0.08563709259033203,
+      "eval_runtime": 47.4226,
+      "eval_samples_per_second": 2.657,
+      "eval_steps_per_second": 2.657,
+      "step": 226
+    },
+    {
+      "epoch": 0.4024822695035461,
+      "grad_norm": 0.5459873080253601,
+      "learning_rate": 0.0002417710196779964,
+      "loss": 0.1341,
+      "step": 227
+    },
+    {
+      "epoch": 0.40425531914893614,
+      "grad_norm": 0.7590123414993286,
+      "learning_rate": 0.00024150268336314848,
+      "loss": 0.2418,
+      "step": 228
+    },
+    {
+      "epoch": 0.40602836879432624,
+      "grad_norm": 0.8846575617790222,
+      "learning_rate": 0.00024123434704830053,
+      "loss": 0.1606,
+      "step": 229
+    },
+    {
+      "epoch": 0.4078014184397163,
+      "grad_norm": 0.7346479296684265,
+      "learning_rate": 0.00024096601073345258,
+      "loss": 0.2458,
+      "step": 230
+    },
+    {
+      "epoch": 0.4095744680851064,
+      "grad_norm": 0.7012773752212524,
+      "learning_rate": 0.00024069767441860462,
+      "loss": 0.1353,
+      "step": 231
+    },
+    {
+      "epoch": 0.41134751773049644,
+      "grad_norm": 0.5895172357559204,
+      "learning_rate": 0.0002404293381037567,
+      "loss": 0.2138,
+      "step": 232
+    },
+    {
+      "epoch": 0.41312056737588654,
+      "grad_norm": 0.6732155680656433,
+      "learning_rate": 0.00024016100178890872,
+      "loss": 0.0906,
+      "step": 233
+    },
+    {
+      "epoch": 0.4148936170212766,
+      "grad_norm": 0.38057422637939453,
+      "learning_rate": 0.0002398926654740608,
+      "loss": 0.0973,
+      "step": 234
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.4375483989715576,
+      "learning_rate": 0.00023962432915921285,
+      "loss": 0.0895,
+      "step": 235
+    },
+    {
+      "epoch": 0.41843971631205673,
+      "grad_norm": 0.45380473136901855,
+      "learning_rate": 0.00023935599284436492,
+      "loss": 0.1046,
+      "step": 236
+    },
+    {
+      "epoch": 0.42021276595744683,
+      "grad_norm": 0.38564634323120117,
+      "learning_rate": 0.00023908765652951695,
+      "loss": 0.0899,
+      "step": 237
+    },
+    {
+      "epoch": 0.4219858156028369,
+      "grad_norm": 0.837587296962738,
+      "learning_rate": 0.00023881932021466902,
+      "loss": 0.2611,
+      "step": 238
+    },
+    {
+      "epoch": 0.4237588652482269,
+      "grad_norm": 0.9749343991279602,
+      "learning_rate": 0.0002385509838998211,
+      "loss": 0.1351,
+      "step": 239
+    },
+    {
+      "epoch": 0.425531914893617,
+      "grad_norm": 0.7198343873023987,
+      "learning_rate": 0.00023828264758497315,
+      "loss": 0.1716,
+      "step": 240
+    },
+    {
+      "epoch": 0.42730496453900707,
+      "grad_norm": 0.6409445405006409,
+      "learning_rate": 0.00023801431127012522,
+      "loss": 0.1539,
+      "step": 241
+    },
+    {
+      "epoch": 0.42907801418439717,
+      "grad_norm": 0.6468048095703125,
+      "learning_rate": 0.00023774597495527724,
+      "loss": 0.2017,
+      "step": 242
+    },
+    {
+      "epoch": 0.4308510638297872,
+      "grad_norm": 0.3550443947315216,
+      "learning_rate": 0.00023747763864042932,
+      "loss": 0.102,
+      "step": 243
+    },
+    {
+      "epoch": 0.4326241134751773,
+      "grad_norm": 0.5282692313194275,
+      "learning_rate": 0.00023720930232558137,
+      "loss": 0.0795,
+      "step": 244
+    },
+    {
+      "epoch": 0.43439716312056736,
+      "grad_norm": 0.4462401568889618,
+      "learning_rate": 0.00023694096601073345,
+      "loss": 0.1051,
+      "step": 245
+    },
+    {
+      "epoch": 0.43617021276595747,
+      "grad_norm": 0.33794382214546204,
+      "learning_rate": 0.00023667262969588547,
+      "loss": 0.0688,
+      "step": 246
+    },
+    {
+      "epoch": 0.4379432624113475,
+      "grad_norm": 0.3550000786781311,
+      "learning_rate": 0.00023640429338103754,
+      "loss": 0.0511,
+      "step": 247
+    },
+    {
+      "epoch": 0.4397163120567376,
+      "grad_norm": 0.2764873206615448,
+      "learning_rate": 0.00023613595706618962,
+      "loss": 0.0336,
+      "step": 248
+    },
+    {
+      "epoch": 0.44148936170212766,
+      "grad_norm": 0.2158241719007492,
+      "learning_rate": 0.00023586762075134167,
+      "loss": 0.0275,
+      "step": 249
+    },
+    {
+      "epoch": 0.4432624113475177,
+      "grad_norm": 0.41014501452445984,
+      "learning_rate": 0.00023559928443649372,
+      "loss": 0.1071,
+      "step": 250
+    },
+    {
+      "epoch": 0.4450354609929078,
+      "grad_norm": 1.6365233659744263,
+      "learning_rate": 0.00023533094812164577,
+      "loss": 0.5651,
+      "step": 251
+    },
+    {
+      "epoch": 0.44680851063829785,
+      "grad_norm": 2.2182910442352295,
+      "learning_rate": 0.00023506261180679784,
+      "loss": 0.4931,
+      "step": 252
+    },
+    {
+      "epoch": 0.44858156028368795,
+      "grad_norm": 0.7922513484954834,
+      "learning_rate": 0.0002347942754919499,
+      "loss": 0.165,
+      "step": 253
+    },
+    {
+      "epoch": 0.450354609929078,
+      "grad_norm": 1.0960763692855835,
+      "learning_rate": 0.00023452593917710194,
+      "loss": 0.2092,
+      "step": 254
+    },
+    {
+      "epoch": 0.4521276595744681,
+      "grad_norm": 0.9665970802307129,
+      "learning_rate": 0.000234257602862254,
+      "loss": 0.3383,
+      "step": 255
+    },
+    {
+      "epoch": 0.45390070921985815,
+      "grad_norm": 0.9593271613121033,
+      "learning_rate": 0.00023398926654740607,
+      "loss": 0.2608,
+      "step": 256
+    },
+    {
+      "epoch": 0.45567375886524825,
+      "grad_norm": 0.9603182673454285,
+      "learning_rate": 0.00023372093023255814,
+      "loss": 0.3786,
+      "step": 257
+    },
+    {
+      "epoch": 0.4574468085106383,
+      "grad_norm": 0.6080289483070374,
+      "learning_rate": 0.00023345259391771016,
+      "loss": 0.1903,
+      "step": 258
+    },
+    {
+      "epoch": 0.4592198581560284,
+      "grad_norm": 0.4713785946369171,
+      "learning_rate": 0.00023318425760286224,
+      "loss": 0.1438,
+      "step": 259
+    },
+    {
+      "epoch": 0.46099290780141844,
+      "grad_norm": 0.7215020656585693,
+      "learning_rate": 0.0002329159212880143,
+      "loss": 0.1979,
+      "step": 260
+    },
+    {
+      "epoch": 0.4627659574468085,
+      "grad_norm": 0.7383154630661011,
+      "learning_rate": 0.00023264758497316637,
+      "loss": 0.3316,
+      "step": 261
+    },
+    {
+      "epoch": 0.4645390070921986,
+      "grad_norm": 0.8826196789741516,
+      "learning_rate": 0.0002323792486583184,
+      "loss": 0.3337,
+      "step": 262
+    },
+    {
+      "epoch": 0.46631205673758863,
+      "grad_norm": 0.6072385907173157,
+      "learning_rate": 0.00023211091234347046,
+      "loss": 0.2585,
+      "step": 263
+    },
+    {
+      "epoch": 0.46808510638297873,
+      "grad_norm": 0.5861011743545532,
+      "learning_rate": 0.0002318425760286225,
+      "loss": 0.1952,
+      "step": 264
+    },
+    {
+      "epoch": 0.4698581560283688,
+      "grad_norm": 0.8113440871238708,
+      "learning_rate": 0.0002315742397137746,
+      "loss": 0.1877,
+      "step": 265
+    },
+    {
+      "epoch": 0.4716312056737589,
+      "grad_norm": 0.3667961359024048,
+      "learning_rate": 0.00023130590339892664,
+      "loss": 0.0977,
+      "step": 266
+    },
+    {
+      "epoch": 0.4734042553191489,
+      "grad_norm": 0.4013451933860779,
+      "learning_rate": 0.0002310375670840787,
+      "loss": 0.0988,
+      "step": 267
+    },
+    {
+      "epoch": 0.475177304964539,
+      "grad_norm": 0.4402759373188019,
+      "learning_rate": 0.00023076923076923076,
+      "loss": 0.1747,
+      "step": 268
+    },
+    {
+      "epoch": 0.4769503546099291,
+      "grad_norm": 0.5780351161956787,
+      "learning_rate": 0.0002305008944543828,
+      "loss": 0.1446,
+      "step": 269
+    },
+    {
+      "epoch": 0.4787234042553192,
+      "grad_norm": 1.0400667190551758,
+      "learning_rate": 0.00023023255813953486,
+      "loss": 0.3404,
+      "step": 270
+    },
+    {
+      "epoch": 0.4804964539007092,
+      "grad_norm": 0.734530508518219,
+      "learning_rate": 0.0002299642218246869,
+      "loss": 0.1938,
+      "step": 271
+    },
+    {
+      "epoch": 0.48226950354609927,
+      "grad_norm": 0.4358377158641815,
+      "learning_rate": 0.000229695885509839,
+      "loss": 0.1514,
+      "step": 272
+    },
+    {
+      "epoch": 0.48404255319148937,
+      "grad_norm": 0.766135036945343,
+      "learning_rate": 0.00022942754919499104,
+      "loss": 0.2335,
+      "step": 273
+    },
+    {
+      "epoch": 0.4858156028368794,
+      "grad_norm": 0.8013383746147156,
+      "learning_rate": 0.00022915921288014308,
+      "loss": 0.2433,
+      "step": 274
+    },
+    {
+      "epoch": 0.4875886524822695,
+      "grad_norm": 0.48911821842193604,
+      "learning_rate": 0.00022889087656529513,
+      "loss": 0.1466,
+      "step": 275
+    },
+    {
+      "epoch": 0.48936170212765956,
+      "grad_norm": 0.3731197118759155,
+      "learning_rate": 0.0002286225402504472,
+      "loss": 0.095,
+      "step": 276
+    },
+    {
+      "epoch": 0.49113475177304966,
+      "grad_norm": 0.4375539720058441,
+      "learning_rate": 0.00022835420393559929,
+      "loss": 0.1066,
+      "step": 277
+    },
+    {
+      "epoch": 0.4929078014184397,
+      "grad_norm": 0.5769888758659363,
+      "learning_rate": 0.0002280858676207513,
+      "loss": 0.1959,
+      "step": 278
+    },
+    {
+      "epoch": 0.4946808510638298,
+      "grad_norm": 0.6133276224136353,
+      "learning_rate": 0.00022781753130590338,
+      "loss": 0.1216,
+      "step": 279
+    },
+    {
+      "epoch": 0.49645390070921985,
+      "grad_norm": 0.8098652362823486,
+      "learning_rate": 0.00022754919499105543,
+      "loss": 0.223,
+      "step": 280
+    },
+    {
+      "epoch": 0.49822695035460995,
+      "grad_norm": 0.38705819845199585,
+      "learning_rate": 0.0002272808586762075,
+      "loss": 0.0888,
+      "step": 281
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.37354519963264465,
+      "learning_rate": 0.00022701252236135953,
+      "loss": 0.1013,
+      "step": 282
+    },
+    {
+      "epoch": 0.50177304964539,
+      "grad_norm": 0.6955541968345642,
+      "learning_rate": 0.0002267441860465116,
+      "loss": 0.1746,
+      "step": 283
+    },
+    {
+      "epoch": 0.5035460992907801,
+      "grad_norm": 0.44892531633377075,
+      "learning_rate": 0.00022647584973166366,
+      "loss": 0.123,
+      "step": 284
+    },
+    {
+      "epoch": 0.5053191489361702,
+      "grad_norm": 0.7901389598846436,
+      "learning_rate": 0.00022620751341681573,
+      "loss": 0.2863,
+      "step": 285
+    },
+    {
+      "epoch": 0.5070921985815603,
+      "grad_norm": 0.6648136973381042,
+      "learning_rate": 0.00022593917710196778,
+      "loss": 0.1577,
+      "step": 286
+    },
+    {
+      "epoch": 0.5088652482269503,
+      "grad_norm": 0.3378174602985382,
+      "learning_rate": 0.00022567084078711983,
+      "loss": 0.0916,
+      "step": 287
+    },
+    {
+      "epoch": 0.5106382978723404,
+      "grad_norm": 0.46413183212280273,
+      "learning_rate": 0.0002254025044722719,
+      "loss": 0.1292,
+      "step": 288
+    },
+    {
+      "epoch": 0.5124113475177305,
+      "grad_norm": 0.5988323092460632,
+      "learning_rate": 0.00022513416815742396,
+      "loss": 0.1586,
+      "step": 289
+    },
+    {
+      "epoch": 0.5141843971631206,
+      "grad_norm": 0.5859420299530029,
+      "learning_rate": 0.000224865831842576,
+      "loss": 0.1296,
+      "step": 290
+    },
+    {
+      "epoch": 0.5159574468085106,
+      "grad_norm": 0.4029462933540344,
+      "learning_rate": 0.00022459749552772805,
+      "loss": 0.1072,
+      "step": 291
+    },
+    {
+      "epoch": 0.5177304964539007,
+      "grad_norm": 0.28882545232772827,
+      "learning_rate": 0.00022432915921288013,
+      "loss": 0.0747,
+      "step": 292
+    },
+    {
+      "epoch": 0.5195035460992907,
+      "grad_norm": 0.35556840896606445,
+      "learning_rate": 0.00022406082289803218,
+      "loss": 0.0463,
+      "step": 293
+    },
+    {
+      "epoch": 0.5212765957446809,
+      "grad_norm": 0.5565528273582458,
+      "learning_rate": 0.00022379248658318423,
+      "loss": 0.0675,
+      "step": 294
+    },
+    {
+      "epoch": 0.5230496453900709,
+      "grad_norm": 0.5911768078804016,
+      "learning_rate": 0.0002235241502683363,
+      "loss": 0.1442,
+      "step": 295
+    },
+    {
+      "epoch": 0.524822695035461,
+      "grad_norm": 0.4034654498100281,
+      "learning_rate": 0.00022325581395348835,
+      "loss": 0.1111,
+      "step": 296
+    },
+    {
+      "epoch": 0.526595744680851,
+      "grad_norm": 0.35688602924346924,
+      "learning_rate": 0.00022298747763864043,
+      "loss": 0.0828,
+      "step": 297
+    },
+    {
+      "epoch": 0.5283687943262412,
+      "grad_norm": 0.6449804306030273,
+      "learning_rate": 0.00022271914132379245,
+      "loss": 0.2104,
+      "step": 298
+    },
+    {
+      "epoch": 0.5301418439716312,
+      "grad_norm": 0.5375797152519226,
+      "learning_rate": 0.00022245080500894453,
+      "loss": 0.1546,
+      "step": 299
+    },
+    {
+      "epoch": 0.5319148936170213,
+      "grad_norm": 0.18811915814876556,
+      "learning_rate": 0.00022218246869409658,
+      "loss": 0.0246,
+      "step": 300
+    },
+    {
+      "epoch": 0.5336879432624113,
+      "grad_norm": 1.5281504392623901,
+      "learning_rate": 0.00022191413237924865,
+      "loss": 0.5446,
+      "step": 301
+    },
+    {
+      "epoch": 0.5354609929078015,
+      "grad_norm": 1.5097373723983765,
+      "learning_rate": 0.00022164579606440067,
+      "loss": 0.4033,
+      "step": 302
+    },
+    {
+      "epoch": 0.5372340425531915,
+      "grad_norm": 1.4294488430023193,
+      "learning_rate": 0.00022137745974955275,
+      "loss": 0.3835,
+      "step": 303
+    },
+    {
+      "epoch": 0.5390070921985816,
+      "grad_norm": 0.7668700218200684,
+      "learning_rate": 0.00022110912343470483,
+      "loss": 0.2103,
+      "step": 304
+    },
+    {
+      "epoch": 0.5407801418439716,
+      "grad_norm": 0.9126233458518982,
+      "learning_rate": 0.00022084078711985688,
+      "loss": 0.39,
+      "step": 305
+    },
+    {
+      "epoch": 0.5425531914893617,
+      "grad_norm": 0.7275278568267822,
+      "learning_rate": 0.00022057245080500892,
+      "loss": 0.2128,
+      "step": 306
+    },
+    {
+      "epoch": 0.5443262411347518,
+      "grad_norm": 0.6632657647132874,
+      "learning_rate": 0.00022030411449016097,
+      "loss": 0.1819,
+      "step": 307
+    },
+    {
+      "epoch": 0.5460992907801419,
+      "grad_norm": 0.5265213847160339,
+      "learning_rate": 0.00022003577817531305,
+      "loss": 0.1879,
+      "step": 308
+    },
+    {
+      "epoch": 0.5478723404255319,
+      "grad_norm": 0.4046447277069092,
+      "learning_rate": 0.0002197674418604651,
+      "loss": 0.1225,
+      "step": 309
+    },
+    {
+      "epoch": 0.549645390070922,
+      "grad_norm": 0.7889742851257324,
+      "learning_rate": 0.00021949910554561715,
+      "loss": 0.3022,
+      "step": 310
+    },
+    {
+      "epoch": 0.5514184397163121,
+      "grad_norm": 0.739321768283844,
+      "learning_rate": 0.0002192307692307692,
+      "loss": 0.1621,
+      "step": 311
+    },
+    {
+      "epoch": 0.5531914893617021,
+      "grad_norm": 0.5458974838256836,
+      "learning_rate": 0.00021896243291592127,
+      "loss": 0.1505,
+      "step": 312
+    },
+    {
+      "epoch": 0.5549645390070922,
+      "grad_norm": 0.6029899716377258,
+      "learning_rate": 0.00021869409660107332,
+      "loss": 0.2114,
+      "step": 313
+    },
+    {
+      "epoch": 0.5567375886524822,
+      "grad_norm": 0.830324113368988,
+      "learning_rate": 0.00021842576028622537,
+      "loss": 0.2397,
+      "step": 314
+    },
+    {
+      "epoch": 0.5585106382978723,
+      "grad_norm": 0.7340556383132935,
+      "learning_rate": 0.00021815742397137745,
+      "loss": 0.1606,
+      "step": 315
+    },
+    {
+      "epoch": 0.5602836879432624,
+      "grad_norm": 0.44604039192199707,
+      "learning_rate": 0.0002178890876565295,
+      "loss": 0.1192,
+      "step": 316
+    },
+    {
+      "epoch": 0.5620567375886525,
+      "grad_norm": 0.37659141421318054,
+      "learning_rate": 0.00021762075134168157,
+      "loss": 0.1112,
+      "step": 317
+    },
+    {
+      "epoch": 0.5638297872340425,
+      "grad_norm": 0.8448424935340881,
+      "learning_rate": 0.0002173524150268336,
+      "loss": 0.1799,
+      "step": 318
+    },
+    {
+      "epoch": 0.5656028368794326,
+      "grad_norm": 0.5874999761581421,
+      "learning_rate": 0.00021708407871198567,
+      "loss": 0.2011,
+      "step": 319
+    },
+    {
+      "epoch": 0.5673758865248227,
+      "grad_norm": 0.5802026391029358,
+      "learning_rate": 0.00021681574239713772,
+      "loss": 0.1725,
+      "step": 320
+    },
+    {
+      "epoch": 0.5691489361702128,
+      "grad_norm": 0.5646753907203674,
+      "learning_rate": 0.0002165474060822898,
+      "loss": 0.1464,
+      "step": 321
+    },
+    {
+      "epoch": 0.5709219858156028,
+      "grad_norm": 0.5598042011260986,
+      "learning_rate": 0.00021627906976744182,
+      "loss": 0.1135,
+      "step": 322
+    },
+    {
+      "epoch": 0.5726950354609929,
+      "grad_norm": 0.35759246349334717,
+      "learning_rate": 0.0002160107334525939,
+      "loss": 0.079,
+      "step": 323
+    },
+    {
+      "epoch": 0.574468085106383,
+      "grad_norm": 0.8581865429878235,
+      "learning_rate": 0.00021574239713774597,
+      "loss": 0.2422,
+      "step": 324
+    },
+    {
+      "epoch": 0.5762411347517731,
+      "grad_norm": 0.7060332298278809,
+      "learning_rate": 0.00021547406082289802,
+      "loss": 0.2283,
+      "step": 325
+    },
+    {
+      "epoch": 0.5780141843971631,
+      "grad_norm": 0.7340462803840637,
+      "learning_rate": 0.00021520572450805007,
+      "loss": 0.1915,
+      "step": 326
+    },
+    {
+      "epoch": 0.5797872340425532,
+      "grad_norm": 0.42532655596733093,
+      "learning_rate": 0.00021493738819320212,
+      "loss": 0.1125,
+      "step": 327
+    },
+    {
+      "epoch": 0.5815602836879432,
+      "grad_norm": 0.2977962791919708,
+      "learning_rate": 0.0002146690518783542,
+      "loss": 0.0812,
+      "step": 328
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "grad_norm": 0.3650486171245575,
+      "learning_rate": 0.00021440071556350624,
+      "loss": 0.0999,
+      "step": 329
+    },
+    {
+      "epoch": 0.5851063829787234,
+      "grad_norm": 0.6662601232528687,
+      "learning_rate": 0.0002141323792486583,
+      "loss": 0.2518,
+      "step": 330
+    },
+    {
+      "epoch": 0.5868794326241135,
+      "grad_norm": 0.46315595507621765,
+      "learning_rate": 0.00021386404293381034,
+      "loss": 0.1164,
+      "step": 331
+    },
+    {
+      "epoch": 0.5886524822695035,
+      "grad_norm": 0.5522826910018921,
+      "learning_rate": 0.00021359570661896242,
+      "loss": 0.1563,
+      "step": 332
+    },
+    {
+      "epoch": 0.5904255319148937,
+      "grad_norm": 0.6960483193397522,
+      "learning_rate": 0.0002133273703041145,
+      "loss": 0.1998,
+      "step": 333
+    },
+    {
+      "epoch": 0.5921985815602837,
+      "grad_norm": 0.46060803532600403,
+      "learning_rate": 0.00021305903398926651,
+      "loss": 0.1235,
+      "step": 334
+    },
+    {
+      "epoch": 0.5939716312056738,
+      "grad_norm": 0.8451377749443054,
+      "learning_rate": 0.0002127906976744186,
+      "loss": 0.2462,
+      "step": 335
+    },
+    {
+      "epoch": 0.5957446808510638,
+      "grad_norm": 0.4865632653236389,
+      "learning_rate": 0.00021252236135957064,
+      "loss": 0.1759,
+      "step": 336
+    },
+    {
+      "epoch": 0.5975177304964538,
+      "grad_norm": 0.5368924736976624,
+      "learning_rate": 0.00021225402504472272,
+      "loss": 0.1632,
+      "step": 337
+    },
+    {
+      "epoch": 0.599290780141844,
+      "grad_norm": 0.5031744837760925,
+      "learning_rate": 0.00021198568872987474,
+      "loss": 0.2001,
+      "step": 338
+    },
+    {
+      "epoch": 0.601063829787234,
+      "grad_norm": 0.35617366433143616,
+      "learning_rate": 0.00021171735241502681,
+      "loss": 0.0905,
+      "step": 339
+    },
+    {
+      "epoch": 0.6028368794326241,
+      "grad_norm": 0.4215497076511383,
+      "learning_rate": 0.00021144901610017886,
+      "loss": 0.1474,
+      "step": 340
+    },
+    {
+      "epoch": 0.6046099290780141,
+      "grad_norm": 0.3073301911354065,
+      "learning_rate": 0.00021118067978533094,
+      "loss": 0.053,
+      "step": 341
+    },
+    {
+      "epoch": 0.6063829787234043,
+      "grad_norm": 0.30299848318099976,
+      "learning_rate": 0.00021091234347048301,
+      "loss": 0.0715,
+      "step": 342
+    },
+    {
+      "epoch": 0.6081560283687943,
+      "grad_norm": 0.26558998227119446,
+      "learning_rate": 0.00021064400715563504,
+      "loss": 0.0573,
+      "step": 343
+    },
+    {
+      "epoch": 0.6099290780141844,
+      "grad_norm": 0.21946674585342407,
+      "learning_rate": 0.0002103756708407871,
+      "loss": 0.0502,
+      "step": 344
+    },
+    {
+      "epoch": 0.6117021276595744,
+      "grad_norm": 0.48567119240760803,
+      "learning_rate": 0.00021010733452593916,
+      "loss": 0.1224,
+      "step": 345
+    },
+    {
+      "epoch": 0.6134751773049646,
+      "grad_norm": 0.444938600063324,
+      "learning_rate": 0.00020983899821109124,
+      "loss": 0.0899,
+      "step": 346
+    },
+    {
+      "epoch": 0.6152482269503546,
+      "grad_norm": 0.3576812744140625,
+      "learning_rate": 0.00020957066189624326,
+      "loss": 0.1104,
+      "step": 347
+    },
+    {
+      "epoch": 0.6170212765957447,
+      "grad_norm": 0.5801045298576355,
+      "learning_rate": 0.00020930232558139534,
+      "loss": 0.2192,
+      "step": 348
+    },
+    {
+      "epoch": 0.6187943262411347,
+      "grad_norm": 0.3965771198272705,
+      "learning_rate": 0.00020903398926654739,
+      "loss": 0.1025,
+      "step": 349
+    },
+    {
+      "epoch": 0.6205673758865248,
+      "grad_norm": 0.32997211813926697,
+      "learning_rate": 0.00020876565295169943,
+      "loss": 0.0392,
+      "step": 350
+    },
+    {
+      "epoch": 0.6223404255319149,
+      "grad_norm": 2.3874707221984863,
+      "learning_rate": 0.00020849731663685148,
+      "loss": 0.4267,
+      "step": 351
+    },
+    {
+      "epoch": 0.624113475177305,
+      "grad_norm": 1.0561052560806274,
+      "learning_rate": 0.00020822898032200356,
+      "loss": 0.3531,
+      "step": 352
+    },
+    {
+      "epoch": 0.625886524822695,
+      "grad_norm": 1.0714682340621948,
+      "learning_rate": 0.00020796064400715564,
+      "loss": 0.208,
+      "step": 353
+    },
+    {
+      "epoch": 0.6276595744680851,
+      "grad_norm": 1.0135624408721924,
+      "learning_rate": 0.00020769230769230766,
+      "loss": 0.2722,
+      "step": 354
+    },
+    {
+      "epoch": 0.6294326241134752,
+      "grad_norm": 0.4171317219734192,
+      "learning_rate": 0.00020742397137745973,
+      "loss": 0.1137,
+      "step": 355
+    },
+    {
+      "epoch": 0.6312056737588653,
+      "grad_norm": 0.7069351673126221,
+      "learning_rate": 0.00020715563506261178,
+      "loss": 0.1919,
+      "step": 356
+    },
+    {
+      "epoch": 0.6329787234042553,
+      "grad_norm": 0.5680619478225708,
+      "learning_rate": 0.00020688729874776386,
+      "loss": 0.1884,
+      "step": 357
+    },
+    {
+      "epoch": 0.6347517730496454,
+      "grad_norm": 0.8620048761367798,
+      "learning_rate": 0.00020661896243291588,
+      "loss": 0.3209,
+      "step": 358
+    },
+    {
+      "epoch": 0.6365248226950354,
+      "grad_norm": 0.4163174331188202,
+      "learning_rate": 0.00020635062611806796,
+      "loss": 0.1553,
+      "step": 359
+    },
+    {
+      "epoch": 0.6382978723404256,
+      "grad_norm": 0.7022566199302673,
+      "learning_rate": 0.00020608228980322,
+      "loss": 0.2839,
+      "step": 360
+    },
+    {
+      "epoch": 0.6400709219858156,
+      "grad_norm": 0.35048022866249084,
+      "learning_rate": 0.00020581395348837208,
+      "loss": 0.1168,
+      "step": 361
+    },
+    {
+      "epoch": 0.6418439716312057,
+      "grad_norm": 0.6058384776115417,
+      "learning_rate": 0.00020554561717352416,
+      "loss": 0.2336,
+      "step": 362
+    },
+    {
+      "epoch": 0.6436170212765957,
+      "grad_norm": 0.5632460117340088,
+      "learning_rate": 0.00020527728085867618,
+      "loss": 0.1866,
+      "step": 363
+    },
+    {
+      "epoch": 0.6453900709219859,
+      "grad_norm": 0.8508513569831848,
+      "learning_rate": 0.00020500894454382826,
+      "loss": 0.2795,
+      "step": 364
+    },
+    {
+      "epoch": 0.6471631205673759,
+      "grad_norm": 0.3261764645576477,
+      "learning_rate": 0.0002047406082289803,
+      "loss": 0.0695,
+      "step": 365
+    },
+    {
+      "epoch": 0.648936170212766,
+      "grad_norm": 0.6185038089752197,
+      "learning_rate": 0.00020447227191413238,
+      "loss": 0.2125,
+      "step": 366
+    },
+    {
+      "epoch": 0.650709219858156,
+      "grad_norm": 0.9981937408447266,
+      "learning_rate": 0.0002042039355992844,
+      "loss": 0.179,
+      "step": 367
+    },
+    {
+      "epoch": 0.6524822695035462,
+      "grad_norm": 0.625804603099823,
+      "learning_rate": 0.00020393559928443648,
+      "loss": 0.2315,
+      "step": 368
+    },
+    {
+      "epoch": 0.6542553191489362,
+      "grad_norm": 0.460195928812027,
+      "learning_rate": 0.00020366726296958853,
+      "loss": 0.1352,
+      "step": 369
+    },
+    {
+      "epoch": 0.6560283687943262,
+      "grad_norm": 0.6661739349365234,
+      "learning_rate": 0.00020339892665474058,
+      "loss": 0.2227,
+      "step": 370
+    },
+    {
+      "epoch": 0.6578014184397163,
+      "grad_norm": 0.5500937700271606,
+      "learning_rate": 0.00020313059033989265,
+      "loss": 0.1519,
+      "step": 371
+    },
+    {
+      "epoch": 0.6595744680851063,
+      "grad_norm": 0.8262995481491089,
+      "learning_rate": 0.0002028622540250447,
+      "loss": 0.2965,
+      "step": 372
+    },
+    {
+      "epoch": 0.6613475177304965,
+      "grad_norm": 0.6892026662826538,
+      "learning_rate": 0.00020259391771019678,
+      "loss": 0.2244,
+      "step": 373
+    },
+    {
+      "epoch": 0.6631205673758865,
+      "grad_norm": 0.9893831014633179,
+      "learning_rate": 0.0002023255813953488,
+      "loss": 0.27,
+      "step": 374
+    },
+    {
+      "epoch": 0.6648936170212766,
+      "grad_norm": 0.26278918981552124,
+      "learning_rate": 0.00020205724508050088,
+      "loss": 0.0958,
+      "step": 375
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.4749915301799774,
+      "learning_rate": 0.00020178890876565293,
+      "loss": 0.182,
+      "step": 376
+    },
+    {
+      "epoch": 0.6684397163120568,
+      "grad_norm": 0.36534926295280457,
+      "learning_rate": 0.000201520572450805,
+      "loss": 0.0955,
+      "step": 377
+    },
+    {
+      "epoch": 0.6702127659574468,
+      "grad_norm": 0.28374701738357544,
+      "learning_rate": 0.00020125223613595702,
+      "loss": 0.083,
+      "step": 378
+    },
+    {
+      "epoch": 0.6719858156028369,
+      "grad_norm": 0.5432383418083191,
+      "learning_rate": 0.0002009838998211091,
+      "loss": 0.1533,
+      "step": 379
+    },
+    {
+      "epoch": 0.6737588652482269,
+      "grad_norm": 0.6451116800308228,
+      "learning_rate": 0.00020071556350626118,
+      "loss": 0.2346,
+      "step": 380
+    },
+    {
+      "epoch": 0.675531914893617,
+      "grad_norm": 0.3411319851875305,
+      "learning_rate": 0.00020044722719141323,
+      "loss": 0.0937,
+      "step": 381
+    },
+    {
+      "epoch": 0.6773049645390071,
+      "grad_norm": 0.40049058198928833,
+      "learning_rate": 0.0002001788908765653,
+      "loss": 0.1219,
+      "step": 382
+    },
+    {
+      "epoch": 0.6790780141843972,
+      "grad_norm": 0.3120768070220947,
+      "learning_rate": 0.00019991055456171732,
+      "loss": 0.1086,
+      "step": 383
+    },
+    {
+      "epoch": 0.6808510638297872,
+      "grad_norm": 0.3227909207344055,
+      "learning_rate": 0.0001996422182468694,
+      "loss": 0.0911,
+      "step": 384
+    },
+    {
+      "epoch": 0.6826241134751773,
+      "grad_norm": 0.22975340485572815,
+      "learning_rate": 0.00019937388193202145,
+      "loss": 0.0818,
+      "step": 385
+    },
+    {
+      "epoch": 0.6843971631205674,
+      "grad_norm": 0.3845633268356323,
+      "learning_rate": 0.00019910554561717352,
+      "loss": 0.1278,
+      "step": 386
+    },
+    {
+      "epoch": 0.6861702127659575,
+      "grad_norm": 0.4206799864768982,
+      "learning_rate": 0.00019883720930232555,
+      "loss": 0.1137,
+      "step": 387
+    },
+    {
+      "epoch": 0.6879432624113475,
+      "grad_norm": 0.4134272634983063,
+      "learning_rate": 0.00019856887298747762,
+      "loss": 0.1299,
+      "step": 388
+    },
+    {
+      "epoch": 0.6897163120567376,
+      "grad_norm": 0.44249987602233887,
+      "learning_rate": 0.0001983005366726297,
+      "loss": 0.1211,
+      "step": 389
+    },
+    {
+      "epoch": 0.6914893617021277,
+      "grad_norm": 0.3611302971839905,
+      "learning_rate": 0.00019803220035778172,
+      "loss": 0.114,
+      "step": 390
+    },
+    {
+      "epoch": 0.6932624113475178,
+      "grad_norm": 0.47269973158836365,
+      "learning_rate": 0.0001977638640429338,
+      "loss": 0.1405,
+      "step": 391
+    },
+    {
+      "epoch": 0.6950354609929078,
+      "grad_norm": 0.47546666860580444,
+      "learning_rate": 0.00019749552772808585,
+      "loss": 0.1447,
+      "step": 392
+    },
+    {
+      "epoch": 0.6968085106382979,
+      "grad_norm": 0.3153839111328125,
+      "learning_rate": 0.00019722719141323792,
+      "loss": 0.0972,
+      "step": 393
+    },
+    {
+      "epoch": 0.6985815602836879,
+      "grad_norm": 0.23356540501117706,
+      "learning_rate": 0.00019695885509838994,
+      "loss": 0.0429,
+      "step": 394
+    },
+    {
+      "epoch": 0.700354609929078,
+      "grad_norm": 0.399263858795166,
+      "learning_rate": 0.00019669051878354202,
+      "loss": 0.1401,
+      "step": 395
+    },
+    {
+      "epoch": 0.7021276595744681,
+      "grad_norm": 0.36282962560653687,
+      "learning_rate": 0.00019642218246869407,
+      "loss": 0.11,
+      "step": 396
+    },
+    {
+      "epoch": 0.7039007092198581,
+      "grad_norm": 0.4403574764728546,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 0.0706,
+      "step": 397
+    },
+    {
+      "epoch": 0.7056737588652482,
+      "grad_norm": 0.5901324152946472,
+      "learning_rate": 0.00019588550983899817,
+      "loss": 0.1191,
+      "step": 398
+    },
+    {
+      "epoch": 0.7074468085106383,
+      "grad_norm": 0.2563900351524353,
+      "learning_rate": 0.00019561717352415024,
+      "loss": 0.0731,
+      "step": 399
+    },
+    {
+      "epoch": 0.7092198581560284,
+      "grad_norm": 0.4444797933101654,
+      "learning_rate": 0.00019534883720930232,
+      "loss": 0.1425,
+      "step": 400
+    },
+    {
+      "epoch": 0.7109929078014184,
+      "grad_norm": 1.0342155694961548,
+      "learning_rate": 0.00019508050089445437,
+      "loss": 0.3125,
+      "step": 401
+    },
+    {
+      "epoch": 0.7127659574468085,
+      "grad_norm": 0.6980201005935669,
+      "learning_rate": 0.00019481216457960644,
+      "loss": 0.2188,
+      "step": 402
+    },
+    {
+      "epoch": 0.7145390070921985,
+      "grad_norm": 0.740844190120697,
+      "learning_rate": 0.00019454382826475847,
+      "loss": 0.2467,
+      "step": 403
+    },
+    {
+      "epoch": 0.7163120567375887,
+      "grad_norm": 0.513054370880127,
+      "learning_rate": 0.00019427549194991054,
+      "loss": 0.1531,
+      "step": 404
+    },
+    {
+      "epoch": 0.7180851063829787,
+      "grad_norm": 0.4994530975818634,
+      "learning_rate": 0.0001940071556350626,
+      "loss": 0.1065,
+      "step": 405
+    },
+    {
+      "epoch": 0.7198581560283688,
+      "grad_norm": 0.9895260334014893,
+      "learning_rate": 0.00019373881932021467,
+      "loss": 0.4544,
+      "step": 406
+    },
+    {
+      "epoch": 0.7216312056737588,
+      "grad_norm": 0.422955721616745,
+      "learning_rate": 0.0001934704830053667,
+      "loss": 0.1079,
+      "step": 407
+    },
+    {
+      "epoch": 0.723404255319149,
+      "grad_norm": 0.5568337440490723,
+      "learning_rate": 0.00019320214669051877,
+      "loss": 0.2162,
+      "step": 408
+    },
+    {
+      "epoch": 0.725177304964539,
+      "grad_norm": 0.4763684868812561,
+      "learning_rate": 0.00019293381037567084,
+      "loss": 0.1643,
+      "step": 409
+    },
+    {
+      "epoch": 0.7269503546099291,
+      "grad_norm": 0.4117484390735626,
+      "learning_rate": 0.0001926654740608229,
+      "loss": 0.1393,
+      "step": 410
+    },
+    {
+      "epoch": 0.7287234042553191,
+      "grad_norm": 0.4180298447608948,
+      "learning_rate": 0.00019239713774597494,
+      "loss": 0.183,
+      "step": 411
+    },
+    {
+      "epoch": 0.7304964539007093,
+      "grad_norm": 0.42761579155921936,
+      "learning_rate": 0.000192128801431127,
+      "loss": 0.1463,
+      "step": 412
+    },
+    {
+      "epoch": 0.7322695035460993,
+      "grad_norm": 0.36335864663124084,
+      "learning_rate": 0.00019186046511627906,
+      "loss": 0.1032,
+      "step": 413
+    },
+    {
+      "epoch": 0.7340425531914894,
+      "grad_norm": 0.4711301326751709,
+      "learning_rate": 0.0001915921288014311,
+      "loss": 0.17,
+      "step": 414
+    },
+    {
+      "epoch": 0.7358156028368794,
+      "grad_norm": 0.38681793212890625,
+      "learning_rate": 0.00019132379248658316,
+      "loss": 0.1371,
+      "step": 415
+    },
+    {
+      "epoch": 0.7375886524822695,
+      "grad_norm": 0.38631147146224976,
+      "learning_rate": 0.0001910554561717352,
+      "loss": 0.1183,
+      "step": 416
+    },
+    {
+      "epoch": 0.7393617021276596,
+      "grad_norm": 0.4759841561317444,
+      "learning_rate": 0.0001907871198568873,
+      "loss": 0.1441,
+      "step": 417
+    },
+    {
+      "epoch": 0.7411347517730497,
+      "grad_norm": 0.42136093974113464,
+      "learning_rate": 0.00019051878354203936,
+      "loss": 0.1093,
+      "step": 418
+    },
+    {
+      "epoch": 0.7429078014184397,
+      "grad_norm": 0.5578286051750183,
+      "learning_rate": 0.00019025044722719139,
+      "loss": 0.1521,
+      "step": 419
+    },
+    {
+      "epoch": 0.7446808510638298,
+      "grad_norm": 0.32407546043395996,
+      "learning_rate": 0.00018998211091234346,
+      "loss": 0.0614,
+      "step": 420
+    },
+    {
+      "epoch": 0.7464539007092199,
+      "grad_norm": 0.621771514415741,
+      "learning_rate": 0.0001897137745974955,
+      "loss": 0.2111,
+      "step": 421
+    },
+    {
+      "epoch": 0.74822695035461,
+      "grad_norm": 0.4861868619918823,
+      "learning_rate": 0.0001894454382826476,
+      "loss": 0.1138,
+      "step": 422
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6632353067398071,
+      "learning_rate": 0.0001891771019677996,
+      "loss": 0.2492,
+      "step": 423
+    },
+    {
+      "epoch": 0.75177304964539,
+      "grad_norm": 0.3670256733894348,
+      "learning_rate": 0.00018890876565295169,
+      "loss": 0.0927,
+      "step": 424
+    },
+    {
+      "epoch": 0.7535460992907801,
+      "grad_norm": 0.48544085025787354,
+      "learning_rate": 0.00018864042933810373,
+      "loss": 0.1422,
+      "step": 425
+    },
+    {
+      "epoch": 0.7553191489361702,
+      "grad_norm": 0.3215571939945221,
+      "learning_rate": 0.0001883720930232558,
+      "loss": 0.0729,
+      "step": 426
+    },
+    {
+      "epoch": 0.7570921985815603,
+      "grad_norm": 0.6928660869598389,
+      "learning_rate": 0.00018810375670840786,
+      "loss": 0.2033,
+      "step": 427
+    },
+    {
+      "epoch": 0.7588652482269503,
+      "grad_norm": 0.42845651507377625,
+      "learning_rate": 0.0001878354203935599,
+      "loss": 0.1536,
+      "step": 428
+    },
+    {
+      "epoch": 0.7606382978723404,
+      "grad_norm": 0.5783790349960327,
+      "learning_rate": 0.00018756708407871198,
+      "loss": 0.1573,
+      "step": 429
+    },
+    {
+      "epoch": 0.7624113475177305,
+      "grad_norm": 0.34763625264167786,
+      "learning_rate": 0.00018729874776386403,
+      "loss": 0.0891,
+      "step": 430
+    },
+    {
+      "epoch": 0.7641843971631206,
+      "grad_norm": 0.3298535943031311,
+      "learning_rate": 0.00018703041144901608,
+      "loss": 0.0836,
+      "step": 431
+    },
+    {
+      "epoch": 0.7659574468085106,
+      "grad_norm": 0.48353129625320435,
+      "learning_rate": 0.00018676207513416813,
+      "loss": 0.1624,
+      "step": 432
+    },
+    {
+      "epoch": 0.7677304964539007,
+      "grad_norm": 0.3654095530509949,
+      "learning_rate": 0.0001864937388193202,
+      "loss": 0.074,
+      "step": 433
+    },
+    {
+      "epoch": 0.7695035460992907,
+      "grad_norm": 0.6465492248535156,
+      "learning_rate": 0.00018622540250447223,
+      "loss": 0.2221,
+      "step": 434
+    },
+    {
+      "epoch": 0.7712765957446809,
+      "grad_norm": 0.29519444704055786,
+      "learning_rate": 0.0001859570661896243,
+      "loss": 0.073,
+      "step": 435
+    },
+    {
+      "epoch": 0.7730496453900709,
+      "grad_norm": 0.39645931124687195,
+      "learning_rate": 0.00018568872987477636,
+      "loss": 0.0799,
+      "step": 436
+    },
+    {
+      "epoch": 0.774822695035461,
+      "grad_norm": 0.7929151058197021,
+      "learning_rate": 0.00018542039355992843,
+      "loss": 0.244,
+      "step": 437
+    },
+    {
+      "epoch": 0.776595744680851,
+      "grad_norm": 0.5418475866317749,
+      "learning_rate": 0.0001851520572450805,
+      "loss": 0.1439,
+      "step": 438
+    },
+    {
+      "epoch": 0.7783687943262412,
+      "grad_norm": 0.4036068320274353,
+      "learning_rate": 0.00018488372093023253,
+      "loss": 0.1158,
+      "step": 439
+    },
+    {
+      "epoch": 0.7801418439716312,
+      "grad_norm": 0.33123520016670227,
+      "learning_rate": 0.0001846153846153846,
+      "loss": 0.1182,
+      "step": 440
+    },
+    {
+      "epoch": 0.7819148936170213,
+      "grad_norm": 0.3179425299167633,
+      "learning_rate": 0.00018434704830053665,
+      "loss": 0.0763,
+      "step": 441
+    },
+    {
+      "epoch": 0.7836879432624113,
+      "grad_norm": 0.25326409935951233,
+      "learning_rate": 0.00018407871198568873,
+      "loss": 0.0388,
+      "step": 442
+    },
+    {
+      "epoch": 0.7854609929078015,
+      "grad_norm": 0.4816826581954956,
+      "learning_rate": 0.00018381037567084075,
+      "loss": 0.1308,
+      "step": 443
+    },
+    {
+      "epoch": 0.7872340425531915,
+      "grad_norm": 0.2538895010948181,
+      "learning_rate": 0.00018354203935599283,
+      "loss": 0.0607,
+      "step": 444
+    },
+    {
+      "epoch": 0.7890070921985816,
+      "grad_norm": 0.3773513436317444,
+      "learning_rate": 0.00018327370304114488,
+      "loss": 0.1168,
+      "step": 445
+    },
+    {
+      "epoch": 0.7907801418439716,
+      "grad_norm": 0.48236358165740967,
+      "learning_rate": 0.00018300536672629695,
+      "loss": 0.1078,
+      "step": 446
+    },
+    {
+      "epoch": 0.7925531914893617,
+      "grad_norm": 0.37481409311294556,
+      "learning_rate": 0.000182737030411449,
+      "loss": 0.0847,
+      "step": 447
+    },
+    {
+      "epoch": 0.7943262411347518,
+      "grad_norm": 0.25582361221313477,
+      "learning_rate": 0.00018246869409660105,
+      "loss": 0.0518,
+      "step": 448
+    },
+    {
+      "epoch": 0.7960992907801419,
+      "grad_norm": 0.582918643951416,
+      "learning_rate": 0.00018220035778175313,
+      "loss": 0.199,
+      "step": 449
+    },
+    {
+      "epoch": 0.7978723404255319,
+      "grad_norm": 0.3821042478084564,
+      "learning_rate": 0.00018193202146690518,
+      "loss": 0.1308,
+      "step": 450
+    },
+    {
+      "epoch": 0.799645390070922,
+      "grad_norm": 0.9046875834465027,
+      "learning_rate": 0.00018166368515205723,
+      "loss": 0.3907,
+      "step": 451
+    },
+    {
+      "epoch": 0.8014184397163121,
+      "grad_norm": 0.9091282486915588,
+      "learning_rate": 0.00018139534883720928,
+      "loss": 0.421,
+      "step": 452
+    },
+    {
+      "epoch": 0.8014184397163121,
+      "eval_loss": 0.07596753537654877,
+      "eval_runtime": 47.273,
+      "eval_samples_per_second": 2.665,
+      "eval_steps_per_second": 2.665,
+      "step": 452
+    },
+    {
+      "epoch": 0.8031914893617021,
+      "grad_norm": 0.5998420715332031,
+      "learning_rate": 0.00018112701252236135,
+      "loss": 0.1944,
+      "step": 453
+    },
+    {
+      "epoch": 0.8049645390070922,
+      "grad_norm": 0.6365135908126831,
+      "learning_rate": 0.00018085867620751337,
+      "loss": 0.2445,
+      "step": 454
+    },
+    {
+      "epoch": 0.8067375886524822,
+      "grad_norm": 0.5594285726547241,
+      "learning_rate": 0.00018059033989266545,
+      "loss": 0.2096,
+      "step": 455
+    },
+    {
+      "epoch": 0.8085106382978723,
+      "grad_norm": 0.8409189581871033,
+      "learning_rate": 0.00018032200357781753,
+      "loss": 0.4065,
+      "step": 456
+    },
+    {
+      "epoch": 0.8102836879432624,
+      "grad_norm": 0.5661482214927673,
+      "learning_rate": 0.00018005366726296957,
+      "loss": 0.25,
+      "step": 457
+    },
+    {
+      "epoch": 0.8120567375886525,
+      "grad_norm": 0.5428692102432251,
+      "learning_rate": 0.00017978533094812165,
+      "loss": 0.2254,
+      "step": 458
+    },
+    {
+      "epoch": 0.8138297872340425,
+      "grad_norm": 0.4579595923423767,
+      "learning_rate": 0.00017951699463327367,
+      "loss": 0.1008,
+      "step": 459
+    },
+    {
+      "epoch": 0.8156028368794326,
+      "grad_norm": 0.4390023946762085,
+      "learning_rate": 0.00017924865831842575,
+      "loss": 0.1144,
+      "step": 460
+    },
+    {
+      "epoch": 0.8173758865248227,
+      "grad_norm": 0.5851403474807739,
+      "learning_rate": 0.0001789803220035778,
+      "loss": 0.2355,
+      "step": 461
+    },
+    {
+      "epoch": 0.8191489361702128,
+      "grad_norm": 0.31094157695770264,
+      "learning_rate": 0.00017871198568872987,
+      "loss": 0.1078,
+      "step": 462
+    },
+    {
+      "epoch": 0.8209219858156028,
+      "grad_norm": 0.5213750600814819,
+      "learning_rate": 0.0001784436493738819,
+      "loss": 0.1546,
+      "step": 463
+    },
+    {
+      "epoch": 0.8226950354609929,
+      "grad_norm": 0.3972003161907196,
+      "learning_rate": 0.00017817531305903397,
+      "loss": 0.1379,
+      "step": 464
+    },
+    {
+      "epoch": 0.824468085106383,
+      "grad_norm": 0.35936617851257324,
+      "learning_rate": 0.00017790697674418605,
+      "loss": 0.1238,
+      "step": 465
+    },
+    {
+      "epoch": 0.8262411347517731,
+      "grad_norm": 0.31084805727005005,
+      "learning_rate": 0.0001776386404293381,
+      "loss": 0.0898,
+      "step": 466
+    },
+    {
+      "epoch": 0.8280141843971631,
+      "grad_norm": 0.35060107707977295,
+      "learning_rate": 0.00017737030411449015,
+      "loss": 0.1444,
+      "step": 467
+    },
+    {
+      "epoch": 0.8297872340425532,
+      "grad_norm": 0.31308284401893616,
+      "learning_rate": 0.0001771019677996422,
+      "loss": 0.0722,
+      "step": 468
+    },
+    {
+      "epoch": 0.8315602836879432,
+      "grad_norm": 0.4981883466243744,
+      "learning_rate": 0.00017683363148479427,
+      "loss": 0.1764,
+      "step": 469
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.5609257817268372,
+      "learning_rate": 0.00017656529516994632,
+      "loss": 0.175,
+      "step": 470
+    },
+    {
+      "epoch": 0.8351063829787234,
+      "grad_norm": 0.5611094832420349,
+      "learning_rate": 0.00017629695885509837,
+      "loss": 0.242,
+      "step": 471
+    },
+    {
+      "epoch": 0.8368794326241135,
+      "grad_norm": 0.7660516500473022,
+      "learning_rate": 0.00017602862254025042,
+      "loss": 0.2488,
+      "step": 472
+    },
+    {
+      "epoch": 0.8386524822695035,
+      "grad_norm": 0.3760780096054077,
+      "learning_rate": 0.0001757602862254025,
+      "loss": 0.1253,
+      "step": 473
+    },
+    {
+      "epoch": 0.8404255319148937,
+      "grad_norm": 0.5254774689674377,
+      "learning_rate": 0.00017549194991055454,
+      "loss": 0.2273,
+      "step": 474
+    },
+    {
+      "epoch": 0.8421985815602837,
+      "grad_norm": 0.34555211663246155,
+      "learning_rate": 0.0001752236135957066,
+      "loss": 0.1294,
+      "step": 475
+    },
+    {
+      "epoch": 0.8439716312056738,
+      "grad_norm": 0.4075317978858948,
+      "learning_rate": 0.00017495527728085867,
+      "loss": 0.1258,
+      "step": 476
+    },
+    {
+      "epoch": 0.8457446808510638,
+      "grad_norm": 0.4469994604587555,
+      "learning_rate": 0.00017468694096601072,
+      "loss": 0.155,
+      "step": 477
+    },
+    {
+      "epoch": 0.8475177304964538,
+      "grad_norm": 0.3752269744873047,
+      "learning_rate": 0.0001744186046511628,
+      "loss": 0.1403,
+      "step": 478
+    },
+    {
+      "epoch": 0.849290780141844,
+      "grad_norm": 0.3446385860443115,
+      "learning_rate": 0.00017415026833631482,
+      "loss": 0.0925,
+      "step": 479
+    },
+    {
+      "epoch": 0.851063829787234,
+      "grad_norm": 0.5445675253868103,
+      "learning_rate": 0.0001738819320214669,
+      "loss": 0.1791,
+      "step": 480
+    },
+    {
+      "epoch": 0.8528368794326241,
+      "grad_norm": 0.4845312535762787,
+      "learning_rate": 0.00017361359570661894,
+      "loss": 0.236,
+      "step": 481
+    },
+    {
+      "epoch": 0.8546099290780141,
+      "grad_norm": 0.21976758539676666,
+      "learning_rate": 0.00017334525939177102,
+      "loss": 0.0785,
+      "step": 482
+    },
+    {
+      "epoch": 0.8563829787234043,
+      "grad_norm": 0.3595951497554779,
+      "learning_rate": 0.00017307692307692304,
+      "loss": 0.1297,
+      "step": 483
+    },
+    {
+      "epoch": 0.8581560283687943,
+      "grad_norm": 0.34340062737464905,
+      "learning_rate": 0.00017280858676207511,
+      "loss": 0.1437,
+      "step": 484
+    },
+    {
+      "epoch": 0.8599290780141844,
+      "grad_norm": 0.40024513006210327,
+      "learning_rate": 0.0001725402504472272,
+      "loss": 0.111,
+      "step": 485
+    },
+    {
+      "epoch": 0.8617021276595744,
+      "grad_norm": 0.4997851252555847,
+      "learning_rate": 0.00017227191413237924,
+      "loss": 0.1406,
+      "step": 486
+    },
+    {
+      "epoch": 0.8634751773049646,
+      "grad_norm": 0.37670308351516724,
+      "learning_rate": 0.0001720035778175313,
+      "loss": 0.1653,
+      "step": 487
+    },
+    {
+      "epoch": 0.8652482269503546,
+      "grad_norm": 0.4352554380893707,
+      "learning_rate": 0.00017173524150268334,
+      "loss": 0.1275,
+      "step": 488
+    },
+    {
+      "epoch": 0.8670212765957447,
+      "grad_norm": 0.31643253564834595,
+      "learning_rate": 0.00017146690518783541,
+      "loss": 0.0982,
+      "step": 489
+    },
+    {
+      "epoch": 0.8687943262411347,
+      "grad_norm": 0.3775823712348938,
+      "learning_rate": 0.00017119856887298746,
+      "loss": 0.1087,
+      "step": 490
+    },
+    {
+      "epoch": 0.8705673758865248,
+      "grad_norm": 0.714429497718811,
+      "learning_rate": 0.0001709302325581395,
+      "loss": 0.1743,
+      "step": 491
+    },
+    {
+      "epoch": 0.8723404255319149,
+      "grad_norm": 0.24479807913303375,
+      "learning_rate": 0.00017066189624329156,
+      "loss": 0.0616,
+      "step": 492
+    },
+    {
+      "epoch": 0.874113475177305,
+      "grad_norm": 0.4242885708808899,
+      "learning_rate": 0.00017039355992844364,
+      "loss": 0.1266,
+      "step": 493
+    },
+    {
+      "epoch": 0.875886524822695,
+      "grad_norm": 0.4247332513332367,
+      "learning_rate": 0.00017012522361359571,
+      "loss": 0.1287,
+      "step": 494
+    },
+    {
+      "epoch": 0.8776595744680851,
+      "grad_norm": 0.49524983763694763,
+      "learning_rate": 0.00016985688729874774,
+      "loss": 0.1957,
+      "step": 495
+    },
+    {
+      "epoch": 0.8794326241134752,
+      "grad_norm": 0.2352031022310257,
+      "learning_rate": 0.0001695885509838998,
+      "loss": 0.0585,
+      "step": 496
+    },
+    {
+      "epoch": 0.8812056737588653,
+      "grad_norm": 0.23934312164783478,
+      "learning_rate": 0.00016932021466905186,
+      "loss": 0.0477,
+      "step": 497
+    },
+    {
+      "epoch": 0.8829787234042553,
+      "grad_norm": 0.3493366539478302,
+      "learning_rate": 0.00016905187835420394,
+      "loss": 0.1074,
+      "step": 498
+    },
+    {
+      "epoch": 0.8847517730496454,
+      "grad_norm": 0.28466084599494934,
+      "learning_rate": 0.00016878354203935596,
+      "loss": 0.0575,
+      "step": 499
+    },
+    {
+      "epoch": 0.8865248226950354,
+      "grad_norm": 0.33539536595344543,
+      "learning_rate": 0.00016851520572450803,
+      "loss": 0.0637,
+      "step": 500
+    },
+    {
+      "epoch": 0.8882978723404256,
+      "grad_norm": 0.8851572275161743,
+      "learning_rate": 0.00016824686940966008,
+      "loss": 0.3208,
+      "step": 501
+    },
+    {
+      "epoch": 0.8900709219858156,
+      "grad_norm": 0.8678922057151794,
+      "learning_rate": 0.00016797853309481216,
+      "loss": 0.3818,
+      "step": 502
+    },
+    {
+      "epoch": 0.8918439716312057,
+      "grad_norm": 0.6528666019439697,
+      "learning_rate": 0.00016771019677996424,
+      "loss": 0.2207,
+      "step": 503
+    },
+    {
+      "epoch": 0.8936170212765957,
+      "grad_norm": 0.550101637840271,
+      "learning_rate": 0.00016744186046511626,
+      "loss": 0.1612,
+      "step": 504
+    },
+    {
+      "epoch": 0.8953900709219859,
+      "grad_norm": 0.4004516303539276,
+      "learning_rate": 0.00016717352415026833,
+      "loss": 0.1158,
+      "step": 505
+    },
+    {
+      "epoch": 0.8971631205673759,
+      "grad_norm": 0.3911718428134918,
+      "learning_rate": 0.00016690518783542038,
+      "loss": 0.1702,
+      "step": 506
+    },
+    {
+      "epoch": 0.898936170212766,
+      "grad_norm": 0.45639804005622864,
+      "learning_rate": 0.00016663685152057243,
+      "loss": 0.194,
+      "step": 507
+    },
+    {
+      "epoch": 0.900709219858156,
+      "grad_norm": 0.38425612449645996,
+      "learning_rate": 0.00016636851520572448,
+      "loss": 0.1112,
+      "step": 508
+    },
+    {
+      "epoch": 0.9024822695035462,
+      "grad_norm": 0.3748437464237213,
+      "learning_rate": 0.00016610017889087656,
+      "loss": 0.1234,
+      "step": 509
+    },
+    {
+      "epoch": 0.9042553191489362,
+      "grad_norm": 0.29861482977867126,
+      "learning_rate": 0.0001658318425760286,
+      "loss": 0.1295,
+      "step": 510
+    },
+    {
+      "epoch": 0.9060283687943262,
+      "grad_norm": 0.36903858184814453,
+      "learning_rate": 0.00016556350626118066,
+      "loss": 0.119,
+      "step": 511
+    },
+    {
+      "epoch": 0.9078014184397163,
+      "grad_norm": 0.37858816981315613,
+      "learning_rate": 0.0001652951699463327,
+      "loss": 0.1686,
+      "step": 512
+    },
+    {
+      "epoch": 0.9095744680851063,
+      "grad_norm": 0.34037184715270996,
+      "learning_rate": 0.00016502683363148478,
+      "loss": 0.1046,
+      "step": 513
+    },
+    {
+      "epoch": 0.9113475177304965,
+      "grad_norm": 0.8230399489402771,
+      "learning_rate": 0.00016475849731663686,
+      "loss": 0.301,
+      "step": 514
+    },
+    {
+      "epoch": 0.9131205673758865,
+      "grad_norm": 0.4620438814163208,
+      "learning_rate": 0.00016449016100178888,
+      "loss": 0.1323,
+      "step": 515
+    },
+    {
+      "epoch": 0.9148936170212766,
+      "grad_norm": 0.3759715259075165,
+      "learning_rate": 0.00016422182468694095,
+      "loss": 0.1192,
+      "step": 516
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "grad_norm": 0.6008899211883545,
+      "learning_rate": 0.000163953488372093,
+      "loss": 0.2129,
+      "step": 517
+    },
+    {
+      "epoch": 0.9184397163120568,
+      "grad_norm": 0.4496185779571533,
+      "learning_rate": 0.00016368515205724508,
+      "loss": 0.1597,
+      "step": 518
+    },
+    {
+      "epoch": 0.9202127659574468,
+      "grad_norm": 0.34270530939102173,
+      "learning_rate": 0.0001634168157423971,
+      "loss": 0.1263,
+      "step": 519
+    },
+    {
+      "epoch": 0.9219858156028369,
+      "grad_norm": 0.7845521569252014,
+      "learning_rate": 0.00016314847942754918,
+      "loss": 0.2071,
+      "step": 520
+    },
+    {
+      "epoch": 0.9237588652482269,
+      "grad_norm": 0.4366159737110138,
+      "learning_rate": 0.00016288014311270123,
+      "loss": 0.1805,
+      "step": 521
+    },
+    {
+      "epoch": 0.925531914893617,
+      "grad_norm": 0.3796030580997467,
+      "learning_rate": 0.0001626118067978533,
+      "loss": 0.1017,
+      "step": 522
+    },
+    {
+      "epoch": 0.9273049645390071,
+      "grad_norm": 0.4644930064678192,
+      "learning_rate": 0.00016234347048300538,
+      "loss": 0.1297,
+      "step": 523
+    },
+    {
+      "epoch": 0.9290780141843972,
+      "grad_norm": 0.34396037459373474,
+      "learning_rate": 0.0001620751341681574,
+      "loss": 0.0827,
+      "step": 524
+    },
+    {
+      "epoch": 0.9308510638297872,
+      "grad_norm": 0.3743005096912384,
+      "learning_rate": 0.00016180679785330948,
+      "loss": 0.1467,
+      "step": 525
+    },
+    {
+      "epoch": 0.9326241134751773,
+      "grad_norm": 0.35721150040626526,
+      "learning_rate": 0.00016153846153846153,
+      "loss": 0.0911,
+      "step": 526
+    },
+    {
+      "epoch": 0.9343971631205674,
+      "grad_norm": 0.42456573247909546,
+      "learning_rate": 0.00016127012522361358,
+      "loss": 0.1744,
+      "step": 527
+    },
+    {
+      "epoch": 0.9361702127659575,
+      "grad_norm": 0.4045068919658661,
+      "learning_rate": 0.00016100178890876562,
+      "loss": 0.1234,
+      "step": 528
+    },
+    {
+      "epoch": 0.9379432624113475,
+      "grad_norm": 0.3121665418148041,
+      "learning_rate": 0.0001607334525939177,
+      "loss": 0.0744,
+      "step": 529
+    },
+    {
+      "epoch": 0.9397163120567376,
+      "grad_norm": 0.3464220464229584,
+      "learning_rate": 0.00016046511627906975,
+      "loss": 0.1392,
+      "step": 530
+    },
+    {
+      "epoch": 0.9414893617021277,
+      "grad_norm": 0.3165525496006012,
+      "learning_rate": 0.0001601967799642218,
+      "loss": 0.0688,
+      "step": 531
+    },
+    {
+      "epoch": 0.9432624113475178,
+      "grad_norm": 0.3919833302497864,
+      "learning_rate": 0.00015992844364937387,
+      "loss": 0.1155,
+      "step": 532
+    },
+    {
+      "epoch": 0.9450354609929078,
+      "grad_norm": 0.34956642985343933,
+      "learning_rate": 0.00015966010733452592,
+      "loss": 0.0881,
+      "step": 533
+    },
+    {
+      "epoch": 0.9468085106382979,
+      "grad_norm": 0.3646623492240906,
+      "learning_rate": 0.000159391771019678,
+      "loss": 0.1212,
+      "step": 534
+    },
+    {
+      "epoch": 0.9485815602836879,
+      "grad_norm": 0.4821237325668335,
+      "learning_rate": 0.00015912343470483002,
+      "loss": 0.1594,
+      "step": 535
+    },
+    {
+      "epoch": 0.950354609929078,
+      "grad_norm": 0.28735411167144775,
+      "learning_rate": 0.0001588550983899821,
+      "loss": 0.084,
+      "step": 536
+    },
+    {
+      "epoch": 0.9521276595744681,
+      "grad_norm": 0.5679156184196472,
+      "learning_rate": 0.00015858676207513415,
+      "loss": 0.1563,
+      "step": 537
+    },
+    {
+      "epoch": 0.9539007092198581,
+      "grad_norm": 0.5127171874046326,
+      "learning_rate": 0.00015831842576028622,
+      "loss": 0.1513,
+      "step": 538
+    },
+    {
+      "epoch": 0.9556737588652482,
+      "grad_norm": 0.30556991696357727,
+      "learning_rate": 0.00015805008944543825,
+      "loss": 0.1026,
+      "step": 539
+    },
+    {
+      "epoch": 0.9574468085106383,
+      "grad_norm": 0.41368523240089417,
+      "learning_rate": 0.00015778175313059032,
+      "loss": 0.1205,
+      "step": 540
+    },
+    {
+      "epoch": 0.9592198581560284,
+      "grad_norm": 0.3814534842967987,
+      "learning_rate": 0.0001575134168157424,
+      "loss": 0.1103,
+      "step": 541
+    },
+    {
+      "epoch": 0.9609929078014184,
+      "grad_norm": 0.3183111548423767,
+      "learning_rate": 0.00015724508050089445,
+      "loss": 0.0965,
+      "step": 542
+    },
+    {
+      "epoch": 0.9627659574468085,
+      "grad_norm": 0.276720255613327,
+      "learning_rate": 0.00015697674418604652,
+      "loss": 0.0684,
+      "step": 543
+    },
+    {
+      "epoch": 0.9645390070921985,
+      "grad_norm": 0.2961701452732086,
+      "learning_rate": 0.00015670840787119854,
+      "loss": 0.04,
+      "step": 544
+    },
+    {
+      "epoch": 0.9663120567375887,
+      "grad_norm": 0.3687111437320709,
+      "learning_rate": 0.00015644007155635062,
+      "loss": 0.0921,
+      "step": 545
+    },
+    {
+      "epoch": 0.9680851063829787,
+      "grad_norm": 0.2294873148202896,
+      "learning_rate": 0.00015617173524150267,
+      "loss": 0.0694,
+      "step": 546
+    },
+    {
+      "epoch": 0.9698581560283688,
+      "grad_norm": 0.33963993191719055,
+      "learning_rate": 0.00015590339892665475,
+      "loss": 0.1287,
+      "step": 547
+    },
+    {
+      "epoch": 0.9716312056737588,
+      "grad_norm": 0.4909413456916809,
+      "learning_rate": 0.00015563506261180677,
+      "loss": 0.168,
+      "step": 548
+    },
+    {
+      "epoch": 0.973404255319149,
+      "grad_norm": 0.27783486247062683,
+      "learning_rate": 0.00015536672629695884,
+      "loss": 0.054,
+      "step": 549
+    },
+    {
+      "epoch": 0.975177304964539,
+      "grad_norm": 0.23697739839553833,
+      "learning_rate": 0.0001550983899821109,
+      "loss": 0.0399,
+      "step": 550
+    },
+    {
+      "epoch": 0.9769503546099291,
+      "grad_norm": 0.7063981890678406,
+      "learning_rate": 0.00015483005366726294,
+      "loss": 0.1962,
+      "step": 551
+    },
+    {
+      "epoch": 0.9787234042553191,
+      "grad_norm": 0.6349850296974182,
+      "learning_rate": 0.00015456171735241502,
+      "loss": 0.1456,
+      "step": 552
+    },
+    {
+      "epoch": 0.9804964539007093,
+      "grad_norm": 0.8495027422904968,
+      "learning_rate": 0.00015429338103756707,
+      "loss": 0.2952,
+      "step": 553
+    },
+    {
+      "epoch": 0.9822695035460993,
+      "grad_norm": 0.4368564784526825,
+      "learning_rate": 0.00015402504472271914,
+      "loss": 0.1814,
+      "step": 554
+    },
+    {
+      "epoch": 0.9840425531914894,
+      "grad_norm": 0.5131334662437439,
+      "learning_rate": 0.00015375670840787116,
+      "loss": 0.1503,
+      "step": 555
+    },
+    {
+      "epoch": 0.9858156028368794,
+      "grad_norm": 0.40509024262428284,
+      "learning_rate": 0.00015348837209302324,
+      "loss": 0.1643,
+      "step": 556
+    },
+    {
+      "epoch": 0.9875886524822695,
+      "grad_norm": 0.37984341382980347,
+      "learning_rate": 0.0001532200357781753,
+      "loss": 0.1496,
+      "step": 557
+    },
+    {
+      "epoch": 0.9893617021276596,
+      "grad_norm": 0.39520135521888733,
+      "learning_rate": 0.00015295169946332737,
+      "loss": 0.1251,
+      "step": 558
+    },
+    {
+      "epoch": 0.9911347517730497,
+      "grad_norm": 0.2659216821193695,
+      "learning_rate": 0.0001526833631484794,
+      "loss": 0.0907,
+      "step": 559
+    },
+    {
+      "epoch": 0.9929078014184397,
+      "grad_norm": 0.42887184023857117,
+      "learning_rate": 0.00015241502683363146,
+      "loss": 0.1358,
+      "step": 560
+    },
+    {
+      "epoch": 0.9946808510638298,
+      "grad_norm": 0.2191324532032013,
+      "learning_rate": 0.00015214669051878354,
+      "loss": 0.0713,
+      "step": 561
+    },
+    {
+      "epoch": 0.9964539007092199,
+      "grad_norm": 0.26999935507774353,
+      "learning_rate": 0.0001518783542039356,
+      "loss": 0.0859,
+      "step": 562
+    },
+    {
+      "epoch": 0.99822695035461,
+      "grad_norm": 0.30264967679977417,
+      "learning_rate": 0.00015161001788908767,
+      "loss": 0.1131,
+      "step": 563
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2457718849182129,
+      "learning_rate": 0.0001513416815742397,
+      "loss": 0.0857,
+      "step": 564
+    },
+    {
+      "epoch": 1.00177304964539,
+      "grad_norm": 0.6467576622962952,
+      "learning_rate": 0.00015107334525939176,
+      "loss": 0.2397,
+      "step": 565
+    },
+    {
+      "epoch": 1.00354609929078,
+      "grad_norm": 0.5576766133308411,
+      "learning_rate": 0.0001508050089445438,
+      "loss": 0.2687,
+      "step": 566
+    },
+    {
+      "epoch": 1.0053191489361701,
+      "grad_norm": 0.5558743476867676,
+      "learning_rate": 0.0001505366726296959,
+      "loss": 0.1957,
+      "step": 567
+    },
+    {
+      "epoch": 1.0070921985815602,
+      "grad_norm": 0.3179946541786194,
+      "learning_rate": 0.0001502683363148479,
+      "loss": 0.1073,
+      "step": 568
+    },
+    {
+      "epoch": 1.0088652482269505,
+      "grad_norm": 0.36975720524787903,
+      "learning_rate": 0.00015,
+      "loss": 0.142,
+      "step": 569
+    },
+    {
+      "epoch": 1.0106382978723405,
+      "grad_norm": 0.2883351743221283,
+      "learning_rate": 0.00014973166368515204,
+      "loss": 0.0991,
+      "step": 570
+    },
+    {
+      "epoch": 1.0124113475177305,
+      "grad_norm": 0.36468443274497986,
+      "learning_rate": 0.00014946332737030408,
+      "loss": 0.1154,
+      "step": 571
+    },
+    {
+      "epoch": 1.0141843971631206,
+      "grad_norm": 0.2813965082168579,
+      "learning_rate": 0.00014919499105545616,
+      "loss": 0.1039,
+      "step": 572
+    },
+    {
+      "epoch": 1.0159574468085106,
+      "grad_norm": 0.3948434293270111,
+      "learning_rate": 0.0001489266547406082,
+      "loss": 0.1177,
+      "step": 573
+    },
+    {
+      "epoch": 1.0177304964539007,
+      "grad_norm": 0.4562970697879791,
+      "learning_rate": 0.00014865831842576029,
+      "loss": 0.1622,
+      "step": 574
+    },
+    {
+      "epoch": 1.0195035460992907,
+      "grad_norm": 0.3354407250881195,
+      "learning_rate": 0.00014838998211091234,
+      "loss": 0.1234,
+      "step": 575
+    },
+    {
+      "epoch": 1.0212765957446808,
+      "grad_norm": 0.4788317382335663,
+      "learning_rate": 0.00014812164579606438,
+      "loss": 0.2213,
+      "step": 576
+    },
+    {
+      "epoch": 1.0230496453900708,
+      "grad_norm": 0.3917693495750427,
+      "learning_rate": 0.00014785330948121646,
+      "loss": 0.1178,
+      "step": 577
+    },
+    {
+      "epoch": 1.024822695035461,
+      "grad_norm": 0.27874428033828735,
+      "learning_rate": 0.0001475849731663685,
+      "loss": 0.1009,
+      "step": 578
+    },
+    {
+      "epoch": 1.0265957446808511,
+      "grad_norm": 0.5051771998405457,
+      "learning_rate": 0.00014731663685152056,
+      "loss": 0.1656,
+      "step": 579
+    },
+    {
+      "epoch": 1.0283687943262412,
+      "grad_norm": 0.38514822721481323,
+      "learning_rate": 0.0001470483005366726,
+      "loss": 0.134,
+      "step": 580
+    },
+    {
+      "epoch": 1.0301418439716312,
+      "grad_norm": 0.44142523407936096,
+      "learning_rate": 0.00014677996422182466,
+      "loss": 0.1344,
+      "step": 581
+    },
+    {
+      "epoch": 1.0319148936170213,
+      "grad_norm": 0.27137696743011475,
+      "learning_rate": 0.00014651162790697673,
+      "loss": 0.0951,
+      "step": 582
+    },
+    {
+      "epoch": 1.0336879432624113,
+      "grad_norm": 0.2701858580112457,
+      "learning_rate": 0.00014624329159212878,
+      "loss": 0.0917,
+      "step": 583
+    },
+    {
+      "epoch": 1.0354609929078014,
+      "grad_norm": 0.3253329396247864,
+      "learning_rate": 0.00014597495527728086,
+      "loss": 0.1056,
+      "step": 584
+    },
+    {
+      "epoch": 1.0372340425531914,
+      "grad_norm": 0.31853559613227844,
+      "learning_rate": 0.0001457066189624329,
+      "loss": 0.1003,
+      "step": 585
+    },
+    {
+      "epoch": 1.0390070921985815,
+      "grad_norm": 0.3343375623226166,
+      "learning_rate": 0.00014543828264758496,
+      "loss": 0.109,
+      "step": 586
+    },
+    {
+      "epoch": 1.0407801418439717,
+      "grad_norm": 0.22761645913124084,
+      "learning_rate": 0.00014516994633273703,
+      "loss": 0.08,
+      "step": 587
+    },
+    {
+      "epoch": 1.0425531914893618,
+      "grad_norm": 0.24303674697875977,
+      "learning_rate": 0.00014490161001788908,
+      "loss": 0.089,
+      "step": 588
+    },
+    {
+      "epoch": 1.0443262411347518,
+      "grad_norm": 0.36809274554252625,
+      "learning_rate": 0.00014463327370304113,
+      "loss": 0.1198,
+      "step": 589
+    },
+    {
+      "epoch": 1.0460992907801419,
+      "grad_norm": 0.35131344199180603,
+      "learning_rate": 0.00014436493738819318,
+      "loss": 0.1082,
+      "step": 590
+    },
+    {
+      "epoch": 1.047872340425532,
+      "grad_norm": 0.23665814101696014,
+      "learning_rate": 0.00014409660107334523,
+      "loss": 0.0704,
+      "step": 591
+    },
+    {
+      "epoch": 1.049645390070922,
+      "grad_norm": 0.34136760234832764,
+      "learning_rate": 0.0001438282647584973,
+      "loss": 0.1344,
+      "step": 592
+    },
+    {
+      "epoch": 1.051418439716312,
+      "grad_norm": 0.3390890657901764,
+      "learning_rate": 0.00014355992844364938,
+      "loss": 0.1147,
+      "step": 593
+    },
+    {
+      "epoch": 1.053191489361702,
+      "grad_norm": 0.33551934361457825,
+      "learning_rate": 0.00014329159212880143,
+      "loss": 0.1205,
+      "step": 594
+    },
+    {
+      "epoch": 1.0549645390070923,
+      "grad_norm": 0.3362303674221039,
+      "learning_rate": 0.00014302325581395348,
+      "loss": 0.1012,
+      "step": 595
+    },
+    {
+      "epoch": 1.0567375886524824,
+      "grad_norm": 0.40094855427742004,
+      "learning_rate": 0.00014275491949910553,
+      "loss": 0.1217,
+      "step": 596
+    },
+    {
+      "epoch": 1.0585106382978724,
+      "grad_norm": 0.3621748685836792,
+      "learning_rate": 0.0001424865831842576,
+      "loss": 0.128,
+      "step": 597
+    },
+    {
+      "epoch": 1.0602836879432624,
+      "grad_norm": 0.2884347140789032,
+      "learning_rate": 0.00014221824686940965,
+      "loss": 0.0745,
+      "step": 598
+    },
+    {
+      "epoch": 1.0620567375886525,
+      "grad_norm": 0.31141483783721924,
+      "learning_rate": 0.0001419499105545617,
+      "loss": 0.0932,
+      "step": 599
+    },
+    {
+      "epoch": 1.0638297872340425,
+      "grad_norm": 0.24083460867404938,
+      "learning_rate": 0.00014168157423971375,
+      "loss": 0.0735,
+      "step": 600
+    },
+    {
+      "epoch": 1.0656028368794326,
+      "grad_norm": 0.35196706652641296,
+      "learning_rate": 0.0001414132379248658,
+      "loss": 0.0865,
+      "step": 601
+    },
+    {
+      "epoch": 1.0673758865248226,
+      "grad_norm": 0.2785693407058716,
+      "learning_rate": 0.00014114490161001788,
+      "loss": 0.0832,
+      "step": 602
+    },
+    {
+      "epoch": 1.0691489361702127,
+      "grad_norm": 0.20616450905799866,
+      "learning_rate": 0.00014087656529516995,
+      "loss": 0.0498,
+      "step": 603
+    },
+    {
+      "epoch": 1.070921985815603,
+      "grad_norm": 0.3255903422832489,
+      "learning_rate": 0.000140608228980322,
+      "loss": 0.0971,
+      "step": 604
+    },
+    {
+      "epoch": 1.072695035460993,
+      "grad_norm": 0.24554111063480377,
+      "learning_rate": 0.00014033989266547405,
+      "loss": 0.0497,
+      "step": 605
+    },
+    {
+      "epoch": 1.074468085106383,
+      "grad_norm": 0.4421694576740265,
+      "learning_rate": 0.0001400715563506261,
+      "loss": 0.1649,
+      "step": 606
+    },
+    {
+      "epoch": 1.076241134751773,
+      "grad_norm": 0.24358749389648438,
+      "learning_rate": 0.00013980322003577818,
+      "loss": 0.0573,
+      "step": 607
+    },
+    {
+      "epoch": 1.0780141843971631,
+      "grad_norm": 0.28591474890708923,
+      "learning_rate": 0.00013953488372093022,
+      "loss": 0.0976,
+      "step": 608
+    },
+    {
+      "epoch": 1.0797872340425532,
+      "grad_norm": 0.26130688190460205,
+      "learning_rate": 0.00013926654740608227,
+      "loss": 0.0992,
+      "step": 609
+    },
+    {
+      "epoch": 1.0815602836879432,
+      "grad_norm": 0.2845771908760071,
+      "learning_rate": 0.00013899821109123432,
+      "loss": 0.0581,
+      "step": 610
+    },
+    {
+      "epoch": 1.0833333333333333,
+      "grad_norm": 0.2530725598335266,
+      "learning_rate": 0.0001387298747763864,
+      "loss": 0.0747,
+      "step": 611
+    },
+    {
+      "epoch": 1.0851063829787233,
+      "grad_norm": 0.29305538535118103,
+      "learning_rate": 0.00013846153846153845,
+      "loss": 0.0901,
+      "step": 612
+    },
+    {
+      "epoch": 1.0868794326241136,
+      "grad_norm": 0.23423607647418976,
+      "learning_rate": 0.00013819320214669052,
+      "loss": 0.0743,
+      "step": 613
+    },
+    {
+      "epoch": 1.0886524822695036,
+      "grad_norm": 0.20492520928382874,
+      "learning_rate": 0.00013792486583184257,
+      "loss": 0.0358,
+      "step": 614
+    },
+    {
+      "epoch": 1.0904255319148937,
+      "grad_norm": 0.6422183513641357,
+      "learning_rate": 0.00013765652951699462,
+      "loss": 0.1572,
+      "step": 615
+    },
+    {
+      "epoch": 1.0921985815602837,
+      "grad_norm": 0.8374213576316833,
+      "learning_rate": 0.00013738819320214667,
+      "loss": 0.2574,
+      "step": 616
+    },
+    {
+      "epoch": 1.0939716312056738,
+      "grad_norm": 0.5346600413322449,
+      "learning_rate": 0.00013711985688729875,
+      "loss": 0.2275,
+      "step": 617
+    },
+    {
+      "epoch": 1.0957446808510638,
+      "grad_norm": 0.5961403250694275,
+      "learning_rate": 0.0001368515205724508,
+      "loss": 0.1427,
+      "step": 618
+    },
+    {
+      "epoch": 1.0975177304964538,
+      "grad_norm": 0.4382435381412506,
+      "learning_rate": 0.00013658318425760284,
+      "loss": 0.1515,
+      "step": 619
+    },
+    {
+      "epoch": 1.099290780141844,
+      "grad_norm": 0.49309927225112915,
+      "learning_rate": 0.0001363148479427549,
+      "loss": 0.1894,
+      "step": 620
+    },
+    {
+      "epoch": 1.101063829787234,
+      "grad_norm": 0.3819267153739929,
+      "learning_rate": 0.00013604651162790697,
+      "loss": 0.14,
+      "step": 621
+    },
+    {
+      "epoch": 1.1028368794326242,
+      "grad_norm": 0.3135676383972168,
+      "learning_rate": 0.00013577817531305902,
+      "loss": 0.1468,
+      "step": 622
+    },
+    {
+      "epoch": 1.1046099290780143,
+      "grad_norm": 0.43014463782310486,
+      "learning_rate": 0.0001355098389982111,
+      "loss": 0.152,
+      "step": 623
+    },
+    {
+      "epoch": 1.1063829787234043,
+      "grad_norm": 0.3625108301639557,
+      "learning_rate": 0.00013524150268336314,
+      "loss": 0.1162,
+      "step": 624
+    },
+    {
+      "epoch": 1.1081560283687943,
+      "grad_norm": 0.43387532234191895,
+      "learning_rate": 0.0001349731663685152,
+      "loss": 0.2164,
+      "step": 625
+    },
+    {
+      "epoch": 1.1099290780141844,
+      "grad_norm": 0.49003899097442627,
+      "learning_rate": 0.00013470483005366724,
+      "loss": 0.2018,
+      "step": 626
+    },
+    {
+      "epoch": 1.1117021276595744,
+      "grad_norm": 0.32920360565185547,
+      "learning_rate": 0.00013443649373881932,
+      "loss": 0.1432,
+      "step": 627
+    },
+    {
+      "epoch": 1.1134751773049645,
+      "grad_norm": 0.2235662341117859,
+      "learning_rate": 0.00013416815742397137,
+      "loss": 0.1093,
+      "step": 628
+    },
+    {
+      "epoch": 1.1152482269503545,
+      "grad_norm": 0.32100364565849304,
+      "learning_rate": 0.00013389982110912342,
+      "loss": 0.0847,
+      "step": 629
+    },
+    {
+      "epoch": 1.1170212765957448,
+      "grad_norm": 0.2799128293991089,
+      "learning_rate": 0.00013363148479427547,
+      "loss": 0.12,
+      "step": 630
+    },
+    {
+      "epoch": 1.1187943262411348,
+      "grad_norm": 0.23999084532260895,
+      "learning_rate": 0.00013336314847942754,
+      "loss": 0.073,
+      "step": 631
+    },
+    {
+      "epoch": 1.1205673758865249,
+      "grad_norm": 0.3222167491912842,
+      "learning_rate": 0.0001330948121645796,
+      "loss": 0.1152,
+      "step": 632
+    },
+    {
+      "epoch": 1.122340425531915,
+      "grad_norm": 0.40135595202445984,
+      "learning_rate": 0.00013282647584973167,
+      "loss": 0.1472,
+      "step": 633
+    },
+    {
+      "epoch": 1.124113475177305,
+      "grad_norm": 0.3301258683204651,
+      "learning_rate": 0.00013255813953488372,
+      "loss": 0.1238,
+      "step": 634
+    },
+    {
+      "epoch": 1.125886524822695,
+      "grad_norm": 0.3449542820453644,
+      "learning_rate": 0.00013228980322003576,
+      "loss": 0.1339,
+      "step": 635
+    },
+    {
+      "epoch": 1.127659574468085,
+      "grad_norm": 0.3443644046783447,
+      "learning_rate": 0.00013202146690518781,
+      "loss": 0.1018,
+      "step": 636
+    },
+    {
+      "epoch": 1.1294326241134751,
+      "grad_norm": 0.27248480916023254,
+      "learning_rate": 0.0001317531305903399,
+      "loss": 0.102,
+      "step": 637
+    },
+    {
+      "epoch": 1.1312056737588652,
+      "grad_norm": 0.4105375111103058,
+      "learning_rate": 0.00013148479427549194,
+      "loss": 0.1475,
+      "step": 638
+    },
+    {
+      "epoch": 1.1329787234042552,
+      "grad_norm": 0.468524694442749,
+      "learning_rate": 0.000131216457960644,
+      "loss": 0.11,
+      "step": 639
+    },
+    {
+      "epoch": 1.1347517730496455,
+      "grad_norm": 0.3666204810142517,
+      "learning_rate": 0.00013094812164579604,
+      "loss": 0.1337,
+      "step": 640
+    },
+    {
+      "epoch": 1.1365248226950355,
+      "grad_norm": 0.4041987955570221,
+      "learning_rate": 0.0001306797853309481,
+      "loss": 0.1476,
+      "step": 641
+    },
+    {
+      "epoch": 1.1382978723404256,
+      "grad_norm": 0.2802480459213257,
+      "learning_rate": 0.00013041144901610016,
+      "loss": 0.1008,
+      "step": 642
+    },
+    {
+      "epoch": 1.1400709219858156,
+      "grad_norm": 0.3202970623970032,
+      "learning_rate": 0.00013014311270125224,
+      "loss": 0.1072,
+      "step": 643
+    },
+    {
+      "epoch": 1.1418439716312057,
+      "grad_norm": 0.20990432798862457,
+      "learning_rate": 0.0001298747763864043,
+      "loss": 0.069,
+      "step": 644
+    },
+    {
+      "epoch": 1.1436170212765957,
+      "grad_norm": 0.33487433195114136,
+      "learning_rate": 0.00012960644007155634,
+      "loss": 0.1406,
+      "step": 645
+    },
+    {
+      "epoch": 1.1453900709219857,
+      "grad_norm": 0.2642214298248291,
+      "learning_rate": 0.00012933810375670839,
+      "loss": 0.0974,
+      "step": 646
+    },
+    {
+      "epoch": 1.1471631205673758,
+      "grad_norm": 0.32974326610565186,
+      "learning_rate": 0.00012906976744186046,
+      "loss": 0.0896,
+      "step": 647
+    },
+    {
+      "epoch": 1.148936170212766,
+      "grad_norm": 0.2828482687473297,
+      "learning_rate": 0.0001288014311270125,
+      "loss": 0.0803,
+      "step": 648
+    },
+    {
+      "epoch": 1.150709219858156,
+      "grad_norm": 0.3773040175437927,
+      "learning_rate": 0.00012853309481216456,
+      "loss": 0.1187,
+      "step": 649
+    },
+    {
+      "epoch": 1.1524822695035462,
+      "grad_norm": 0.3649806082248688,
+      "learning_rate": 0.00012826475849731664,
+      "loss": 0.0808,
+      "step": 650
+    },
+    {
+      "epoch": 1.1542553191489362,
+      "grad_norm": 0.3349076509475708,
+      "learning_rate": 0.00012799642218246868,
+      "loss": 0.089,
+      "step": 651
+    },
+    {
+      "epoch": 1.1560283687943262,
+      "grad_norm": 0.2334943264722824,
+      "learning_rate": 0.00012772808586762073,
+      "loss": 0.0706,
+      "step": 652
+    },
+    {
+      "epoch": 1.1578014184397163,
+      "grad_norm": 0.1958976536989212,
+      "learning_rate": 0.0001274597495527728,
+      "loss": 0.05,
+      "step": 653
+    },
+    {
+      "epoch": 1.1595744680851063,
+      "grad_norm": 0.47231465578079224,
+      "learning_rate": 0.00012719141323792486,
+      "loss": 0.165,
+      "step": 654
+    },
+    {
+      "epoch": 1.1613475177304964,
+      "grad_norm": 0.365305632352829,
+      "learning_rate": 0.0001269230769230769,
+      "loss": 0.1267,
+      "step": 655
+    },
+    {
+      "epoch": 1.1631205673758864,
+      "grad_norm": 0.19873836636543274,
+      "learning_rate": 0.00012665474060822896,
+      "loss": 0.0418,
+      "step": 656
+    },
+    {
+      "epoch": 1.1648936170212765,
+      "grad_norm": 0.30495962500572205,
+      "learning_rate": 0.00012638640429338103,
+      "loss": 0.0876,
+      "step": 657
+    },
+    {
+      "epoch": 1.1666666666666667,
+      "grad_norm": 0.28687819838523865,
+      "learning_rate": 0.00012611806797853308,
+      "loss": 0.0658,
+      "step": 658
+    },
+    {
+      "epoch": 1.1684397163120568,
+      "grad_norm": 0.24457816779613495,
+      "learning_rate": 0.00012584973166368513,
+      "loss": 0.081,
+      "step": 659
+    },
+    {
+      "epoch": 1.1702127659574468,
+      "grad_norm": 0.6117145419120789,
+      "learning_rate": 0.0001255813953488372,
+      "loss": 0.124,
+      "step": 660
+    },
+    {
+      "epoch": 1.1719858156028369,
+      "grad_norm": 0.4446345269680023,
+      "learning_rate": 0.00012531305903398926,
+      "loss": 0.0919,
+      "step": 661
+    },
+    {
+      "epoch": 1.173758865248227,
+      "grad_norm": 0.2580985426902771,
+      "learning_rate": 0.0001250447227191413,
+      "loss": 0.0873,
+      "step": 662
+    },
+    {
+      "epoch": 1.175531914893617,
+      "grad_norm": 0.1888839155435562,
+      "learning_rate": 0.00012477638640429338,
+      "loss": 0.0496,
+      "step": 663
+    },
+    {
+      "epoch": 1.177304964539007,
+      "grad_norm": 0.2381056845188141,
+      "learning_rate": 0.00012450805008944543,
+      "loss": 0.0847,
+      "step": 664
+    },
+    {
+      "epoch": 1.1790780141843973,
+      "grad_norm": 0.7352200746536255,
+      "learning_rate": 0.00012423971377459748,
+      "loss": 0.2856,
+      "step": 665
+    },
+    {
+      "epoch": 1.1808510638297873,
+      "grad_norm": 0.40972036123275757,
+      "learning_rate": 0.00012397137745974953,
+      "loss": 0.1386,
+      "step": 666
+    },
+    {
+      "epoch": 1.1826241134751774,
+      "grad_norm": 1.2634938955307007,
+      "learning_rate": 0.0001237030411449016,
+      "loss": 0.2431,
+      "step": 667
+    },
+    {
+      "epoch": 1.1843971631205674,
+      "grad_norm": 0.5589777231216431,
+      "learning_rate": 0.00012343470483005365,
+      "loss": 0.1766,
+      "step": 668
+    },
+    {
+      "epoch": 1.1861702127659575,
+      "grad_norm": 0.4527657926082611,
+      "learning_rate": 0.00012316636851520573,
+      "loss": 0.186,
+      "step": 669
+    },
+    {
+      "epoch": 1.1879432624113475,
+      "grad_norm": 0.507477343082428,
+      "learning_rate": 0.00012289803220035778,
+      "loss": 0.1624,
+      "step": 670
+    },
+    {
+      "epoch": 1.1897163120567376,
+      "grad_norm": 0.5501653552055359,
+      "learning_rate": 0.00012262969588550983,
+      "loss": 0.1805,
+      "step": 671
+    },
+    {
+      "epoch": 1.1914893617021276,
+      "grad_norm": 0.6057561635971069,
+      "learning_rate": 0.00012236135957066188,
+      "loss": 0.2011,
+      "step": 672
+    },
+    {
+      "epoch": 1.1932624113475176,
+      "grad_norm": 0.28531283140182495,
+      "learning_rate": 0.00012209302325581395,
+      "loss": 0.1028,
+      "step": 673
+    },
+    {
+      "epoch": 1.1950354609929077,
+      "grad_norm": 0.4142218828201294,
+      "learning_rate": 0.000121824686940966,
+      "loss": 0.1485,
+      "step": 674
+    },
+    {
+      "epoch": 1.196808510638298,
+      "grad_norm": 0.4527840316295624,
+      "learning_rate": 0.00012155635062611805,
+      "loss": 0.171,
+      "step": 675
+    },
+    {
+      "epoch": 1.198581560283688,
+      "grad_norm": 0.31029072403907776,
+      "learning_rate": 0.00012128801431127011,
+      "loss": 0.116,
+      "step": 676
+    },
+    {
+      "epoch": 1.200354609929078,
+      "grad_norm": 0.5413310527801514,
+      "learning_rate": 0.00012101967799642216,
+      "loss": 0.1969,
+      "step": 677
+    },
+    {
+      "epoch": 1.202127659574468,
+      "grad_norm": 0.24240347743034363,
+      "learning_rate": 0.00012075134168157424,
+      "loss": 0.0809,
+      "step": 678
+    },
+    {
+      "epoch": 1.202127659574468,
+      "eval_loss": 0.0639609843492508,
+      "eval_runtime": 47.1878,
+      "eval_samples_per_second": 2.67,
+      "eval_steps_per_second": 2.67,
+      "step": 678
+    },
+    {
+      "epoch": 1.2039007092198581,
+      "grad_norm": 0.4262109398841858,
+      "learning_rate": 0.00012048300536672629,
+      "loss": 0.1255,
+      "step": 679
+    },
+    {
+      "epoch": 1.2056737588652482,
+      "grad_norm": 0.5289608240127563,
+      "learning_rate": 0.00012021466905187835,
+      "loss": 0.1731,
+      "step": 680
+    },
+    {
+      "epoch": 1.2074468085106382,
+      "grad_norm": 0.32836204767227173,
+      "learning_rate": 0.0001199463327370304,
+      "loss": 0.1366,
+      "step": 681
+    },
+    {
+      "epoch": 1.2092198581560283,
+      "grad_norm": 0.710455060005188,
+      "learning_rate": 0.00011967799642218246,
+      "loss": 0.1448,
+      "step": 682
+    },
+    {
+      "epoch": 1.2109929078014185,
+      "grad_norm": 0.4564105272293091,
+      "learning_rate": 0.00011940966010733451,
+      "loss": 0.1959,
+      "step": 683
+    },
+    {
+      "epoch": 1.2127659574468086,
+      "grad_norm": 0.3768802285194397,
+      "learning_rate": 0.00011914132379248657,
+      "loss": 0.1465,
+      "step": 684
+    },
+    {
+      "epoch": 1.2145390070921986,
+      "grad_norm": 0.3054960072040558,
+      "learning_rate": 0.00011887298747763862,
+      "loss": 0.1059,
+      "step": 685
+    },
+    {
+      "epoch": 1.2163120567375887,
+      "grad_norm": 0.3906306326389313,
+      "learning_rate": 0.00011860465116279069,
+      "loss": 0.1583,
+      "step": 686
+    },
+    {
+      "epoch": 1.2180851063829787,
+      "grad_norm": 0.43236812949180603,
+      "learning_rate": 0.00011833631484794273,
+      "loss": 0.1688,
+      "step": 687
+    },
+    {
+      "epoch": 1.2198581560283688,
+      "grad_norm": 0.39702731370925903,
+      "learning_rate": 0.00011806797853309481,
+      "loss": 0.1422,
+      "step": 688
+    },
+    {
+      "epoch": 1.2216312056737588,
+      "grad_norm": 0.33314409852027893,
+      "learning_rate": 0.00011779964221824686,
+      "loss": 0.0932,
+      "step": 689
+    },
+    {
+      "epoch": 1.2234042553191489,
+      "grad_norm": 0.24420586228370667,
+      "learning_rate": 0.00011753130590339892,
+      "loss": 0.0724,
+      "step": 690
+    },
+    {
+      "epoch": 1.225177304964539,
+      "grad_norm": 0.3487238585948944,
+      "learning_rate": 0.00011726296958855097,
+      "loss": 0.137,
+      "step": 691
+    },
+    {
+      "epoch": 1.226950354609929,
+      "grad_norm": 0.2950010299682617,
+      "learning_rate": 0.00011699463327370303,
+      "loss": 0.12,
+      "step": 692
+    },
+    {
+      "epoch": 1.2287234042553192,
+      "grad_norm": 0.27103373408317566,
+      "learning_rate": 0.00011672629695885508,
+      "loss": 0.0896,
+      "step": 693
+    },
+    {
+      "epoch": 1.2304964539007093,
+      "grad_norm": 0.21480779349803925,
+      "learning_rate": 0.00011645796064400715,
+      "loss": 0.0857,
+      "step": 694
+    },
+    {
+      "epoch": 1.2322695035460993,
+      "grad_norm": 0.27627474069595337,
+      "learning_rate": 0.0001161896243291592,
+      "loss": 0.1184,
+      "step": 695
+    },
+    {
+      "epoch": 1.2340425531914894,
+      "grad_norm": 0.4017396569252014,
+      "learning_rate": 0.00011592128801431126,
+      "loss": 0.1221,
+      "step": 696
+    },
+    {
+      "epoch": 1.2358156028368794,
+      "grad_norm": 0.2959025204181671,
+      "learning_rate": 0.00011565295169946332,
+      "loss": 0.0934,
+      "step": 697
+    },
+    {
+      "epoch": 1.2375886524822695,
+      "grad_norm": 0.48978719115257263,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 0.1601,
+      "step": 698
+    },
+    {
+      "epoch": 1.2393617021276595,
+      "grad_norm": 0.30416980385780334,
+      "learning_rate": 0.00011511627906976743,
+      "loss": 0.1089,
+      "step": 699
+    },
+    {
+      "epoch": 1.2411347517730495,
+      "grad_norm": 0.22977767884731293,
+      "learning_rate": 0.0001148479427549195,
+      "loss": 0.0717,
+      "step": 700
+    },
+    {
+      "epoch": 1.2429078014184398,
+      "grad_norm": 0.45538970828056335,
+      "learning_rate": 0.00011457960644007154,
+      "loss": 0.1824,
+      "step": 701
+    },
+    {
+      "epoch": 1.2446808510638299,
+      "grad_norm": 0.5112214088439941,
+      "learning_rate": 0.0001143112701252236,
+      "loss": 0.1782,
+      "step": 702
+    },
+    {
+      "epoch": 1.24645390070922,
+      "grad_norm": 0.3076941967010498,
+      "learning_rate": 0.00011404293381037565,
+      "loss": 0.1195,
+      "step": 703
+    },
+    {
+      "epoch": 1.24822695035461,
+      "grad_norm": 0.3471357226371765,
+      "learning_rate": 0.00011377459749552772,
+      "loss": 0.133,
+      "step": 704
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.27840855717658997,
+      "learning_rate": 0.00011350626118067977,
+      "loss": 0.0776,
+      "step": 705
+    },
+    {
+      "epoch": 1.25177304964539,
+      "grad_norm": 0.3375721871852875,
+      "learning_rate": 0.00011323792486583183,
+      "loss": 0.1033,
+      "step": 706
+    },
+    {
+      "epoch": 1.25354609929078,
+      "grad_norm": 0.22541607916355133,
+      "learning_rate": 0.00011296958855098389,
+      "loss": 0.0615,
+      "step": 707
+    },
+    {
+      "epoch": 1.2553191489361701,
+      "grad_norm": 0.21740016341209412,
+      "learning_rate": 0.00011270125223613595,
+      "loss": 0.0586,
+      "step": 708
+    },
+    {
+      "epoch": 1.2570921985815602,
+      "grad_norm": 0.23697029054164886,
+      "learning_rate": 0.000112432915921288,
+      "loss": 0.0701,
+      "step": 709
+    },
+    {
+      "epoch": 1.2588652482269502,
+      "grad_norm": 0.21985624730587006,
+      "learning_rate": 0.00011216457960644006,
+      "loss": 0.0617,
+      "step": 710
+    },
+    {
+      "epoch": 1.2606382978723405,
+      "grad_norm": 0.19769056141376495,
+      "learning_rate": 0.00011189624329159211,
+      "loss": 0.0568,
+      "step": 711
+    },
+    {
+      "epoch": 1.2624113475177305,
+      "grad_norm": 0.1668470799922943,
+      "learning_rate": 0.00011162790697674418,
+      "loss": 0.0417,
+      "step": 712
+    },
+    {
+      "epoch": 1.2641843971631206,
+      "grad_norm": 0.1491071879863739,
+      "learning_rate": 0.00011135957066189623,
+      "loss": 0.0338,
+      "step": 713
+    },
+    {
+      "epoch": 1.2659574468085106,
+      "grad_norm": 0.19979891180992126,
+      "learning_rate": 0.00011109123434704829,
+      "loss": 0.0484,
+      "step": 714
+    },
+    {
+      "epoch": 1.2677304964539007,
+      "grad_norm": 0.6793609261512756,
+      "learning_rate": 0.00011082289803220034,
+      "loss": 0.302,
+      "step": 715
+    },
+    {
+      "epoch": 1.2695035460992907,
+      "grad_norm": 0.35137850046157837,
+      "learning_rate": 0.00011055456171735241,
+      "loss": 0.0982,
+      "step": 716
+    },
+    {
+      "epoch": 1.2712765957446808,
+      "grad_norm": 0.35721829533576965,
+      "learning_rate": 0.00011028622540250446,
+      "loss": 0.1059,
+      "step": 717
+    },
+    {
+      "epoch": 1.273049645390071,
+      "grad_norm": 0.44229385256767273,
+      "learning_rate": 0.00011001788908765652,
+      "loss": 0.1698,
+      "step": 718
+    },
+    {
+      "epoch": 1.274822695035461,
+      "grad_norm": 0.5400137305259705,
+      "learning_rate": 0.00010974955277280857,
+      "loss": 0.1205,
+      "step": 719
+    },
+    {
+      "epoch": 1.2765957446808511,
+      "grad_norm": 0.5769519209861755,
+      "learning_rate": 0.00010948121645796064,
+      "loss": 0.1939,
+      "step": 720
+    },
+    {
+      "epoch": 1.2783687943262412,
+      "grad_norm": 0.47844019532203674,
+      "learning_rate": 0.00010921288014311269,
+      "loss": 0.1888,
+      "step": 721
+    },
+    {
+      "epoch": 1.2801418439716312,
+      "grad_norm": 0.297620952129364,
+      "learning_rate": 0.00010894454382826475,
+      "loss": 0.0844,
+      "step": 722
+    },
+    {
+      "epoch": 1.2819148936170213,
+      "grad_norm": 0.550375759601593,
+      "learning_rate": 0.0001086762075134168,
+      "loss": 0.2082,
+      "step": 723
+    },
+    {
+      "epoch": 1.2836879432624113,
+      "grad_norm": 0.5277696847915649,
+      "learning_rate": 0.00010840787119856886,
+      "loss": 0.1826,
+      "step": 724
+    },
+    {
+      "epoch": 1.2854609929078014,
+      "grad_norm": 0.48742157220840454,
+      "learning_rate": 0.00010813953488372091,
+      "loss": 0.1965,
+      "step": 725
+    },
+    {
+      "epoch": 1.2872340425531914,
+      "grad_norm": 0.30468684434890747,
+      "learning_rate": 0.00010787119856887298,
+      "loss": 0.1196,
+      "step": 726
+    },
+    {
+      "epoch": 1.2890070921985815,
+      "grad_norm": 0.33084604144096375,
+      "learning_rate": 0.00010760286225402503,
+      "loss": 0.124,
+      "step": 727
+    },
+    {
+      "epoch": 1.2907801418439715,
+      "grad_norm": 0.34763771295547485,
+      "learning_rate": 0.0001073345259391771,
+      "loss": 0.1066,
+      "step": 728
+    },
+    {
+      "epoch": 1.2925531914893618,
+      "grad_norm": 0.38675862550735474,
+      "learning_rate": 0.00010706618962432915,
+      "loss": 0.1452,
+      "step": 729
+    },
+    {
+      "epoch": 1.2943262411347518,
+      "grad_norm": 0.5646600127220154,
+      "learning_rate": 0.00010679785330948121,
+      "loss": 0.1891,
+      "step": 730
+    },
+    {
+      "epoch": 1.2960992907801419,
+      "grad_norm": 0.35108643770217896,
+      "learning_rate": 0.00010652951699463326,
+      "loss": 0.1223,
+      "step": 731
+    },
+    {
+      "epoch": 1.297872340425532,
+      "grad_norm": 0.3707665801048279,
+      "learning_rate": 0.00010626118067978532,
+      "loss": 0.1441,
+      "step": 732
+    },
+    {
+      "epoch": 1.299645390070922,
+      "grad_norm": 0.27179795503616333,
+      "learning_rate": 0.00010599284436493737,
+      "loss": 0.0839,
+      "step": 733
+    },
+    {
+      "epoch": 1.301418439716312,
+      "grad_norm": 0.37396109104156494,
+      "learning_rate": 0.00010572450805008943,
+      "loss": 0.1495,
+      "step": 734
+    },
+    {
+      "epoch": 1.3031914893617023,
+      "grad_norm": 0.40632113814353943,
+      "learning_rate": 0.00010545617173524151,
+      "loss": 0.1374,
+      "step": 735
+    },
+    {
+      "epoch": 1.3049645390070923,
+      "grad_norm": 0.31261610984802246,
+      "learning_rate": 0.00010518783542039356,
+      "loss": 0.1326,
+      "step": 736
+    },
+    {
+      "epoch": 1.3067375886524824,
+      "grad_norm": 0.4085939824581146,
+      "learning_rate": 0.00010491949910554562,
+      "loss": 0.1394,
+      "step": 737
+    },
+    {
+      "epoch": 1.3085106382978724,
+      "grad_norm": 0.2227245569229126,
+      "learning_rate": 0.00010465116279069767,
+      "loss": 0.0973,
+      "step": 738
+    },
+    {
+      "epoch": 1.3102836879432624,
+      "grad_norm": 0.21776029467582703,
+      "learning_rate": 0.00010438282647584972,
+      "loss": 0.0887,
+      "step": 739
+    },
+    {
+      "epoch": 1.3120567375886525,
+      "grad_norm": 0.4079020619392395,
+      "learning_rate": 0.00010411449016100178,
+      "loss": 0.1516,
+      "step": 740
+    },
+    {
+      "epoch": 1.3138297872340425,
+      "grad_norm": 0.34146982431411743,
+      "learning_rate": 0.00010384615384615383,
+      "loss": 0.1185,
+      "step": 741
+    },
+    {
+      "epoch": 1.3156028368794326,
+      "grad_norm": 0.45869576930999756,
+      "learning_rate": 0.00010357781753130589,
+      "loss": 0.1467,
+      "step": 742
+    },
+    {
+      "epoch": 1.3173758865248226,
+      "grad_norm": 0.23303550481796265,
+      "learning_rate": 0.00010330948121645794,
+      "loss": 0.0855,
+      "step": 743
+    },
+    {
+      "epoch": 1.3191489361702127,
+      "grad_norm": 0.4048026502132416,
+      "learning_rate": 0.00010304114490161,
+      "loss": 0.1394,
+      "step": 744
+    },
+    {
+      "epoch": 1.3209219858156027,
+      "grad_norm": 0.21484145522117615,
+      "learning_rate": 0.00010277280858676208,
+      "loss": 0.0918,
+      "step": 745
+    },
+    {
+      "epoch": 1.322695035460993,
+      "grad_norm": 0.3225015103816986,
+      "learning_rate": 0.00010250447227191413,
+      "loss": 0.1408,
+      "step": 746
+    },
+    {
+      "epoch": 1.324468085106383,
+      "grad_norm": 0.263639360666275,
+      "learning_rate": 0.00010223613595706619,
+      "loss": 0.0796,
+      "step": 747
+    },
+    {
+      "epoch": 1.326241134751773,
+      "grad_norm": 0.5266395211219788,
+      "learning_rate": 0.00010196779964221824,
+      "loss": 0.1809,
+      "step": 748
+    },
+    {
+      "epoch": 1.3280141843971631,
+      "grad_norm": 0.30444666743278503,
+      "learning_rate": 0.00010169946332737029,
+      "loss": 0.1294,
+      "step": 749
+    },
+    {
+      "epoch": 1.3297872340425532,
+      "grad_norm": 0.31879284977912903,
+      "learning_rate": 0.00010143112701252235,
+      "loss": 0.1133,
+      "step": 750
+    },
+    {
+      "epoch": 1.3315602836879432,
+      "grad_norm": 0.5735284686088562,
+      "learning_rate": 0.0001011627906976744,
+      "loss": 0.1922,
+      "step": 751
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.27157899737358093,
+      "learning_rate": 0.00010089445438282646,
+      "loss": 0.1064,
+      "step": 752
+    },
+    {
+      "epoch": 1.3351063829787235,
+      "grad_norm": 0.3107971251010895,
+      "learning_rate": 0.00010062611806797851,
+      "loss": 0.1307,
+      "step": 753
+    },
+    {
+      "epoch": 1.3368794326241136,
+      "grad_norm": 0.3065674304962158,
+      "learning_rate": 0.00010035778175313059,
+      "loss": 0.114,
+      "step": 754
+    },
+    {
+      "epoch": 1.3386524822695036,
+      "grad_norm": 0.3299253582954407,
+      "learning_rate": 0.00010008944543828265,
+      "loss": 0.1296,
+      "step": 755
+    },
+    {
+      "epoch": 1.3404255319148937,
+      "grad_norm": 0.2204119861125946,
+      "learning_rate": 9.98211091234347e-05,
+      "loss": 0.0856,
+      "step": 756
+    },
+    {
+      "epoch": 1.3421985815602837,
+      "grad_norm": 0.27640876173973083,
+      "learning_rate": 9.955277280858676e-05,
+      "loss": 0.1025,
+      "step": 757
+    },
+    {
+      "epoch": 1.3439716312056738,
+      "grad_norm": 0.2008877545595169,
+      "learning_rate": 9.928443649373881e-05,
+      "loss": 0.0698,
+      "step": 758
+    },
+    {
+      "epoch": 1.3457446808510638,
+      "grad_norm": 0.23517268896102905,
+      "learning_rate": 9.901610017889086e-05,
+      "loss": 0.0912,
+      "step": 759
+    },
+    {
+      "epoch": 1.3475177304964538,
+      "grad_norm": 0.24769139289855957,
+      "learning_rate": 9.874776386404292e-05,
+      "loss": 0.092,
+      "step": 760
+    },
+    {
+      "epoch": 1.349290780141844,
+      "grad_norm": 0.25975918769836426,
+      "learning_rate": 9.847942754919497e-05,
+      "loss": 0.0725,
+      "step": 761
+    },
+    {
+      "epoch": 1.351063829787234,
+      "grad_norm": 0.29222846031188965,
+      "learning_rate": 9.821109123434703e-05,
+      "loss": 0.0898,
+      "step": 762
+    },
+    {
+      "epoch": 1.352836879432624,
+      "grad_norm": 0.14782114326953888,
+      "learning_rate": 9.794275491949908e-05,
+      "loss": 0.0427,
+      "step": 763
+    },
+    {
+      "epoch": 1.3546099290780143,
+      "grad_norm": 0.18819881975650787,
+      "learning_rate": 9.767441860465116e-05,
+      "loss": 0.0406,
+      "step": 764
+    },
+    {
+      "epoch": 1.3563829787234043,
+      "grad_norm": 0.6308172345161438,
+      "learning_rate": 9.740608228980322e-05,
+      "loss": 0.2485,
+      "step": 765
+    },
+    {
+      "epoch": 1.3581560283687943,
+      "grad_norm": 0.4238440990447998,
+      "learning_rate": 9.713774597495527e-05,
+      "loss": 0.1535,
+      "step": 766
+    },
+    {
+      "epoch": 1.3599290780141844,
+      "grad_norm": 0.6112564206123352,
+      "learning_rate": 9.686940966010733e-05,
+      "loss": 0.2244,
+      "step": 767
+    },
+    {
+      "epoch": 1.3617021276595744,
+      "grad_norm": 0.3402283489704132,
+      "learning_rate": 9.660107334525938e-05,
+      "loss": 0.1037,
+      "step": 768
+    },
+    {
+      "epoch": 1.3634751773049645,
+      "grad_norm": 0.43715304136276245,
+      "learning_rate": 9.633273703041145e-05,
+      "loss": 0.1632,
+      "step": 769
+    },
+    {
+      "epoch": 1.3652482269503547,
+      "grad_norm": 0.30286452174186707,
+      "learning_rate": 9.60644007155635e-05,
+      "loss": 0.096,
+      "step": 770
+    },
+    {
+      "epoch": 1.3670212765957448,
+      "grad_norm": 0.23162096738815308,
+      "learning_rate": 9.579606440071554e-05,
+      "loss": 0.0945,
+      "step": 771
+    },
+    {
+      "epoch": 1.3687943262411348,
+      "grad_norm": 0.4055822789669037,
+      "learning_rate": 9.55277280858676e-05,
+      "loss": 0.1258,
+      "step": 772
+    },
+    {
+      "epoch": 1.3705673758865249,
+      "grad_norm": 0.37254348397254944,
+      "learning_rate": 9.525939177101968e-05,
+      "loss": 0.1608,
+      "step": 773
+    },
+    {
+      "epoch": 1.372340425531915,
+      "grad_norm": 0.2781223952770233,
+      "learning_rate": 9.499105545617173e-05,
+      "loss": 0.0792,
+      "step": 774
+    },
+    {
+      "epoch": 1.374113475177305,
+      "grad_norm": 0.5268644094467163,
+      "learning_rate": 9.47227191413238e-05,
+      "loss": 0.1891,
+      "step": 775
+    },
+    {
+      "epoch": 1.375886524822695,
+      "grad_norm": 0.27493152022361755,
+      "learning_rate": 9.445438282647584e-05,
+      "loss": 0.1144,
+      "step": 776
+    },
+    {
+      "epoch": 1.377659574468085,
+      "grad_norm": 0.30483171343803406,
+      "learning_rate": 9.41860465116279e-05,
+      "loss": 0.1059,
+      "step": 777
+    },
+    {
+      "epoch": 1.3794326241134751,
+      "grad_norm": 0.35939520597457886,
+      "learning_rate": 9.391771019677995e-05,
+      "loss": 0.1262,
+      "step": 778
+    },
+    {
+      "epoch": 1.3812056737588652,
+      "grad_norm": 0.37438303232192993,
+      "learning_rate": 9.364937388193202e-05,
+      "loss": 0.1527,
+      "step": 779
+    },
+    {
+      "epoch": 1.3829787234042552,
+      "grad_norm": 0.28630080819129944,
+      "learning_rate": 9.338103756708407e-05,
+      "loss": 0.1015,
+      "step": 780
+    },
+    {
+      "epoch": 1.3847517730496453,
+      "grad_norm": 0.2913484275341034,
+      "learning_rate": 9.311270125223611e-05,
+      "loss": 0.0807,
+      "step": 781
+    },
+    {
+      "epoch": 1.3865248226950355,
+      "grad_norm": 0.3168102502822876,
+      "learning_rate": 9.284436493738818e-05,
+      "loss": 0.1148,
+      "step": 782
+    },
+    {
+      "epoch": 1.3882978723404256,
+      "grad_norm": 0.38331836462020874,
+      "learning_rate": 9.257602862254025e-05,
+      "loss": 0.176,
+      "step": 783
+    },
+    {
+      "epoch": 1.3900709219858156,
+      "grad_norm": 0.3016497492790222,
+      "learning_rate": 9.23076923076923e-05,
+      "loss": 0.1292,
+      "step": 784
+    },
+    {
+      "epoch": 1.3918439716312057,
+      "grad_norm": 0.38113200664520264,
+      "learning_rate": 9.203935599284437e-05,
+      "loss": 0.1705,
+      "step": 785
+    },
+    {
+      "epoch": 1.3936170212765957,
+      "grad_norm": 0.3391113877296448,
+      "learning_rate": 9.177101967799641e-05,
+      "loss": 0.1164,
+      "step": 786
+    },
+    {
+      "epoch": 1.3953900709219857,
+      "grad_norm": 0.36019831895828247,
+      "learning_rate": 9.150268336314848e-05,
+      "loss": 0.1608,
+      "step": 787
+    },
+    {
+      "epoch": 1.397163120567376,
+      "grad_norm": 0.5200515389442444,
+      "learning_rate": 9.123434704830053e-05,
+      "loss": 0.1884,
+      "step": 788
+    },
+    {
+      "epoch": 1.398936170212766,
+      "grad_norm": 0.32156822085380554,
+      "learning_rate": 9.096601073345259e-05,
+      "loss": 0.1317,
+      "step": 789
+    },
+    {
+      "epoch": 1.400709219858156,
+      "grad_norm": 0.3899897634983063,
+      "learning_rate": 9.069767441860464e-05,
+      "loss": 0.1483,
+      "step": 790
+    },
+    {
+      "epoch": 1.4024822695035462,
+      "grad_norm": 0.2134401947259903,
+      "learning_rate": 9.042933810375669e-05,
+      "loss": 0.0794,
+      "step": 791
+    },
+    {
+      "epoch": 1.4042553191489362,
+      "grad_norm": 0.5371233224868774,
+      "learning_rate": 9.016100178890876e-05,
+      "loss": 0.1098,
+      "step": 792
+    },
+    {
+      "epoch": 1.4060283687943262,
+      "grad_norm": 0.23417764902114868,
+      "learning_rate": 8.989266547406083e-05,
+      "loss": 0.0846,
+      "step": 793
+    },
+    {
+      "epoch": 1.4078014184397163,
+      "grad_norm": 0.32313209772109985,
+      "learning_rate": 8.962432915921287e-05,
+      "loss": 0.1005,
+      "step": 794
+    },
+    {
+      "epoch": 1.4095744680851063,
+      "grad_norm": 0.3287942111492157,
+      "learning_rate": 8.935599284436494e-05,
+      "loss": 0.138,
+      "step": 795
+    },
+    {
+      "epoch": 1.4113475177304964,
+      "grad_norm": 0.4062165915966034,
+      "learning_rate": 8.908765652951699e-05,
+      "loss": 0.1219,
+      "step": 796
+    },
+    {
+      "epoch": 1.4131205673758864,
+      "grad_norm": 0.25136035680770874,
+      "learning_rate": 8.881932021466905e-05,
+      "loss": 0.0987,
+      "step": 797
+    },
+    {
+      "epoch": 1.4148936170212765,
+      "grad_norm": 0.3869902193546295,
+      "learning_rate": 8.85509838998211e-05,
+      "loss": 0.151,
+      "step": 798
+    },
+    {
+      "epoch": 1.4166666666666667,
+      "grad_norm": 0.22914816439151764,
+      "learning_rate": 8.828264758497316e-05,
+      "loss": 0.0755,
+      "step": 799
+    },
+    {
+      "epoch": 1.4184397163120568,
+      "grad_norm": 0.32536619901657104,
+      "learning_rate": 8.801431127012521e-05,
+      "loss": 0.1082,
+      "step": 800
+    },
+    {
+      "epoch": 1.4202127659574468,
+      "grad_norm": 0.19335246086120605,
+      "learning_rate": 8.774597495527727e-05,
+      "loss": 0.0669,
+      "step": 801
+    },
+    {
+      "epoch": 1.4219858156028369,
+      "grad_norm": 0.2427796870470047,
+      "learning_rate": 8.747763864042933e-05,
+      "loss": 0.0848,
+      "step": 802
+    },
+    {
+      "epoch": 1.423758865248227,
+      "grad_norm": 0.2558480203151703,
+      "learning_rate": 8.72093023255814e-05,
+      "loss": 0.0879,
+      "step": 803
+    },
+    {
+      "epoch": 1.425531914893617,
+      "grad_norm": 0.2617630064487457,
+      "learning_rate": 8.694096601073345e-05,
+      "loss": 0.0849,
+      "step": 804
+    },
+    {
+      "epoch": 1.427304964539007,
+      "grad_norm": 0.2902472913265228,
+      "learning_rate": 8.667262969588551e-05,
+      "loss": 0.091,
+      "step": 805
+    },
+    {
+      "epoch": 1.4290780141843973,
+      "grad_norm": 0.2947092652320862,
+      "learning_rate": 8.640429338103756e-05,
+      "loss": 0.1139,
+      "step": 806
+    },
+    {
+      "epoch": 1.4308510638297873,
+      "grad_norm": 0.23282337188720703,
+      "learning_rate": 8.613595706618962e-05,
+      "loss": 0.0963,
+      "step": 807
+    },
+    {
+      "epoch": 1.4326241134751774,
+      "grad_norm": 0.37111896276474,
+      "learning_rate": 8.586762075134167e-05,
+      "loss": 0.1556,
+      "step": 808
+    },
+    {
+      "epoch": 1.4343971631205674,
+      "grad_norm": 0.2589118778705597,
+      "learning_rate": 8.559928443649373e-05,
+      "loss": 0.0903,
+      "step": 809
+    },
+    {
+      "epoch": 1.4361702127659575,
+      "grad_norm": 0.20402726531028748,
+      "learning_rate": 8.533094812164578e-05,
+      "loss": 0.0566,
+      "step": 810
+    },
+    {
+      "epoch": 1.4379432624113475,
+      "grad_norm": 0.18823835253715515,
+      "learning_rate": 8.506261180679786e-05,
+      "loss": 0.0861,
+      "step": 811
+    },
+    {
+      "epoch": 1.4397163120567376,
+      "grad_norm": 0.27323687076568604,
+      "learning_rate": 8.47942754919499e-05,
+      "loss": 0.1117,
+      "step": 812
+    },
+    {
+      "epoch": 1.4414893617021276,
+      "grad_norm": 0.18384765088558197,
+      "learning_rate": 8.452593917710197e-05,
+      "loss": 0.0429,
+      "step": 813
+    },
+    {
+      "epoch": 1.4432624113475176,
+      "grad_norm": 0.202216237783432,
+      "learning_rate": 8.425760286225402e-05,
+      "loss": 0.0426,
+      "step": 814
+    },
+    {
+      "epoch": 1.4450354609929077,
+      "grad_norm": 0.4300391674041748,
+      "learning_rate": 8.398926654740608e-05,
+      "loss": 0.1519,
+      "step": 815
+    },
+    {
+      "epoch": 1.4468085106382977,
+      "grad_norm": 0.4760010242462158,
+      "learning_rate": 8.372093023255813e-05,
+      "loss": 0.1508,
+      "step": 816
+    },
+    {
+      "epoch": 1.448581560283688,
+      "grad_norm": 0.2631239593029022,
+      "learning_rate": 8.345259391771019e-05,
+      "loss": 0.0754,
+      "step": 817
+    },
+    {
+      "epoch": 1.450354609929078,
+      "grad_norm": 0.37113967537879944,
+      "learning_rate": 8.318425760286224e-05,
+      "loss": 0.128,
+      "step": 818
+    },
+    {
+      "epoch": 1.452127659574468,
+      "grad_norm": 0.21397681534290314,
+      "learning_rate": 8.29159212880143e-05,
+      "loss": 0.0823,
+      "step": 819
+    },
+    {
+      "epoch": 1.4539007092198581,
+      "grad_norm": 0.27529531717300415,
+      "learning_rate": 8.264758497316635e-05,
+      "loss": 0.0967,
+      "step": 820
+    },
+    {
+      "epoch": 1.4556737588652482,
+      "grad_norm": 0.5994647741317749,
+      "learning_rate": 8.237924865831843e-05,
+      "loss": 0.1117,
+      "step": 821
+    },
+    {
+      "epoch": 1.4574468085106382,
+      "grad_norm": 0.22065456211566925,
+      "learning_rate": 8.211091234347048e-05,
+      "loss": 0.0928,
+      "step": 822
+    },
+    {
+      "epoch": 1.4592198581560285,
+      "grad_norm": 0.6050121188163757,
+      "learning_rate": 8.184257602862254e-05,
+      "loss": 0.2078,
+      "step": 823
+    },
+    {
+      "epoch": 1.4609929078014185,
+      "grad_norm": 0.763676106929779,
+      "learning_rate": 8.157423971377459e-05,
+      "loss": 0.2562,
+      "step": 824
+    },
+    {
+      "epoch": 1.4627659574468086,
+      "grad_norm": 0.2788752019405365,
+      "learning_rate": 8.130590339892665e-05,
+      "loss": 0.1154,
+      "step": 825
+    },
+    {
+      "epoch": 1.4645390070921986,
+      "grad_norm": 0.4570877254009247,
+      "learning_rate": 8.10375670840787e-05,
+      "loss": 0.1463,
+      "step": 826
+    },
+    {
+      "epoch": 1.4663120567375887,
+      "grad_norm": 0.40701577067375183,
+      "learning_rate": 8.076923076923076e-05,
+      "loss": 0.1529,
+      "step": 827
+    },
+    {
+      "epoch": 1.4680851063829787,
+      "grad_norm": 0.2114609330892563,
+      "learning_rate": 8.050089445438281e-05,
+      "loss": 0.0828,
+      "step": 828
+    },
+    {
+      "epoch": 1.4698581560283688,
+      "grad_norm": 0.38288062810897827,
+      "learning_rate": 8.023255813953487e-05,
+      "loss": 0.1594,
+      "step": 829
+    },
+    {
+      "epoch": 1.4716312056737588,
+      "grad_norm": 0.26571860909461975,
+      "learning_rate": 7.996422182468694e-05,
+      "loss": 0.1094,
+      "step": 830
+    },
+    {
+      "epoch": 1.4734042553191489,
+      "grad_norm": 0.41206905245780945,
+      "learning_rate": 7.9695885509839e-05,
+      "loss": 0.1454,
+      "step": 831
+    },
+    {
+      "epoch": 1.475177304964539,
+      "grad_norm": 0.26968613266944885,
+      "learning_rate": 7.942754919499105e-05,
+      "loss": 0.1204,
+      "step": 832
+    },
+    {
+      "epoch": 1.476950354609929,
+      "grad_norm": 0.3330513834953308,
+      "learning_rate": 7.915921288014311e-05,
+      "loss": 0.1212,
+      "step": 833
+    },
+    {
+      "epoch": 1.4787234042553192,
+      "grad_norm": 0.3439812660217285,
+      "learning_rate": 7.889087656529516e-05,
+      "loss": 0.151,
+      "step": 834
+    },
+    {
+      "epoch": 1.4804964539007093,
+      "grad_norm": 0.34060341119766235,
+      "learning_rate": 7.862254025044722e-05,
+      "loss": 0.0984,
+      "step": 835
+    },
+    {
+      "epoch": 1.4822695035460993,
+      "grad_norm": 0.391315758228302,
+      "learning_rate": 7.835420393559927e-05,
+      "loss": 0.1381,
+      "step": 836
+    },
+    {
+      "epoch": 1.4840425531914894,
+      "grad_norm": 0.27594539523124695,
+      "learning_rate": 7.808586762075133e-05,
+      "loss": 0.0773,
+      "step": 837
+    },
+    {
+      "epoch": 1.4858156028368794,
+      "grad_norm": 0.2562049925327301,
+      "learning_rate": 7.781753130590338e-05,
+      "loss": 0.085,
+      "step": 838
+    },
+    {
+      "epoch": 1.4875886524822695,
+      "grad_norm": 0.5238338112831116,
+      "learning_rate": 7.754919499105545e-05,
+      "loss": 0.1607,
+      "step": 839
+    },
+    {
+      "epoch": 1.4893617021276595,
+      "grad_norm": 0.35886940360069275,
+      "learning_rate": 7.728085867620751e-05,
+      "loss": 0.1452,
+      "step": 840
+    },
+    {
+      "epoch": 1.4911347517730498,
+      "grad_norm": 0.2899888753890991,
+      "learning_rate": 7.701252236135957e-05,
+      "loss": 0.0896,
+      "step": 841
+    },
+    {
+      "epoch": 1.4929078014184398,
+      "grad_norm": 0.25350284576416016,
+      "learning_rate": 7.674418604651162e-05,
+      "loss": 0.1055,
+      "step": 842
+    },
+    {
+      "epoch": 1.4946808510638299,
+      "grad_norm": 0.2904045283794403,
+      "learning_rate": 7.647584973166368e-05,
+      "loss": 0.1178,
+      "step": 843
+    },
+    {
+      "epoch": 1.49645390070922,
+      "grad_norm": 0.39105409383773804,
+      "learning_rate": 7.620751341681573e-05,
+      "loss": 0.1772,
+      "step": 844
+    },
+    {
+      "epoch": 1.49822695035461,
+      "grad_norm": 0.20958948135375977,
+      "learning_rate": 7.59391771019678e-05,
+      "loss": 0.0848,
+      "step": 845
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.2883208990097046,
+      "learning_rate": 7.567084078711984e-05,
+      "loss": 0.1166,
+      "step": 846
+    },
+    {
+      "epoch": 1.50177304964539,
+      "grad_norm": 0.2059636414051056,
+      "learning_rate": 7.54025044722719e-05,
+      "loss": 0.0782,
+      "step": 847
+    },
+    {
+      "epoch": 1.50354609929078,
+      "grad_norm": 0.30345362424850464,
+      "learning_rate": 7.513416815742396e-05,
+      "loss": 0.1018,
+      "step": 848
+    },
+    {
+      "epoch": 1.5053191489361701,
+      "grad_norm": 0.27595144510269165,
+      "learning_rate": 7.486583184257602e-05,
+      "loss": 0.116,
+      "step": 849
+    },
+    {
+      "epoch": 1.5070921985815602,
+      "grad_norm": 0.2541591227054596,
+      "learning_rate": 7.459749552772808e-05,
+      "loss": 0.1124,
+      "step": 850
+    },
+    {
+      "epoch": 1.5088652482269502,
+      "grad_norm": 0.2614005506038666,
+      "learning_rate": 7.432915921288014e-05,
+      "loss": 0.0797,
+      "step": 851
+    },
+    {
+      "epoch": 1.5106382978723403,
+      "grad_norm": 0.29117223620414734,
+      "learning_rate": 7.406082289803219e-05,
+      "loss": 0.1023,
+      "step": 852
+    },
+    {
+      "epoch": 1.5124113475177305,
+      "grad_norm": 0.18029992282390594,
+      "learning_rate": 7.379248658318425e-05,
+      "loss": 0.0695,
+      "step": 853
+    },
+    {
+      "epoch": 1.5141843971631206,
+      "grad_norm": 0.2022331953048706,
+      "learning_rate": 7.35241502683363e-05,
+      "loss": 0.0632,
+      "step": 854
+    },
+    {
+      "epoch": 1.5159574468085106,
+      "grad_norm": 0.1515171378850937,
+      "learning_rate": 7.325581395348837e-05,
+      "loss": 0.0447,
+      "step": 855
+    },
+    {
+      "epoch": 1.5177304964539007,
+      "grad_norm": 0.14520789682865143,
+      "learning_rate": 7.298747763864043e-05,
+      "loss": 0.0454,
+      "step": 856
+    },
+    {
+      "epoch": 1.5195035460992907,
+      "grad_norm": 0.18276812136173248,
+      "learning_rate": 7.271914132379248e-05,
+      "loss": 0.0651,
+      "step": 857
+    },
+    {
+      "epoch": 1.521276595744681,
+      "grad_norm": 0.2109200656414032,
+      "learning_rate": 7.245080500894454e-05,
+      "loss": 0.0658,
+      "step": 858
+    },
+    {
+      "epoch": 1.523049645390071,
+      "grad_norm": 0.3034365475177765,
+      "learning_rate": 7.218246869409659e-05,
+      "loss": 0.1011,
+      "step": 859
+    },
+    {
+      "epoch": 1.524822695035461,
+      "grad_norm": 0.3006940484046936,
+      "learning_rate": 7.191413237924865e-05,
+      "loss": 0.1229,
+      "step": 860
+    },
+    {
+      "epoch": 1.5265957446808511,
+      "grad_norm": 0.23068998754024506,
+      "learning_rate": 7.164579606440071e-05,
+      "loss": 0.0839,
+      "step": 861
+    },
+    {
+      "epoch": 1.5283687943262412,
+      "grad_norm": 0.15325228869915009,
+      "learning_rate": 7.137745974955276e-05,
+      "loss": 0.0579,
+      "step": 862
+    },
+    {
+      "epoch": 1.5301418439716312,
+      "grad_norm": 0.22034241259098053,
+      "learning_rate": 7.110912343470483e-05,
+      "loss": 0.0681,
+      "step": 863
+    },
+    {
+      "epoch": 1.5319148936170213,
+      "grad_norm": 0.16374541819095612,
+      "learning_rate": 7.084078711985688e-05,
+      "loss": 0.0401,
+      "step": 864
+    },
+    {
+      "epoch": 1.5336879432624113,
+      "grad_norm": 0.7110052108764648,
+      "learning_rate": 7.057245080500894e-05,
+      "loss": 0.2821,
+      "step": 865
+    },
+    {
+      "epoch": 1.5354609929078014,
+      "grad_norm": 0.5873008966445923,
+      "learning_rate": 7.0304114490161e-05,
+      "loss": 0.2055,
+      "step": 866
+    },
+    {
+      "epoch": 1.5372340425531914,
+      "grad_norm": 0.40463680028915405,
+      "learning_rate": 7.003577817531305e-05,
+      "loss": 0.1514,
+      "step": 867
+    },
+    {
+      "epoch": 1.5390070921985815,
+      "grad_norm": 0.411138653755188,
+      "learning_rate": 6.976744186046511e-05,
+      "loss": 0.1374,
+      "step": 868
+    },
+    {
+      "epoch": 1.5407801418439715,
+      "grad_norm": 0.3094438910484314,
+      "learning_rate": 6.949910554561716e-05,
+      "loss": 0.1114,
+      "step": 869
+    },
+    {
+      "epoch": 1.5425531914893615,
+      "grad_norm": 0.3126083016395569,
+      "learning_rate": 6.923076923076922e-05,
+      "loss": 0.104,
+      "step": 870
+    },
+    {
+      "epoch": 1.5443262411347518,
+      "grad_norm": 0.2306557297706604,
+      "learning_rate": 6.896243291592129e-05,
+      "loss": 0.0824,
+      "step": 871
+    },
+    {
+      "epoch": 1.5460992907801419,
+      "grad_norm": 1.1792900562286377,
+      "learning_rate": 6.869409660107334e-05,
+      "loss": 0.2877,
+      "step": 872
+    },
+    {
+      "epoch": 1.547872340425532,
+      "grad_norm": 0.25982725620269775,
+      "learning_rate": 6.84257602862254e-05,
+      "loss": 0.1083,
+      "step": 873
+    },
+    {
+      "epoch": 1.549645390070922,
+      "grad_norm": 0.33999934792518616,
+      "learning_rate": 6.815742397137745e-05,
+      "loss": 0.1439,
+      "step": 874
+    },
+    {
+      "epoch": 1.5514184397163122,
+      "grad_norm": 0.2777253985404968,
+      "learning_rate": 6.788908765652951e-05,
+      "loss": 0.1171,
+      "step": 875
+    },
+    {
+      "epoch": 1.5531914893617023,
+      "grad_norm": 0.29223597049713135,
+      "learning_rate": 6.762075134168157e-05,
+      "loss": 0.0913,
+      "step": 876
+    },
+    {
+      "epoch": 1.5549645390070923,
+      "grad_norm": 0.3456422984600067,
+      "learning_rate": 6.735241502683362e-05,
+      "loss": 0.1324,
+      "step": 877
+    },
+    {
+      "epoch": 1.5567375886524824,
+      "grad_norm": 0.27550065517425537,
+      "learning_rate": 6.708407871198568e-05,
+      "loss": 0.1061,
+      "step": 878
+    },
+    {
+      "epoch": 1.5585106382978724,
+      "grad_norm": 0.23130980134010315,
+      "learning_rate": 6.681574239713773e-05,
+      "loss": 0.0879,
+      "step": 879
+    },
+    {
+      "epoch": 1.5602836879432624,
+      "grad_norm": 0.2471001297235489,
+      "learning_rate": 6.65474060822898e-05,
+      "loss": 0.1028,
+      "step": 880
+    },
+    {
+      "epoch": 1.5620567375886525,
+      "grad_norm": 0.4121532440185547,
+      "learning_rate": 6.627906976744186e-05,
+      "loss": 0.1524,
+      "step": 881
+    },
+    {
+      "epoch": 1.5638297872340425,
+      "grad_norm": 0.34416720271110535,
+      "learning_rate": 6.601073345259391e-05,
+      "loss": 0.153,
+      "step": 882
+    },
+    {
+      "epoch": 1.5656028368794326,
+      "grad_norm": 0.37047117948532104,
+      "learning_rate": 6.574239713774597e-05,
+      "loss": 0.1382,
+      "step": 883
+    },
+    {
+      "epoch": 1.5673758865248226,
+      "grad_norm": 0.25409069657325745,
+      "learning_rate": 6.547406082289802e-05,
+      "loss": 0.0929,
+      "step": 884
+    },
+    {
+      "epoch": 1.5691489361702127,
+      "grad_norm": 0.16979749500751495,
+      "learning_rate": 6.520572450805008e-05,
+      "loss": 0.0661,
+      "step": 885
+    },
+    {
+      "epoch": 1.5709219858156027,
+      "grad_norm": 0.24894487857818604,
+      "learning_rate": 6.493738819320214e-05,
+      "loss": 0.0879,
+      "step": 886
+    },
+    {
+      "epoch": 1.5726950354609928,
+      "grad_norm": 0.23900751769542694,
+      "learning_rate": 6.466905187835419e-05,
+      "loss": 0.083,
+      "step": 887
+    },
+    {
+      "epoch": 1.574468085106383,
+      "grad_norm": 0.21001844108104706,
+      "learning_rate": 6.440071556350626e-05,
+      "loss": 0.0846,
+      "step": 888
+    },
+    {
+      "epoch": 1.576241134751773,
+      "grad_norm": 0.4241371750831604,
+      "learning_rate": 6.413237924865832e-05,
+      "loss": 0.1436,
+      "step": 889
+    },
+    {
+      "epoch": 1.5780141843971631,
+      "grad_norm": 0.36801424622535706,
+      "learning_rate": 6.386404293381037e-05,
+      "loss": 0.144,
+      "step": 890
+    },
+    {
+      "epoch": 1.5797872340425532,
+      "grad_norm": 0.3026575744152069,
+      "learning_rate": 6.359570661896243e-05,
+      "loss": 0.1313,
+      "step": 891
+    },
+    {
+      "epoch": 1.5815602836879432,
+      "grad_norm": 0.6115602254867554,
+      "learning_rate": 6.332737030411448e-05,
+      "loss": 0.2098,
+      "step": 892
+    },
+    {
+      "epoch": 1.5833333333333335,
+      "grad_norm": 0.21278461813926697,
+      "learning_rate": 6.305903398926654e-05,
+      "loss": 0.0877,
+      "step": 893
+    },
+    {
+      "epoch": 1.5851063829787235,
+      "grad_norm": 0.1702726036310196,
+      "learning_rate": 6.27906976744186e-05,
+      "loss": 0.0776,
+      "step": 894
+    },
+    {
+      "epoch": 1.5868794326241136,
+      "grad_norm": 0.2994159758090973,
+      "learning_rate": 6.252236135957065e-05,
+      "loss": 0.1155,
+      "step": 895
+    },
+    {
+      "epoch": 1.5886524822695036,
+      "grad_norm": 0.19650763273239136,
+      "learning_rate": 6.225402504472272e-05,
+      "loss": 0.0779,
+      "step": 896
+    },
+    {
+      "epoch": 1.5904255319148937,
+      "grad_norm": 0.2170541137456894,
+      "learning_rate": 6.198568872987476e-05,
+      "loss": 0.0598,
+      "step": 897
+    },
+    {
+      "epoch": 1.5921985815602837,
+      "grad_norm": 0.19491881132125854,
+      "learning_rate": 6.171735241502683e-05,
+      "loss": 0.0681,
+      "step": 898
+    },
+    {
+      "epoch": 1.5939716312056738,
+      "grad_norm": 0.16843058168888092,
+      "learning_rate": 6.144901610017889e-05,
+      "loss": 0.0564,
+      "step": 899
+    },
+    {
+      "epoch": 1.5957446808510638,
+      "grad_norm": 0.2948700487613678,
+      "learning_rate": 6.118067978533094e-05,
+      "loss": 0.0942,
+      "step": 900
+    },
+    {
+      "epoch": 1.5975177304964538,
+      "grad_norm": 0.20082975924015045,
+      "learning_rate": 6.0912343470483e-05,
+      "loss": 0.0827,
+      "step": 901
+    },
+    {
+      "epoch": 1.599290780141844,
+      "grad_norm": 0.11655837297439575,
+      "learning_rate": 6.064400715563506e-05,
+      "loss": 0.0436,
+      "step": 902
+    },
+    {
+      "epoch": 1.601063829787234,
+      "grad_norm": 0.3066547214984894,
+      "learning_rate": 6.037567084078712e-05,
+      "loss": 0.1092,
+      "step": 903
+    },
+    {
+      "epoch": 1.602836879432624,
+      "grad_norm": 0.14038467407226562,
+      "learning_rate": 6.0107334525939175e-05,
+      "loss": 0.0393,
+      "step": 904
+    },
+    {
+      "epoch": 1.602836879432624,
+      "eval_loss": 0.05965956300497055,
+      "eval_runtime": 47.1387,
+      "eval_samples_per_second": 2.673,
+      "eval_steps_per_second": 2.673,
+      "step": 904
+    },
+    {
+      "epoch": 1.604609929078014,
+      "grad_norm": 0.20005856454372406,
+      "learning_rate": 5.983899821109123e-05,
+      "loss": 0.0717,
+      "step": 905
+    },
+    {
+      "epoch": 1.6063829787234043,
+      "grad_norm": 0.16580653190612793,
+      "learning_rate": 5.957066189624329e-05,
+      "loss": 0.0669,
+      "step": 906
+    },
+    {
+      "epoch": 1.6081560283687943,
+      "grad_norm": 0.2209959626197815,
+      "learning_rate": 5.930232558139534e-05,
+      "loss": 0.0829,
+      "step": 907
+    },
+    {
+      "epoch": 1.6099290780141844,
+      "grad_norm": 0.2260005921125412,
+      "learning_rate": 5.9033989266547405e-05,
+      "loss": 0.0732,
+      "step": 908
+    },
+    {
+      "epoch": 1.6117021276595744,
+      "grad_norm": 0.5214055180549622,
+      "learning_rate": 5.876565295169946e-05,
+      "loss": 0.1302,
+      "step": 909
+    },
+    {
+      "epoch": 1.6134751773049647,
+      "grad_norm": 0.20019935071468353,
+      "learning_rate": 5.849731663685152e-05,
+      "loss": 0.0535,
+      "step": 910
+    },
+    {
+      "epoch": 1.6152482269503547,
+      "grad_norm": 0.24594898521900177,
+      "learning_rate": 5.822898032200357e-05,
+      "loss": 0.0799,
+      "step": 911
+    },
+    {
+      "epoch": 1.6170212765957448,
+      "grad_norm": 0.19057996571063995,
+      "learning_rate": 5.796064400715563e-05,
+      "loss": 0.0338,
+      "step": 912
+    },
+    {
+      "epoch": 1.6187943262411348,
+      "grad_norm": 0.16017338633537292,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 0.0698,
+      "step": 913
+    },
+    {
+      "epoch": 1.6205673758865249,
+      "grad_norm": 0.16351930797100067,
+      "learning_rate": 5.742397137745975e-05,
+      "loss": 0.033,
+      "step": 914
+    },
+    {
+      "epoch": 1.622340425531915,
+      "grad_norm": 0.4897412657737732,
+      "learning_rate": 5.71556350626118e-05,
+      "loss": 0.2265,
+      "step": 915
+    },
+    {
+      "epoch": 1.624113475177305,
+      "grad_norm": 0.546875,
+      "learning_rate": 5.688729874776386e-05,
+      "loss": 0.2423,
+      "step": 916
+    },
+    {
+      "epoch": 1.625886524822695,
+      "grad_norm": 0.3986557722091675,
+      "learning_rate": 5.6618962432915914e-05,
+      "loss": 0.1331,
+      "step": 917
+    },
+    {
+      "epoch": 1.627659574468085,
+      "grad_norm": 0.5060171484947205,
+      "learning_rate": 5.635062611806798e-05,
+      "loss": 0.1931,
+      "step": 918
+    },
+    {
+      "epoch": 1.6294326241134751,
+      "grad_norm": 0.3128643035888672,
+      "learning_rate": 5.608228980322003e-05,
+      "loss": 0.1167,
+      "step": 919
+    },
+    {
+      "epoch": 1.6312056737588652,
+      "grad_norm": 0.42538610100746155,
+      "learning_rate": 5.581395348837209e-05,
+      "loss": 0.1636,
+      "step": 920
+    },
+    {
+      "epoch": 1.6329787234042552,
+      "grad_norm": 0.28647974133491516,
+      "learning_rate": 5.5545617173524144e-05,
+      "loss": 0.0699,
+      "step": 921
+    },
+    {
+      "epoch": 1.6347517730496453,
+      "grad_norm": 0.39569562673568726,
+      "learning_rate": 5.527728085867621e-05,
+      "loss": 0.2067,
+      "step": 922
+    },
+    {
+      "epoch": 1.6365248226950353,
+      "grad_norm": 0.2089187651872635,
+      "learning_rate": 5.500894454382826e-05,
+      "loss": 0.0992,
+      "step": 923
+    },
+    {
+      "epoch": 1.6382978723404256,
+      "grad_norm": 0.24980148673057556,
+      "learning_rate": 5.474060822898032e-05,
+      "loss": 0.0856,
+      "step": 924
+    },
+    {
+      "epoch": 1.6400709219858156,
+      "grad_norm": 0.2893138825893402,
+      "learning_rate": 5.4472271914132374e-05,
+      "loss": 0.124,
+      "step": 925
+    },
+    {
+      "epoch": 1.6418439716312057,
+      "grad_norm": 0.3664391338825226,
+      "learning_rate": 5.420393559928443e-05,
+      "loss": 0.1636,
+      "step": 926
+    },
+    {
+      "epoch": 1.6436170212765957,
+      "grad_norm": 0.31083714962005615,
+      "learning_rate": 5.393559928443649e-05,
+      "loss": 0.1389,
+      "step": 927
+    },
+    {
+      "epoch": 1.645390070921986,
+      "grad_norm": 0.335844486951828,
+      "learning_rate": 5.366726296958855e-05,
+      "loss": 0.1405,
+      "step": 928
+    },
+    {
+      "epoch": 1.647163120567376,
+      "grad_norm": 0.6708166599273682,
+      "learning_rate": 5.3398926654740604e-05,
+      "loss": 0.1981,
+      "step": 929
+    },
+    {
+      "epoch": 1.648936170212766,
+      "grad_norm": 0.3606095612049103,
+      "learning_rate": 5.313059033989266e-05,
+      "loss": 0.1331,
+      "step": 930
+    },
+    {
+      "epoch": 1.650709219858156,
+      "grad_norm": 0.265484094619751,
+      "learning_rate": 5.2862254025044716e-05,
+      "loss": 0.1015,
+      "step": 931
+    },
+    {
+      "epoch": 1.6524822695035462,
+      "grad_norm": 0.2551873028278351,
+      "learning_rate": 5.259391771019678e-05,
+      "loss": 0.113,
+      "step": 932
+    },
+    {
+      "epoch": 1.6542553191489362,
+      "grad_norm": 0.5668807625770569,
+      "learning_rate": 5.2325581395348834e-05,
+      "loss": 0.1927,
+      "step": 933
+    },
+    {
+      "epoch": 1.6560283687943262,
+      "grad_norm": 0.39409738779067993,
+      "learning_rate": 5.205724508050089e-05,
+      "loss": 0.1265,
+      "step": 934
+    },
+    {
+      "epoch": 1.6578014184397163,
+      "grad_norm": 0.3022906184196472,
+      "learning_rate": 5.1788908765652946e-05,
+      "loss": 0.1071,
+      "step": 935
+    },
+    {
+      "epoch": 1.6595744680851063,
+      "grad_norm": 0.24279522895812988,
+      "learning_rate": 5.1520572450805e-05,
+      "loss": 0.0784,
+      "step": 936
+    },
+    {
+      "epoch": 1.6613475177304964,
+      "grad_norm": 0.39648398756980896,
+      "learning_rate": 5.1252236135957064e-05,
+      "loss": 0.1585,
+      "step": 937
+    },
+    {
+      "epoch": 1.6631205673758864,
+      "grad_norm": 0.25539273023605347,
+      "learning_rate": 5.098389982110912e-05,
+      "loss": 0.0872,
+      "step": 938
+    },
+    {
+      "epoch": 1.6648936170212765,
+      "grad_norm": 0.253682404756546,
+      "learning_rate": 5.0715563506261176e-05,
+      "loss": 0.0817,
+      "step": 939
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.19512628018856049,
+      "learning_rate": 5.044722719141323e-05,
+      "loss": 0.0768,
+      "step": 940
+    },
+    {
+      "epoch": 1.6684397163120568,
+      "grad_norm": 0.25534772872924805,
+      "learning_rate": 5.0178890876565294e-05,
+      "loss": 0.0944,
+      "step": 941
+    },
+    {
+      "epoch": 1.6702127659574468,
+      "grad_norm": 0.23525482416152954,
+      "learning_rate": 4.991055456171735e-05,
+      "loss": 0.073,
+      "step": 942
+    },
+    {
+      "epoch": 1.6719858156028369,
+      "grad_norm": 0.2775448262691498,
+      "learning_rate": 4.9642218246869406e-05,
+      "loss": 0.1318,
+      "step": 943
+    },
+    {
+      "epoch": 1.673758865248227,
+      "grad_norm": 0.29373791813850403,
+      "learning_rate": 4.937388193202146e-05,
+      "loss": 0.1047,
+      "step": 944
+    },
+    {
+      "epoch": 1.675531914893617,
+      "grad_norm": 0.22552639245986938,
+      "learning_rate": 4.910554561717352e-05,
+      "loss": 0.073,
+      "step": 945
+    },
+    {
+      "epoch": 1.6773049645390072,
+      "grad_norm": 0.2918212413787842,
+      "learning_rate": 4.883720930232558e-05,
+      "loss": 0.1008,
+      "step": 946
+    },
+    {
+      "epoch": 1.6790780141843973,
+      "grad_norm": 0.3461683690547943,
+      "learning_rate": 4.8568872987477636e-05,
+      "loss": 0.114,
+      "step": 947
+    },
+    {
+      "epoch": 1.6808510638297873,
+      "grad_norm": 0.18103665113449097,
+      "learning_rate": 4.830053667262969e-05,
+      "loss": 0.0614,
+      "step": 948
+    },
+    {
+      "epoch": 1.6826241134751774,
+      "grad_norm": 0.28270408511161804,
+      "learning_rate": 4.803220035778175e-05,
+      "loss": 0.0943,
+      "step": 949
+    },
+    {
+      "epoch": 1.6843971631205674,
+      "grad_norm": 0.2964995503425598,
+      "learning_rate": 4.77638640429338e-05,
+      "loss": 0.1005,
+      "step": 950
+    },
+    {
+      "epoch": 1.6861702127659575,
+      "grad_norm": 0.19928471744060516,
+      "learning_rate": 4.7495527728085866e-05,
+      "loss": 0.0776,
+      "step": 951
+    },
+    {
+      "epoch": 1.6879432624113475,
+      "grad_norm": 0.21275044977664948,
+      "learning_rate": 4.722719141323792e-05,
+      "loss": 0.0763,
+      "step": 952
+    },
+    {
+      "epoch": 1.6897163120567376,
+      "grad_norm": 0.13716483116149902,
+      "learning_rate": 4.695885509838998e-05,
+      "loss": 0.042,
+      "step": 953
+    },
+    {
+      "epoch": 1.6914893617021276,
+      "grad_norm": 0.27446210384368896,
+      "learning_rate": 4.669051878354203e-05,
+      "loss": 0.1125,
+      "step": 954
+    },
+    {
+      "epoch": 1.6932624113475176,
+      "grad_norm": 0.17361652851104736,
+      "learning_rate": 4.642218246869409e-05,
+      "loss": 0.0656,
+      "step": 955
+    },
+    {
+      "epoch": 1.6950354609929077,
+      "grad_norm": 0.22053229808807373,
+      "learning_rate": 4.615384615384615e-05,
+      "loss": 0.085,
+      "step": 956
+    },
+    {
+      "epoch": 1.6968085106382977,
+      "grad_norm": 0.2688538730144501,
+      "learning_rate": 4.588550983899821e-05,
+      "loss": 0.1084,
+      "step": 957
+    },
+    {
+      "epoch": 1.6985815602836878,
+      "grad_norm": 0.17203108966350555,
+      "learning_rate": 4.561717352415026e-05,
+      "loss": 0.0711,
+      "step": 958
+    },
+    {
+      "epoch": 1.700354609929078,
+      "grad_norm": 0.2185698002576828,
+      "learning_rate": 4.534883720930232e-05,
+      "loss": 0.0906,
+      "step": 959
+    },
+    {
+      "epoch": 1.702127659574468,
+      "grad_norm": 0.3515457808971405,
+      "learning_rate": 4.508050089445438e-05,
+      "loss": 0.1529,
+      "step": 960
+    },
+    {
+      "epoch": 1.7039007092198581,
+      "grad_norm": 0.1280314177274704,
+      "learning_rate": 4.481216457960644e-05,
+      "loss": 0.0425,
+      "step": 961
+    },
+    {
+      "epoch": 1.7056737588652482,
+      "grad_norm": 0.16565589606761932,
+      "learning_rate": 4.454382826475849e-05,
+      "loss": 0.0406,
+      "step": 962
+    },
+    {
+      "epoch": 1.7074468085106385,
+      "grad_norm": 0.17467603087425232,
+      "learning_rate": 4.427549194991055e-05,
+      "loss": 0.0369,
+      "step": 963
+    },
+    {
+      "epoch": 1.7092198581560285,
+      "grad_norm": 0.17544378340244293,
+      "learning_rate": 4.4007155635062605e-05,
+      "loss": 0.0434,
+      "step": 964
+    },
+    {
+      "epoch": 1.7109929078014185,
+      "grad_norm": 0.42301613092422485,
+      "learning_rate": 4.373881932021467e-05,
+      "loss": 0.1475,
+      "step": 965
+    },
+    {
+      "epoch": 1.7127659574468086,
+      "grad_norm": 0.3776220977306366,
+      "learning_rate": 4.347048300536672e-05,
+      "loss": 0.1585,
+      "step": 966
+    },
+    {
+      "epoch": 1.7145390070921986,
+      "grad_norm": 0.35067784786224365,
+      "learning_rate": 4.320214669051878e-05,
+      "loss": 0.1329,
+      "step": 967
+    },
+    {
+      "epoch": 1.7163120567375887,
+      "grad_norm": 0.2998995780944824,
+      "learning_rate": 4.2933810375670835e-05,
+      "loss": 0.0946,
+      "step": 968
+    },
+    {
+      "epoch": 1.7180851063829787,
+      "grad_norm": 0.31156042218208313,
+      "learning_rate": 4.266547406082289e-05,
+      "loss": 0.1103,
+      "step": 969
+    },
+    {
+      "epoch": 1.7198581560283688,
+      "grad_norm": 0.4217228889465332,
+      "learning_rate": 4.239713774597495e-05,
+      "loss": 0.1722,
+      "step": 970
+    },
+    {
+      "epoch": 1.7216312056737588,
+      "grad_norm": 0.4580366015434265,
+      "learning_rate": 4.212880143112701e-05,
+      "loss": 0.1871,
+      "step": 971
+    },
+    {
+      "epoch": 1.7234042553191489,
+      "grad_norm": 0.3506891429424286,
+      "learning_rate": 4.1860465116279065e-05,
+      "loss": 0.134,
+      "step": 972
+    },
+    {
+      "epoch": 1.725177304964539,
+      "grad_norm": 0.5243451595306396,
+      "learning_rate": 4.159212880143112e-05,
+      "loss": 0.147,
+      "step": 973
+    },
+    {
+      "epoch": 1.726950354609929,
+      "grad_norm": 0.3330957889556885,
+      "learning_rate": 4.1323792486583176e-05,
+      "loss": 0.1151,
+      "step": 974
+    },
+    {
+      "epoch": 1.728723404255319,
+      "grad_norm": 0.2872239053249359,
+      "learning_rate": 4.105545617173524e-05,
+      "loss": 0.1414,
+      "step": 975
+    },
+    {
+      "epoch": 1.7304964539007093,
+      "grad_norm": 0.31004229187965393,
+      "learning_rate": 4.0787119856887295e-05,
+      "loss": 0.1331,
+      "step": 976
+    },
+    {
+      "epoch": 1.7322695035460993,
+      "grad_norm": 0.3493654131889343,
+      "learning_rate": 4.051878354203935e-05,
+      "loss": 0.1438,
+      "step": 977
+    },
+    {
+      "epoch": 1.7340425531914894,
+      "grad_norm": 0.4803168773651123,
+      "learning_rate": 4.0250447227191406e-05,
+      "loss": 0.1709,
+      "step": 978
+    },
+    {
+      "epoch": 1.7358156028368794,
+      "grad_norm": 0.2807621657848358,
+      "learning_rate": 3.998211091234347e-05,
+      "loss": 0.112,
+      "step": 979
+    },
+    {
+      "epoch": 1.7375886524822695,
+      "grad_norm": 0.4310925006866455,
+      "learning_rate": 3.9713774597495524e-05,
+      "loss": 0.1529,
+      "step": 980
+    },
+    {
+      "epoch": 1.7393617021276597,
+      "grad_norm": 0.5579214692115784,
+      "learning_rate": 3.944543828264758e-05,
+      "loss": 0.2199,
+      "step": 981
+    },
+    {
+      "epoch": 1.7411347517730498,
+      "grad_norm": 0.24788203835487366,
+      "learning_rate": 3.9177101967799636e-05,
+      "loss": 0.0795,
+      "step": 982
+    },
+    {
+      "epoch": 1.7429078014184398,
+      "grad_norm": 0.35569068789482117,
+      "learning_rate": 3.890876565295169e-05,
+      "loss": 0.1494,
+      "step": 983
+    },
+    {
+      "epoch": 1.7446808510638299,
+      "grad_norm": 0.26108360290527344,
+      "learning_rate": 3.8640429338103754e-05,
+      "loss": 0.1172,
+      "step": 984
+    },
+    {
+      "epoch": 1.74645390070922,
+      "grad_norm": 0.2628544569015503,
+      "learning_rate": 3.837209302325581e-05,
+      "loss": 0.1146,
+      "step": 985
+    },
+    {
+      "epoch": 1.74822695035461,
+      "grad_norm": 0.267870157957077,
+      "learning_rate": 3.8103756708407866e-05,
+      "loss": 0.1053,
+      "step": 986
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.3323216140270233,
+      "learning_rate": 3.783542039355992e-05,
+      "loss": 0.1278,
+      "step": 987
+    },
+    {
+      "epoch": 1.75177304964539,
+      "grad_norm": 0.36032265424728394,
+      "learning_rate": 3.756708407871198e-05,
+      "loss": 0.1392,
+      "step": 988
+    },
+    {
+      "epoch": 1.75354609929078,
+      "grad_norm": 0.23055288195610046,
+      "learning_rate": 3.729874776386404e-05,
+      "loss": 0.0875,
+      "step": 989
+    },
+    {
+      "epoch": 1.7553191489361701,
+      "grad_norm": 0.19896911084651947,
+      "learning_rate": 3.7030411449016096e-05,
+      "loss": 0.0744,
+      "step": 990
+    },
+    {
+      "epoch": 1.7570921985815602,
+      "grad_norm": 0.2982289791107178,
+      "learning_rate": 3.676207513416815e-05,
+      "loss": 0.1312,
+      "step": 991
+    },
+    {
+      "epoch": 1.7588652482269502,
+      "grad_norm": 0.18045595288276672,
+      "learning_rate": 3.6493738819320214e-05,
+      "loss": 0.0621,
+      "step": 992
+    },
+    {
+      "epoch": 1.7606382978723403,
+      "grad_norm": 0.3549979627132416,
+      "learning_rate": 3.622540250447227e-05,
+      "loss": 0.147,
+      "step": 993
+    },
+    {
+      "epoch": 1.7624113475177305,
+      "grad_norm": 0.2849663496017456,
+      "learning_rate": 3.5957066189624326e-05,
+      "loss": 0.0958,
+      "step": 994
+    },
+    {
+      "epoch": 1.7641843971631206,
+      "grad_norm": 0.19186846911907196,
+      "learning_rate": 3.568872987477638e-05,
+      "loss": 0.0717,
+      "step": 995
+    },
+    {
+      "epoch": 1.7659574468085106,
+      "grad_norm": 0.19222399592399597,
+      "learning_rate": 3.542039355992844e-05,
+      "loss": 0.0448,
+      "step": 996
+    },
+    {
+      "epoch": 1.7677304964539007,
+      "grad_norm": 0.36738601326942444,
+      "learning_rate": 3.51520572450805e-05,
+      "loss": 0.1569,
+      "step": 997
+    },
+    {
+      "epoch": 1.7695035460992907,
+      "grad_norm": 0.3227214813232422,
+      "learning_rate": 3.4883720930232556e-05,
+      "loss": 0.1288,
+      "step": 998
+    },
+    {
+      "epoch": 1.771276595744681,
+      "grad_norm": 0.2572925090789795,
+      "learning_rate": 3.461538461538461e-05,
+      "loss": 0.0916,
+      "step": 999
+    },
+    {
+      "epoch": 1.773049645390071,
+      "grad_norm": 0.23549784719944,
+      "learning_rate": 3.434704830053667e-05,
+      "loss": 0.0891,
+      "step": 1000
+    },
+    {
+      "epoch": 1.774822695035461,
+      "grad_norm": 0.37821611762046814,
+      "learning_rate": 3.4078711985688723e-05,
+      "loss": 0.1633,
+      "step": 1001
+    },
+    {
+      "epoch": 1.7765957446808511,
+      "grad_norm": 0.24105395376682281,
+      "learning_rate": 3.3810375670840786e-05,
+      "loss": 0.104,
+      "step": 1002
+    },
+    {
+      "epoch": 1.7783687943262412,
+      "grad_norm": 0.21173512935638428,
+      "learning_rate": 3.354203935599284e-05,
+      "loss": 0.0903,
+      "step": 1003
+    },
+    {
+      "epoch": 1.7801418439716312,
+      "grad_norm": 0.4547687768936157,
+      "learning_rate": 3.32737030411449e-05,
+      "loss": 0.1414,
+      "step": 1004
+    },
+    {
+      "epoch": 1.7819148936170213,
+      "grad_norm": 0.2635003328323364,
+      "learning_rate": 3.3005366726296953e-05,
+      "loss": 0.115,
+      "step": 1005
+    },
+    {
+      "epoch": 1.7836879432624113,
+      "grad_norm": 0.20460587739944458,
+      "learning_rate": 3.273703041144901e-05,
+      "loss": 0.066,
+      "step": 1006
+    },
+    {
+      "epoch": 1.7854609929078014,
+      "grad_norm": 0.2040269821882248,
+      "learning_rate": 3.246869409660107e-05,
+      "loss": 0.0607,
+      "step": 1007
+    },
+    {
+      "epoch": 1.7872340425531914,
+      "grad_norm": 0.29658254981040955,
+      "learning_rate": 3.220035778175313e-05,
+      "loss": 0.1239,
+      "step": 1008
+    },
+    {
+      "epoch": 1.7890070921985815,
+      "grad_norm": 0.1401108056306839,
+      "learning_rate": 3.1932021466905183e-05,
+      "loss": 0.0552,
+      "step": 1009
+    },
+    {
+      "epoch": 1.7907801418439715,
+      "grad_norm": 0.433570921421051,
+      "learning_rate": 3.166368515205724e-05,
+      "loss": 0.1549,
+      "step": 1010
+    },
+    {
+      "epoch": 1.7925531914893615,
+      "grad_norm": 0.24940761923789978,
+      "learning_rate": 3.13953488372093e-05,
+      "loss": 0.0958,
+      "step": 1011
+    },
+    {
+      "epoch": 1.7943262411347518,
+      "grad_norm": 0.18111471831798553,
+      "learning_rate": 3.112701252236136e-05,
+      "loss": 0.0498,
+      "step": 1012
+    },
+    {
+      "epoch": 1.7960992907801419,
+      "grad_norm": 0.15693312883377075,
+      "learning_rate": 3.085867620751341e-05,
+      "loss": 0.0358,
+      "step": 1013
+    },
+    {
+      "epoch": 1.797872340425532,
+      "grad_norm": 0.13912533223628998,
+      "learning_rate": 3.059033989266547e-05,
+      "loss": 0.0433,
+      "step": 1014
+    },
+    {
+      "epoch": 1.799645390070922,
+      "grad_norm": 0.39377179741859436,
+      "learning_rate": 3.032200357781753e-05,
+      "loss": 0.1744,
+      "step": 1015
+    },
+    {
+      "epoch": 1.8014184397163122,
+      "grad_norm": 0.558695912361145,
+      "learning_rate": 3.0053667262969588e-05,
+      "loss": 0.1799,
+      "step": 1016
+    },
+    {
+      "epoch": 1.8031914893617023,
+      "grad_norm": 0.49997296929359436,
+      "learning_rate": 2.9785330948121643e-05,
+      "loss": 0.1481,
+      "step": 1017
+    },
+    {
+      "epoch": 1.8049645390070923,
+      "grad_norm": 0.23218946158885956,
+      "learning_rate": 2.9516994633273703e-05,
+      "loss": 0.0783,
+      "step": 1018
+    },
+    {
+      "epoch": 1.8067375886524824,
+      "grad_norm": 0.3141940236091614,
+      "learning_rate": 2.924865831842576e-05,
+      "loss": 0.0942,
+      "step": 1019
+    },
+    {
+      "epoch": 1.8085106382978724,
+      "grad_norm": 0.4221852421760559,
+      "learning_rate": 2.8980322003577814e-05,
+      "loss": 0.1216,
+      "step": 1020
+    },
+    {
+      "epoch": 1.8102836879432624,
+      "grad_norm": 0.21606023609638214,
+      "learning_rate": 2.8711985688729873e-05,
+      "loss": 0.0776,
+      "step": 1021
+    },
+    {
+      "epoch": 1.8120567375886525,
+      "grad_norm": 0.3150366246700287,
+      "learning_rate": 2.844364937388193e-05,
+      "loss": 0.1131,
+      "step": 1022
+    },
+    {
+      "epoch": 1.8138297872340425,
+      "grad_norm": 0.5450868010520935,
+      "learning_rate": 2.817531305903399e-05,
+      "loss": 0.1853,
+      "step": 1023
+    },
+    {
+      "epoch": 1.8156028368794326,
+      "grad_norm": 0.21700796484947205,
+      "learning_rate": 2.7906976744186044e-05,
+      "loss": 0.0988,
+      "step": 1024
+    },
+    {
+      "epoch": 1.8173758865248226,
+      "grad_norm": 0.23696239292621613,
+      "learning_rate": 2.7638640429338103e-05,
+      "loss": 0.1032,
+      "step": 1025
+    },
+    {
+      "epoch": 1.8191489361702127,
+      "grad_norm": 0.4778185486793518,
+      "learning_rate": 2.737030411449016e-05,
+      "loss": 0.1505,
+      "step": 1026
+    },
+    {
+      "epoch": 1.8209219858156027,
+      "grad_norm": 0.5296378135681152,
+      "learning_rate": 2.7101967799642215e-05,
+      "loss": 0.1929,
+      "step": 1027
+    },
+    {
+      "epoch": 1.8226950354609928,
+      "grad_norm": 0.4214850962162018,
+      "learning_rate": 2.6833631484794274e-05,
+      "loss": 0.1353,
+      "step": 1028
+    },
+    {
+      "epoch": 1.824468085106383,
+      "grad_norm": 0.41558897495269775,
+      "learning_rate": 2.656529516994633e-05,
+      "loss": 0.1337,
+      "step": 1029
+    },
+    {
+      "epoch": 1.826241134751773,
+      "grad_norm": 0.18530499935150146,
+      "learning_rate": 2.629695885509839e-05,
+      "loss": 0.0798,
+      "step": 1030
+    },
+    {
+      "epoch": 1.8280141843971631,
+      "grad_norm": 0.26249393820762634,
+      "learning_rate": 2.6028622540250445e-05,
+      "loss": 0.1154,
+      "step": 1031
+    },
+    {
+      "epoch": 1.8297872340425532,
+      "grad_norm": 0.252618670463562,
+      "learning_rate": 2.57602862254025e-05,
+      "loss": 0.0912,
+      "step": 1032
+    },
+    {
+      "epoch": 1.8315602836879432,
+      "grad_norm": 0.24905292689800262,
+      "learning_rate": 2.549194991055456e-05,
+      "loss": 0.0882,
+      "step": 1033
+    },
+    {
+      "epoch": 1.8333333333333335,
+      "grad_norm": 0.37883955240249634,
+      "learning_rate": 2.5223613595706616e-05,
+      "loss": 0.1248,
+      "step": 1034
+    },
+    {
+      "epoch": 1.8351063829787235,
+      "grad_norm": 0.31058332324028015,
+      "learning_rate": 2.4955277280858675e-05,
+      "loss": 0.1237,
+      "step": 1035
+    },
+    {
+      "epoch": 1.8368794326241136,
+      "grad_norm": 0.2710226774215698,
+      "learning_rate": 2.468694096601073e-05,
+      "loss": 0.1072,
+      "step": 1036
+    },
+    {
+      "epoch": 1.8386524822695036,
+      "grad_norm": 0.31133633852005005,
+      "learning_rate": 2.441860465116279e-05,
+      "loss": 0.1199,
+      "step": 1037
+    },
+    {
+      "epoch": 1.8404255319148937,
+      "grad_norm": 0.19131186604499817,
+      "learning_rate": 2.4150268336314846e-05,
+      "loss": 0.0885,
+      "step": 1038
+    },
+    {
+      "epoch": 1.8421985815602837,
+      "grad_norm": 0.34097999334335327,
+      "learning_rate": 2.38819320214669e-05,
+      "loss": 0.1198,
+      "step": 1039
+    },
+    {
+      "epoch": 1.8439716312056738,
+      "grad_norm": 0.1662115752696991,
+      "learning_rate": 2.361359570661896e-05,
+      "loss": 0.0724,
+      "step": 1040
+    },
+    {
+      "epoch": 1.8457446808510638,
+      "grad_norm": 0.16127242147922516,
+      "learning_rate": 2.3345259391771016e-05,
+      "loss": 0.0647,
+      "step": 1041
+    },
+    {
+      "epoch": 1.8475177304964538,
+      "grad_norm": 0.18112505972385406,
+      "learning_rate": 2.3076923076923076e-05,
+      "loss": 0.0651,
+      "step": 1042
+    },
+    {
+      "epoch": 1.849290780141844,
+      "grad_norm": 0.1682027280330658,
+      "learning_rate": 2.280858676207513e-05,
+      "loss": 0.0542,
+      "step": 1043
+    },
+    {
+      "epoch": 1.851063829787234,
+      "grad_norm": 0.3057568669319153,
+      "learning_rate": 2.254025044722719e-05,
+      "loss": 0.1143,
+      "step": 1044
+    },
+    {
+      "epoch": 1.852836879432624,
+      "grad_norm": 0.29910141229629517,
+      "learning_rate": 2.2271914132379246e-05,
+      "loss": 0.1168,
+      "step": 1045
+    },
+    {
+      "epoch": 1.854609929078014,
+      "grad_norm": 0.23932576179504395,
+      "learning_rate": 2.2003577817531302e-05,
+      "loss": 0.0988,
+      "step": 1046
+    },
+    {
+      "epoch": 1.8563829787234043,
+      "grad_norm": 0.25302958488464355,
+      "learning_rate": 2.173524150268336e-05,
+      "loss": 0.0843,
+      "step": 1047
+    },
+    {
+      "epoch": 1.8581560283687943,
+      "grad_norm": 0.30847910046577454,
+      "learning_rate": 2.1466905187835417e-05,
+      "loss": 0.1117,
+      "step": 1048
+    },
+    {
+      "epoch": 1.8599290780141844,
+      "grad_norm": 0.22233125567436218,
+      "learning_rate": 2.1198568872987476e-05,
+      "loss": 0.0926,
+      "step": 1049
+    },
+    {
+      "epoch": 1.8617021276595744,
+      "grad_norm": 0.4215439260005951,
+      "learning_rate": 2.0930232558139532e-05,
+      "loss": 0.1222,
+      "step": 1050
+    },
+    {
+      "epoch": 1.8634751773049647,
+      "grad_norm": 0.25102686882019043,
+      "learning_rate": 2.0661896243291588e-05,
+      "loss": 0.076,
+      "step": 1051
+    },
+    {
+      "epoch": 1.8652482269503547,
+      "grad_norm": 0.20676009356975555,
+      "learning_rate": 2.0393559928443647e-05,
+      "loss": 0.0927,
+      "step": 1052
+    },
+    {
+      "epoch": 1.8670212765957448,
+      "grad_norm": 0.16590143740177155,
+      "learning_rate": 2.0125223613595703e-05,
+      "loss": 0.0437,
+      "step": 1053
+    },
+    {
+      "epoch": 1.8687943262411348,
+      "grad_norm": 0.30797678232192993,
+      "learning_rate": 1.9856887298747762e-05,
+      "loss": 0.1162,
+      "step": 1054
+    },
+    {
+      "epoch": 1.8705673758865249,
+      "grad_norm": 0.24597567319869995,
+      "learning_rate": 1.9588550983899818e-05,
+      "loss": 0.0736,
+      "step": 1055
+    },
+    {
+      "epoch": 1.872340425531915,
+      "grad_norm": 0.21166519820690155,
+      "learning_rate": 1.9320214669051877e-05,
+      "loss": 0.0729,
+      "step": 1056
+    },
+    {
+      "epoch": 1.874113475177305,
+      "grad_norm": 0.18480858206748962,
+      "learning_rate": 1.9051878354203933e-05,
+      "loss": 0.0612,
+      "step": 1057
+    },
+    {
+      "epoch": 1.875886524822695,
+      "grad_norm": 0.2520684599876404,
+      "learning_rate": 1.878354203935599e-05,
+      "loss": 0.0676,
+      "step": 1058
+    },
+    {
+      "epoch": 1.877659574468085,
+      "grad_norm": 0.22201155126094818,
+      "learning_rate": 1.8515205724508048e-05,
+      "loss": 0.0732,
+      "step": 1059
+    },
+    {
+      "epoch": 1.8794326241134751,
+      "grad_norm": 0.24273940920829773,
+      "learning_rate": 1.8246869409660107e-05,
+      "loss": 0.075,
+      "step": 1060
+    },
+    {
+      "epoch": 1.8812056737588652,
+      "grad_norm": 0.33823683857917786,
+      "learning_rate": 1.7978533094812163e-05,
+      "loss": 0.1274,
+      "step": 1061
+    },
+    {
+      "epoch": 1.8829787234042552,
+      "grad_norm": 0.24322524666786194,
+      "learning_rate": 1.771019677996422e-05,
+      "loss": 0.062,
+      "step": 1062
+    },
+    {
+      "epoch": 1.8847517730496453,
+      "grad_norm": 0.2402666211128235,
+      "learning_rate": 1.7441860465116278e-05,
+      "loss": 0.0688,
+      "step": 1063
+    },
+    {
+      "epoch": 1.8865248226950353,
+      "grad_norm": 0.156381294131279,
+      "learning_rate": 1.7173524150268334e-05,
+      "loss": 0.0499,
+      "step": 1064
+    },
+    {
+      "epoch": 1.8882978723404256,
+      "grad_norm": 0.5312046408653259,
+      "learning_rate": 1.6905187835420393e-05,
+      "loss": 0.1921,
+      "step": 1065
+    },
+    {
+      "epoch": 1.8900709219858156,
+      "grad_norm": 0.33548104763031006,
+      "learning_rate": 1.663685152057245e-05,
+      "loss": 0.1373,
+      "step": 1066
+    },
+    {
+      "epoch": 1.8918439716312057,
+      "grad_norm": 0.5181596875190735,
+      "learning_rate": 1.6368515205724505e-05,
+      "loss": 0.092,
+      "step": 1067
+    },
+    {
+      "epoch": 1.8936170212765957,
+      "grad_norm": 0.38026565313339233,
+      "learning_rate": 1.6100178890876564e-05,
+      "loss": 0.1344,
+      "step": 1068
+    },
+    {
+      "epoch": 1.895390070921986,
+      "grad_norm": 0.4322291612625122,
+      "learning_rate": 1.583184257602862e-05,
+      "loss": 0.1421,
+      "step": 1069
+    },
+    {
+      "epoch": 1.897163120567376,
+      "grad_norm": 0.24601012468338013,
+      "learning_rate": 1.556350626118068e-05,
+      "loss": 0.0664,
+      "step": 1070
+    },
+    {
+      "epoch": 1.898936170212766,
+      "grad_norm": 0.2838169038295746,
+      "learning_rate": 1.5295169946332735e-05,
+      "loss": 0.0987,
+      "step": 1071
+    },
+    {
+      "epoch": 1.900709219858156,
+      "grad_norm": 0.2708532512187958,
+      "learning_rate": 1.5026833631484794e-05,
+      "loss": 0.116,
+      "step": 1072
+    },
+    {
+      "epoch": 1.9024822695035462,
+      "grad_norm": 0.48430347442626953,
+      "learning_rate": 1.4758497316636851e-05,
+      "loss": 0.1737,
+      "step": 1073
+    },
+    {
+      "epoch": 1.9042553191489362,
+      "grad_norm": 0.2883581817150116,
+      "learning_rate": 1.4490161001788907e-05,
+      "loss": 0.1058,
+      "step": 1074
+    },
+    {
+      "epoch": 1.9060283687943262,
+      "grad_norm": 0.38699519634246826,
+      "learning_rate": 1.4221824686940965e-05,
+      "loss": 0.1039,
+      "step": 1075
+    },
+    {
+      "epoch": 1.9078014184397163,
+      "grad_norm": 0.4487515389919281,
+      "learning_rate": 1.3953488372093022e-05,
+      "loss": 0.1794,
+      "step": 1076
+    },
+    {
+      "epoch": 1.9095744680851063,
+      "grad_norm": 0.3338315486907959,
+      "learning_rate": 1.368515205724508e-05,
+      "loss": 0.127,
+      "step": 1077
+    },
+    {
+      "epoch": 1.9113475177304964,
+      "grad_norm": 0.32425758242607117,
+      "learning_rate": 1.3416815742397137e-05,
+      "loss": 0.133,
+      "step": 1078
+    },
+    {
+      "epoch": 1.9131205673758864,
+      "grad_norm": 0.25057604908943176,
+      "learning_rate": 1.3148479427549195e-05,
+      "loss": 0.1069,
+      "step": 1079
+    },
+    {
+      "epoch": 1.9148936170212765,
+      "grad_norm": 0.2886715233325958,
+      "learning_rate": 1.288014311270125e-05,
+      "loss": 0.1326,
+      "step": 1080
+    },
+    {
+      "epoch": 1.9166666666666665,
+      "grad_norm": 0.27996861934661865,
+      "learning_rate": 1.2611806797853308e-05,
+      "loss": 0.0981,
+      "step": 1081
+    },
+    {
+      "epoch": 1.9184397163120568,
+      "grad_norm": 0.38353902101516724,
+      "learning_rate": 1.2343470483005365e-05,
+      "loss": 0.1677,
+      "step": 1082
+    },
+    {
+      "epoch": 1.9202127659574468,
+      "grad_norm": 0.18619371950626373,
+      "learning_rate": 1.2075134168157423e-05,
+      "loss": 0.0741,
+      "step": 1083
+    },
+    {
+      "epoch": 1.9219858156028369,
+      "grad_norm": 0.23374900221824646,
+      "learning_rate": 1.180679785330948e-05,
+      "loss": 0.085,
+      "step": 1084
+    },
+    {
+      "epoch": 1.923758865248227,
+      "grad_norm": 0.439928263425827,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.1432,
+      "step": 1085
+    },
+    {
+      "epoch": 1.925531914893617,
+      "grad_norm": 0.3650781512260437,
+      "learning_rate": 1.1270125223613595e-05,
+      "loss": 0.127,
+      "step": 1086
+    },
+    {
+      "epoch": 1.9273049645390072,
+      "grad_norm": 0.47360095381736755,
+      "learning_rate": 1.1001788908765651e-05,
+      "loss": 0.1805,
+      "step": 1087
+    },
+    {
+      "epoch": 1.9290780141843973,
+      "grad_norm": 0.3956616520881653,
+      "learning_rate": 1.0733452593917709e-05,
+      "loss": 0.1409,
+      "step": 1088
+    },
+    {
+      "epoch": 1.9308510638297873,
+      "grad_norm": 0.2528573274612427,
+      "learning_rate": 1.0465116279069766e-05,
+      "loss": 0.0986,
+      "step": 1089
+    },
+    {
+      "epoch": 1.9326241134751774,
+      "grad_norm": 0.5904862880706787,
+      "learning_rate": 1.0196779964221824e-05,
+      "loss": 0.1442,
+      "step": 1090
+    },
+    {
+      "epoch": 1.9343971631205674,
+      "grad_norm": 0.41600659489631653,
+      "learning_rate": 9.928443649373881e-06,
+      "loss": 0.1124,
+      "step": 1091
+    },
+    {
+      "epoch": 1.9361702127659575,
+      "grad_norm": 0.17834167182445526,
+      "learning_rate": 9.660107334525939e-06,
+      "loss": 0.0693,
+      "step": 1092
+    },
+    {
+      "epoch": 1.9379432624113475,
+      "grad_norm": 0.17482277750968933,
+      "learning_rate": 9.391771019677994e-06,
+      "loss": 0.0509,
+      "step": 1093
+    },
+    {
+      "epoch": 1.9397163120567376,
+      "grad_norm": 0.16638445854187012,
+      "learning_rate": 9.123434704830054e-06,
+      "loss": 0.0497,
+      "step": 1094
+    },
+    {
+      "epoch": 1.9414893617021276,
+      "grad_norm": 0.29362753033638,
+      "learning_rate": 8.85509838998211e-06,
+      "loss": 0.1001,
+      "step": 1095
+    },
+    {
+      "epoch": 1.9432624113475176,
+      "grad_norm": 0.13863207399845123,
+      "learning_rate": 8.586762075134167e-06,
+      "loss": 0.0594,
+      "step": 1096
+    },
+    {
+      "epoch": 1.9450354609929077,
+      "grad_norm": 0.33512240648269653,
+      "learning_rate": 8.318425760286224e-06,
+      "loss": 0.1273,
+      "step": 1097
+    },
+    {
+      "epoch": 1.9468085106382977,
+      "grad_norm": 0.3097001612186432,
+      "learning_rate": 8.050089445438282e-06,
+      "loss": 0.1164,
+      "step": 1098
+    },
+    {
+      "epoch": 1.9485815602836878,
+      "grad_norm": 0.2611374855041504,
+      "learning_rate": 7.78175313059034e-06,
+      "loss": 0.1004,
+      "step": 1099
+    },
+    {
+      "epoch": 1.950354609929078,
+      "grad_norm": 0.14972633123397827,
+      "learning_rate": 7.513416815742397e-06,
+      "loss": 0.0577,
+      "step": 1100
+    },
+    {
+      "epoch": 1.952127659574468,
+      "grad_norm": 0.3071126937866211,
+      "learning_rate": 7.2450805008944535e-06,
+      "loss": 0.091,
+      "step": 1101
+    },
+    {
+      "epoch": 1.9539007092198581,
+      "grad_norm": 0.1802784502506256,
+      "learning_rate": 6.976744186046511e-06,
+      "loss": 0.0493,
+      "step": 1102
+    },
+    {
+      "epoch": 1.9556737588652482,
+      "grad_norm": 0.15572038292884827,
+      "learning_rate": 6.7084078711985685e-06,
+      "loss": 0.0579,
+      "step": 1103
+    },
+    {
+      "epoch": 1.9574468085106385,
+      "grad_norm": 0.24851687252521515,
+      "learning_rate": 6.440071556350625e-06,
+      "loss": 0.0871,
+      "step": 1104
+    },
+    {
+      "epoch": 1.9592198581560285,
+      "grad_norm": 0.2088288813829422,
+      "learning_rate": 6.171735241502683e-06,
+      "loss": 0.0733,
+      "step": 1105
+    },
+    {
+      "epoch": 1.9609929078014185,
+      "grad_norm": 0.24039943516254425,
+      "learning_rate": 5.90339892665474e-06,
+      "loss": 0.093,
+      "step": 1106
+    },
+    {
+      "epoch": 1.9627659574468086,
+      "grad_norm": 0.3546561002731323,
+      "learning_rate": 5.635062611806798e-06,
+      "loss": 0.1246,
+      "step": 1107
+    },
+    {
+      "epoch": 1.9645390070921986,
+      "grad_norm": 0.11703565716743469,
+      "learning_rate": 5.366726296958854e-06,
+      "loss": 0.0402,
+      "step": 1108
+    },
+    {
+      "epoch": 1.9663120567375887,
+      "grad_norm": 0.1947447657585144,
+      "learning_rate": 5.098389982110912e-06,
+      "loss": 0.0761,
+      "step": 1109
+    },
+    {
+      "epoch": 1.9680851063829787,
+      "grad_norm": 0.18110178411006927,
+      "learning_rate": 4.830053667262969e-06,
+      "loss": 0.0768,
+      "step": 1110
+    },
+    {
+      "epoch": 1.9698581560283688,
+      "grad_norm": 0.2154490053653717,
+      "learning_rate": 4.561717352415027e-06,
+      "loss": 0.0813,
+      "step": 1111
+    },
+    {
+      "epoch": 1.9716312056737588,
+      "grad_norm": 0.20867429673671722,
+      "learning_rate": 4.2933810375670835e-06,
+      "loss": 0.0812,
+      "step": 1112
+    },
+    {
+      "epoch": 1.9734042553191489,
+      "grad_norm": 0.1496736854314804,
+      "learning_rate": 4.025044722719141e-06,
+      "loss": 0.046,
+      "step": 1113
+    },
+    {
+      "epoch": 1.975177304964539,
+      "grad_norm": 0.10855786502361298,
+      "learning_rate": 3.7567084078711984e-06,
+      "loss": 0.0341,
+      "step": 1114
+    },
+    {
+      "epoch": 1.976950354609929,
+      "grad_norm": 0.2541455626487732,
+      "learning_rate": 3.4883720930232555e-06,
+      "loss": 0.0879,
+      "step": 1115
+    },
+    {
+      "epoch": 1.978723404255319,
+      "grad_norm": 0.30963069200515747,
+      "learning_rate": 3.2200357781753126e-06,
+      "loss": 0.1081,
+      "step": 1116
+    },
+    {
+      "epoch": 1.9804964539007093,
+      "grad_norm": 0.5179100632667542,
+      "learning_rate": 2.95169946332737e-06,
+      "loss": 0.1674,
+      "step": 1117
+    },
+    {
+      "epoch": 1.9822695035460993,
+      "grad_norm": 0.208583265542984,
+      "learning_rate": 2.683363148479427e-06,
+      "loss": 0.0861,
+      "step": 1118
+    },
+    {
+      "epoch": 1.9840425531914894,
+      "grad_norm": 0.32043734192848206,
+      "learning_rate": 2.4150268336314847e-06,
+      "loss": 0.104,
+      "step": 1119
+    },
+    {
+      "epoch": 1.9858156028368794,
+      "grad_norm": 0.16812080144882202,
+      "learning_rate": 2.1466905187835417e-06,
+      "loss": 0.0568,
+      "step": 1120
+    },
+    {
+      "epoch": 1.9875886524822695,
+      "grad_norm": 0.3949272632598877,
+      "learning_rate": 1.8783542039355992e-06,
+      "loss": 0.1243,
+      "step": 1121
+    },
+    {
+      "epoch": 1.9893617021276597,
+      "grad_norm": 0.30382081866264343,
+      "learning_rate": 1.6100178890876563e-06,
+      "loss": 0.0929,
+      "step": 1122
+    },
+    {
+      "epoch": 1.9911347517730498,
+      "grad_norm": 0.3701658844947815,
+      "learning_rate": 1.3416815742397136e-06,
+      "loss": 0.1313,
+      "step": 1123
+    },
+    {
+      "epoch": 1.9929078014184398,
+      "grad_norm": 0.3137078285217285,
+      "learning_rate": 1.0733452593917709e-06,
+      "loss": 0.1288,
+      "step": 1124
+    },
+    {
+      "epoch": 1.9946808510638299,
+      "grad_norm": 0.20329152047634125,
+      "learning_rate": 8.050089445438281e-07,
+      "loss": 0.0661,
+      "step": 1125
+    },
+    {
+      "epoch": 1.99645390070922,
+      "grad_norm": 0.2180982083082199,
+      "learning_rate": 5.366726296958854e-07,
+      "loss": 0.072,
+      "step": 1126
+    },
+    {
+      "epoch": 1.99822695035461,
+      "grad_norm": 0.217008575797081,
+      "learning_rate": 2.683363148479427e-07,
+      "loss": 0.0861,
+      "step": 1127
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.11226160824298859,
+      "learning_rate": 0.0,
+      "loss": 0.0346,
+      "step": 1128
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1128,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.726988197801165e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}