diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/last-checkpoint/trainer_state.json"
@@ -0,0 +1,7033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8902737591809481,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0008902737591809482,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002,
+      "loss": 4.9194,
+      "step": 1
+    },
+    {
+      "epoch": 0.0017805475183618963,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002,
+      "loss": 5.9087,
+      "step": 2
+    },
+    {
+      "epoch": 0.0026708212775428445,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002,
+      "loss": 4.4946,
+      "step": 3
+    },
+    {
+      "epoch": 0.0035610950367237926,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002,
+      "loss": 4.8484,
+      "step": 4
+    },
+    {
+      "epoch": 0.00445136879590474,
+      "grad_norm": 2.1236016750335693,
+      "learning_rate": 0.00019999950652018584,
+      "loss": 4.4794,
+      "step": 5
+    },
+    {
+      "epoch": 0.005341642555085689,
+      "grad_norm": NaN,
+      "learning_rate": 0.00019999950652018584,
+      "loss": 4.9055,
+      "step": 6
+    },
+    {
+      "epoch": 0.006231916314266637,
+      "grad_norm": 4.251764297485352,
+      "learning_rate": 0.0001999980260856137,
+      "loss": 5.8693,
+      "step": 7
+    },
+    {
+      "epoch": 0.007122190073447585,
+      "grad_norm": 6.466914653778076,
+      "learning_rate": 0.000199995558710895,
+      "loss": 4.7998,
+      "step": 8
+    },
+    {
+      "epoch": 0.008012463832628533,
+      "grad_norm": 4.194855690002441,
+      "learning_rate": 0.00019999210442038162,
+      "loss": 4.891,
+      "step": 9
+    },
+    {
+      "epoch": 0.00890273759180948,
+      "grad_norm": 4.219061851501465,
+      "learning_rate": 0.00019998766324816607,
+      "loss": 4.1163,
+      "step": 10
+    },
+    {
+      "epoch": 0.00979301135099043,
+      "grad_norm": 5.155824184417725,
+      "learning_rate": 0.0001999822352380809,
+      "loss": 3.4456,
+      "step": 11
+    },
+    {
+      "epoch": 0.010683285110171378,
+      "grad_norm": 3.160855531692505,
+      "learning_rate": 0.00019997582044369843,
+      "loss": 2.9053,
+      "step": 12
+    },
+    {
+      "epoch": 0.011573558869352326,
+      "grad_norm": 2.9109556674957275,
+      "learning_rate": 0.00019996841892833,
+      "loss": 2.3232,
+      "step": 13
+    },
+    {
+      "epoch": 0.012463832628533273,
+      "grad_norm": 3.342146635055542,
+      "learning_rate": 0.00019996003076502565,
+      "loss": 2.391,
+      "step": 14
+    },
+    {
+      "epoch": 0.013354106387714223,
+      "grad_norm": 2.6786203384399414,
+      "learning_rate": 0.00019995065603657316,
+      "loss": 2.292,
+      "step": 15
+    },
+    {
+      "epoch": 0.01424438014689517,
+      "grad_norm": 3.7010445594787598,
+      "learning_rate": 0.0001999402948354973,
+      "loss": 2.2301,
+      "step": 16
+    },
+    {
+      "epoch": 0.015134653906076118,
+      "grad_norm": 4.940849781036377,
+      "learning_rate": 0.00019992894726405893,
+      "loss": 2.3113,
+      "step": 17
+    },
+    {
+      "epoch": 0.016024927665257066,
+      "grad_norm": 4.58949613571167,
+      "learning_rate": 0.000199916613434254,
+      "loss": 2.3711,
+      "step": 18
+    },
+    {
+      "epoch": 0.016915201424438014,
+      "grad_norm": 5.198491096496582,
+      "learning_rate": 0.0001999032934678125,
+      "loss": 2.4678,
+      "step": 19
+    },
+    {
+      "epoch": 0.01780547518361896,
+      "grad_norm": 4.089502334594727,
+      "learning_rate": 0.00019988898749619702,
+      "loss": 2.104,
+      "step": 20
+    },
+    {
+      "epoch": 0.018695748942799913,
+      "grad_norm": 3.2676820755004883,
+      "learning_rate": 0.00019987369566060176,
+      "loss": 2.5646,
+      "step": 21
+    },
+    {
+      "epoch": 0.01958602270198086,
+      "grad_norm": 2.9748589992523193,
+      "learning_rate": 0.00019985741811195097,
+      "loss": 1.7189,
+      "step": 22
+    },
+    {
+      "epoch": 0.020476296461161808,
+      "grad_norm": 3.193584680557251,
+      "learning_rate": 0.00019984015501089752,
+      "loss": 2.7344,
+      "step": 23
+    },
+    {
+      "epoch": 0.021366570220342756,
+      "grad_norm": 2.9254608154296875,
+      "learning_rate": 0.0001998219065278212,
+      "loss": 1.8742,
+      "step": 24
+    },
+    {
+      "epoch": 0.022256843979523704,
+      "grad_norm": 3.604414701461792,
+      "learning_rate": 0.00019980267284282717,
+      "loss": 2.2673,
+      "step": 25
+    },
+    {
+      "epoch": 0.02314711773870465,
+      "grad_norm": 2.6772382259368896,
+      "learning_rate": 0.00019978245414574417,
+      "loss": 1.6669,
+      "step": 26
+    },
+    {
+      "epoch": 0.0240373914978856,
+      "grad_norm": 3.481696128845215,
+      "learning_rate": 0.00019976125063612252,
+      "loss": 2.3554,
+      "step": 27
+    },
+    {
+      "epoch": 0.024927665257066547,
+      "grad_norm": 2.589164972305298,
+      "learning_rate": 0.00019973906252323238,
+      "loss": 1.9393,
+      "step": 28
+    },
+    {
+      "epoch": 0.025817939016247494,
+      "grad_norm": 2.7414450645446777,
+      "learning_rate": 0.0001997158900260614,
+      "loss": 2.2943,
+      "step": 29
+    },
+    {
+      "epoch": 0.026708212775428446,
+      "grad_norm": 2.633164167404175,
+      "learning_rate": 0.0001996917333733128,
+      "loss": 1.9682,
+      "step": 30
+    },
+    {
+      "epoch": 0.027598486534609393,
+      "grad_norm": 2.2975387573242188,
+      "learning_rate": 0.00019966659280340297,
+      "loss": 1.7236,
+      "step": 31
+    },
+    {
+      "epoch": 0.02848876029379034,
+      "grad_norm": 3.209672689437866,
+      "learning_rate": 0.00019964046856445924,
+      "loss": 2.4324,
+      "step": 32
+    },
+    {
+      "epoch": 0.02937903405297129,
+      "grad_norm": 2.241156816482544,
+      "learning_rate": 0.00019961336091431727,
+      "loss": 1.4512,
+      "step": 33
+    },
+    {
+      "epoch": 0.030269307812152237,
+      "grad_norm": 1.827873706817627,
+      "learning_rate": 0.00019958527012051857,
+      "loss": 1.3515,
+      "step": 34
+    },
+    {
+      "epoch": 0.031159581571333184,
+      "grad_norm": 2.4487650394439697,
+      "learning_rate": 0.00019955619646030802,
+      "loss": 1.7576,
+      "step": 35
+    },
+    {
+      "epoch": 0.03204985533051413,
+      "grad_norm": 2.6001129150390625,
+      "learning_rate": 0.00019952614022063084,
+      "loss": 1.5846,
+      "step": 36
+    },
+    {
+      "epoch": 0.03294012908969508,
+      "grad_norm": 2.0996642112731934,
+      "learning_rate": 0.00019949510169813003,
+      "loss": 2.2082,
+      "step": 37
+    },
+    {
+      "epoch": 0.03383040284887603,
+      "grad_norm": 2.8602750301361084,
+      "learning_rate": 0.00019946308119914323,
+      "loss": 2.2053,
+      "step": 38
+    },
+    {
+      "epoch": 0.03472067660805698,
+      "grad_norm": 2.3887569904327393,
+      "learning_rate": 0.0001994300790396999,
+      "loss": 2.0379,
+      "step": 39
+    },
+    {
+      "epoch": 0.03561095036723792,
+      "grad_norm": 3.160905599594116,
+      "learning_rate": 0.000199396095545518,
+      "loss": 2.7873,
+      "step": 40
+    },
+    {
+      "epoch": 0.036501224126418874,
+      "grad_norm": 2.939594268798828,
+      "learning_rate": 0.00019936113105200085,
+      "loss": 2.0382,
+      "step": 41
+    },
+    {
+      "epoch": 0.037391497885599825,
+      "grad_norm": 4.975297451019287,
+      "learning_rate": 0.00019932518590423394,
+      "loss": 1.5076,
+      "step": 42
+    },
+    {
+      "epoch": 0.03828177164478077,
+      "grad_norm": 1.9363937377929688,
+      "learning_rate": 0.00019928826045698136,
+      "loss": 1.8618,
+      "step": 43
+    },
+    {
+      "epoch": 0.03917204540396172,
+      "grad_norm": 2.7423884868621826,
+      "learning_rate": 0.0001992503550746824,
+      "loss": 2.0692,
+      "step": 44
+    },
+    {
+      "epoch": 0.040062319163142665,
+      "grad_norm": 2.3734116554260254,
+      "learning_rate": 0.0001992114701314478,
+      "loss": 2.3358,
+      "step": 45
+    },
+    {
+      "epoch": 0.040952592922323616,
+      "grad_norm": 2.050905227661133,
+      "learning_rate": 0.0001991716060110563,
+      "loss": 1.873,
+      "step": 46
+    },
+    {
+      "epoch": 0.04184286668150456,
+      "grad_norm": 2.0106308460235596,
+      "learning_rate": 0.00019913076310695068,
+      "loss": 2.0759,
+      "step": 47
+    },
+    {
+      "epoch": 0.04273314044068551,
+      "grad_norm": 1.9395461082458496,
+      "learning_rate": 0.00019908894182223388,
+      "loss": 2.0742,
+      "step": 48
+    },
+    {
+      "epoch": 0.043623414199866456,
+      "grad_norm": 2.643402338027954,
+      "learning_rate": 0.00019904614256966512,
+      "loss": 2.5036,
+      "step": 49
+    },
+    {
+      "epoch": 0.04451368795904741,
+      "grad_norm": 2.7979204654693604,
+      "learning_rate": 0.00019900236577165576,
+      "loss": 2.5311,
+      "step": 50
+    },
+    {
+      "epoch": 0.04540396171822836,
+      "grad_norm": 1.904785394668579,
+      "learning_rate": 0.0001989576118602651,
+      "loss": 1.4152,
+      "step": 51
+    },
+    {
+      "epoch": 0.0462942354774093,
+      "grad_norm": 1.4882405996322632,
+      "learning_rate": 0.00019891188127719618,
+      "loss": 1.4164,
+      "step": 52
+    },
+    {
+      "epoch": 0.047184509236590254,
+      "grad_norm": 6.002376556396484,
+      "learning_rate": 0.0001988651744737914,
+      "loss": 1.9128,
+      "step": 53
+    },
+    {
+      "epoch": 0.0480747829957712,
+      "grad_norm": 1.935181975364685,
+      "learning_rate": 0.00019881749191102808,
+      "loss": 1.981,
+      "step": 54
+    },
+    {
+      "epoch": 0.04896505675495215,
+      "grad_norm": 2.011384963989258,
+      "learning_rate": 0.00019876883405951377,
+      "loss": 1.7107,
+      "step": 55
+    },
+    {
+      "epoch": 0.049855330514133094,
+      "grad_norm": 2.44512939453125,
+      "learning_rate": 0.00019871920139948192,
+      "loss": 2.0596,
+      "step": 56
+    },
+    {
+      "epoch": 0.050745604273314045,
+      "grad_norm": 2.8638131618499756,
+      "learning_rate": 0.0001986685944207868,
+      "loss": 2.5629,
+      "step": 57
+    },
+    {
+      "epoch": 0.05163587803249499,
+      "grad_norm": 1.969163179397583,
+      "learning_rate": 0.0001986170136228989,
+      "loss": 1.6412,
+      "step": 58
+    },
+    {
+      "epoch": 0.05252615179167594,
+      "grad_norm": 1.6036492586135864,
+      "learning_rate": 0.00019856445951489982,
+      "loss": 1.957,
+      "step": 59
+    },
+    {
+      "epoch": 0.05341642555085689,
+      "grad_norm": 2.210430860519409,
+      "learning_rate": 0.0001985109326154774,
+      "loss": 1.4708,
+      "step": 60
+    },
+    {
+      "epoch": 0.054306699310037836,
+      "grad_norm": 3.7153899669647217,
+      "learning_rate": 0.00019845643345292054,
+      "loss": 1.7712,
+      "step": 61
+    },
+    {
+      "epoch": 0.05519697306921879,
+      "grad_norm": 3.576857328414917,
+      "learning_rate": 0.00019840096256511398,
+      "loss": 2.0909,
+      "step": 62
+    },
+    {
+      "epoch": 0.05608724682839973,
+      "grad_norm": 3.2540183067321777,
+      "learning_rate": 0.00019834452049953297,
+      "loss": 2.0778,
+      "step": 63
+    },
+    {
+      "epoch": 0.05697752058758068,
+      "grad_norm": 1.8862560987472534,
+      "learning_rate": 0.00019828710781323792,
+      "loss": 1.9815,
+      "step": 64
+    },
+    {
+      "epoch": 0.05786779434676163,
+      "grad_norm": 3.74940824508667,
+      "learning_rate": 0.0001982287250728689,
+      "loss": 1.9903,
+      "step": 65
+    },
+    {
+      "epoch": 0.05875806810594258,
+      "grad_norm": 2.30810546875,
+      "learning_rate": 0.0001981693728546399,
+      "loss": 2.2339,
+      "step": 66
+    },
+    {
+      "epoch": 0.05964834186512353,
+      "grad_norm": 2.387211561203003,
+      "learning_rate": 0.0001981090517443334,
+      "loss": 2.708,
+      "step": 67
+    },
+    {
+      "epoch": 0.06053861562430447,
+      "grad_norm": 2.4890730381011963,
+      "learning_rate": 0.00019804776233729444,
+      "loss": 2.0848,
+      "step": 68
+    },
+    {
+      "epoch": 0.061428889383485424,
+      "grad_norm": 2.170064687728882,
+      "learning_rate": 0.0001979855052384247,
+      "loss": 2.3971,
+      "step": 69
+    },
+    {
+      "epoch": 0.06231916314266637,
+      "grad_norm": 1.8925060033798218,
+      "learning_rate": 0.00019792228106217658,
+      "loss": 1.8057,
+      "step": 70
+    },
+    {
+      "epoch": 0.06320943690184731,
+      "grad_norm": 2.75600528717041,
+      "learning_rate": 0.00019785809043254722,
+      "loss": 1.8895,
+      "step": 71
+    },
+    {
+      "epoch": 0.06409971066102826,
+      "grad_norm": 2.811318874359131,
+      "learning_rate": 0.0001977929339830722,
+      "loss": 2.0633,
+      "step": 72
+    },
+    {
+      "epoch": 0.06498998442020922,
+      "grad_norm": 2.9272348880767822,
+      "learning_rate": 0.00019772681235681936,
+      "loss": 1.9009,
+      "step": 73
+    },
+    {
+      "epoch": 0.06588025817939017,
+      "grad_norm": 3.1955409049987793,
+      "learning_rate": 0.00019765972620638248,
+      "loss": 2.7632,
+      "step": 74
+    },
+    {
+      "epoch": 0.0667705319385711,
+      "grad_norm": 1.8320505619049072,
+      "learning_rate": 0.00019759167619387476,
+      "loss": 2.0807,
+      "step": 75
+    },
+    {
+      "epoch": 0.06766080569775205,
+      "grad_norm": 1.975226640701294,
+      "learning_rate": 0.00019752266299092236,
+      "loss": 2.3015,
+      "step": 76
+    },
+    {
+      "epoch": 0.068551079456933,
+      "grad_norm": 1.833039402961731,
+      "learning_rate": 0.00019745268727865774,
+      "loss": 2.1897,
+      "step": 77
+    },
+    {
+      "epoch": 0.06944135321611396,
+      "grad_norm": 2.358522891998291,
+      "learning_rate": 0.0001973817497477129,
+      "loss": 2.1918,
+      "step": 78
+    },
+    {
+      "epoch": 0.07033162697529491,
+      "grad_norm": 2.0557165145874023,
+      "learning_rate": 0.00019730985109821266,
+      "loss": 1.9941,
+      "step": 79
+    },
+    {
+      "epoch": 0.07122190073447585,
+      "grad_norm": 2.551722288131714,
+      "learning_rate": 0.00019723699203976766,
+      "loss": 2.0574,
+      "step": 80
+    },
+    {
+      "epoch": 0.0721121744936568,
+      "grad_norm": 2.009219169616699,
+      "learning_rate": 0.0001971631732914674,
+      "loss": 2.0421,
+      "step": 81
+    },
+    {
+      "epoch": 0.07300244825283775,
+      "grad_norm": 1.9363999366760254,
+      "learning_rate": 0.0001970883955818731,
+      "loss": 1.8771,
+      "step": 82
+    },
+    {
+      "epoch": 0.0738927220120187,
+      "grad_norm": 2.5710339546203613,
+      "learning_rate": 0.0001970126596490106,
+      "loss": 2.3888,
+      "step": 83
+    },
+    {
+      "epoch": 0.07478299577119965,
+      "grad_norm": 1.6190274953842163,
+      "learning_rate": 0.00019693596624036292,
+      "loss": 2.0216,
+      "step": 84
+    },
+    {
+      "epoch": 0.07567326953038059,
+      "grad_norm": 4.4114580154418945,
+      "learning_rate": 0.0001968583161128631,
+      "loss": 1.7062,
+      "step": 85
+    },
+    {
+      "epoch": 0.07656354328956154,
+      "grad_norm": 2.1439876556396484,
+      "learning_rate": 0.00019677971003288655,
+      "loss": 2.2109,
+      "step": 86
+    },
+    {
+      "epoch": 0.07745381704874249,
+      "grad_norm": 2.3916540145874023,
+      "learning_rate": 0.00019670014877624353,
+      "loss": 2.083,
+      "step": 87
+    },
+    {
+      "epoch": 0.07834409080792344,
+      "grad_norm": 4.150461673736572,
+      "learning_rate": 0.00019661963312817148,
+      "loss": 1.833,
+      "step": 88
+    },
+    {
+      "epoch": 0.07923436456710438,
+      "grad_norm": 2.2267212867736816,
+      "learning_rate": 0.0001965381638833274,
+      "loss": 1.7452,
+      "step": 89
+    },
+    {
+      "epoch": 0.08012463832628533,
+      "grad_norm": 1.6420040130615234,
+      "learning_rate": 0.00019645574184577982,
+      "loss": 1.4853,
+      "step": 90
+    },
+    {
+      "epoch": 0.08101491208546628,
+      "grad_norm": 3.3110744953155518,
+      "learning_rate": 0.000196372367829001,
+      "loss": 1.7166,
+      "step": 91
+    },
+    {
+      "epoch": 0.08190518584464723,
+      "grad_norm": 2.2621548175811768,
+      "learning_rate": 0.00019628804265585877,
+      "loss": 1.9514,
+      "step": 92
+    },
+    {
+      "epoch": 0.08279545960382818,
+      "grad_norm": 1.7131949663162231,
+      "learning_rate": 0.0001962027671586086,
+      "loss": 1.3943,
+      "step": 93
+    },
+    {
+      "epoch": 0.08368573336300912,
+      "grad_norm": 2.4566714763641357,
+      "learning_rate": 0.0001961165421788852,
+      "loss": 2.0299,
+      "step": 94
+    },
+    {
+      "epoch": 0.08457600712219007,
+      "grad_norm": 2.615260362625122,
+      "learning_rate": 0.0001960293685676943,
+      "loss": 2.1021,
+      "step": 95
+    },
+    {
+      "epoch": 0.08546628088137102,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001960293685676943,
+      "loss": 1.8427,
+      "step": 96
+    },
+    {
+      "epoch": 0.08635655464055197,
+      "grad_norm": 2.9707720279693604,
+      "learning_rate": 0.0001959412471854043,
+      "loss": 2.4674,
+      "step": 97
+    },
+    {
+      "epoch": 0.08724682839973291,
+      "grad_norm": 1.8748257160186768,
+      "learning_rate": 0.0001958521789017376,
+      "loss": 1.8866,
+      "step": 98
+    },
+    {
+      "epoch": 0.08813710215891386,
+      "grad_norm": 2.90433406829834,
+      "learning_rate": 0.00019576216459576222,
+      "loss": 2.4039,
+      "step": 99
+    },
+    {
+      "epoch": 0.08902737591809481,
+      "grad_norm": 2.7328593730926514,
+      "learning_rate": 0.00019567120515588308,
+      "loss": 1.9389,
+      "step": 100
+    },
+    {
+      "epoch": 0.08991764967727577,
+      "grad_norm": 2.1492199897766113,
+      "learning_rate": 0.00019557930147983302,
+      "loss": 2.5141,
+      "step": 101
+    },
+    {
+      "epoch": 0.09080792343645672,
+      "grad_norm": 1.7645761966705322,
+      "learning_rate": 0.00019548645447466431,
+      "loss": 1.8018,
+      "step": 102
+    },
+    {
+      "epoch": 0.09169819719563765,
+      "grad_norm": 2.1450297832489014,
+      "learning_rate": 0.00019539266505673938,
+      "loss": 2.2096,
+      "step": 103
+    },
+    {
+      "epoch": 0.0925884709548186,
+      "grad_norm": 2.1536448001861572,
+      "learning_rate": 0.00019529793415172192,
+      "loss": 2.3354,
+      "step": 104
+    },
+    {
+      "epoch": 0.09347874471399956,
+      "grad_norm": 2.578547716140747,
+      "learning_rate": 0.00019520226269456768,
+      "loss": 2.4987,
+      "step": 105
+    },
+    {
+      "epoch": 0.09436901847318051,
+      "grad_norm": 1.8539806604385376,
+      "learning_rate": 0.00019510565162951537,
+      "loss": 1.6678,
+      "step": 106
+    },
+    {
+      "epoch": 0.09525929223236144,
+      "grad_norm": 3.501614570617676,
+      "learning_rate": 0.00019500810191007718,
+      "loss": 2.011,
+      "step": 107
+    },
+    {
+      "epoch": 0.0961495659915424,
+      "grad_norm": 2.3680636882781982,
+      "learning_rate": 0.00019490961449902946,
+      "loss": 2.2364,
+      "step": 108
+    },
+    {
+      "epoch": 0.09703983975072335,
+      "grad_norm": 2.0228307247161865,
+      "learning_rate": 0.0001948101903684032,
+      "loss": 2.1797,
+      "step": 109
+    },
+    {
+      "epoch": 0.0979301135099043,
+      "grad_norm": 1.9101430177688599,
+      "learning_rate": 0.00019470983049947444,
+      "loss": 1.9929,
+      "step": 110
+    },
+    {
+      "epoch": 0.09882038726908525,
+      "grad_norm": 1.478506088256836,
+      "learning_rate": 0.00019460853588275454,
+      "loss": 1.5515,
+      "step": 111
+    },
+    {
+      "epoch": 0.09971066102826619,
+      "grad_norm": 2.007383346557617,
+      "learning_rate": 0.00019450630751798048,
+      "loss": 2.2393,
+      "step": 112
+    },
+    {
+      "epoch": 0.10060093478744714,
+      "grad_norm": 1.8135859966278076,
+      "learning_rate": 0.000194403146414105,
+      "loss": 1.9253,
+      "step": 113
+    },
+    {
+      "epoch": 0.10149120854662809,
+      "grad_norm": 2.946672201156616,
+      "learning_rate": 0.00019429905358928646,
+      "loss": 2.9314,
+      "step": 114
+    },
+    {
+      "epoch": 0.10238148230580904,
+      "grad_norm": 2.189742088317871,
+      "learning_rate": 0.00019419403007087907,
+      "loss": 1.7929,
+      "step": 115
+    },
+    {
+      "epoch": 0.10327175606498998,
+      "grad_norm": 1.661193609237671,
+      "learning_rate": 0.00019408807689542257,
+      "loss": 1.737,
+      "step": 116
+    },
+    {
+      "epoch": 0.10416202982417093,
+      "grad_norm": 2.0452182292938232,
+      "learning_rate": 0.00019398119510863197,
+      "loss": 2.2965,
+      "step": 117
+    },
+    {
+      "epoch": 0.10505230358335188,
+      "grad_norm": 2.4580020904541016,
+      "learning_rate": 0.00019387338576538744,
+      "loss": 3.0578,
+      "step": 118
+    },
+    {
+      "epoch": 0.10594257734253283,
+      "grad_norm": 1.7867645025253296,
+      "learning_rate": 0.00019376464992972356,
+      "loss": 1.6043,
+      "step": 119
+    },
+    {
+      "epoch": 0.10683285110171378,
+      "grad_norm": 2.074819326400757,
+      "learning_rate": 0.00019365498867481923,
+      "loss": 1.9777,
+      "step": 120
+    },
+    {
+      "epoch": 0.10772312486089472,
+      "grad_norm": 1.3542097806930542,
+      "learning_rate": 0.00019354440308298675,
+      "loss": 1.4176,
+      "step": 121
+    },
+    {
+      "epoch": 0.10861339862007567,
+      "grad_norm": 1.3362765312194824,
+      "learning_rate": 0.00019343289424566122,
+      "loss": 1.2639,
+      "step": 122
+    },
+    {
+      "epoch": 0.10950367237925662,
+      "grad_norm": 2.469452142715454,
+      "learning_rate": 0.00019332046326338986,
+      "loss": 2.6764,
+      "step": 123
+    },
+    {
+      "epoch": 0.11039394613843757,
+      "grad_norm": 2.0411577224731445,
+      "learning_rate": 0.0001932071112458211,
+      "loss": 2.5306,
+      "step": 124
+    },
+    {
+      "epoch": 0.11128421989761851,
+      "grad_norm": 1.3371831178665161,
+      "learning_rate": 0.00019309283931169356,
+      "loss": 1.647,
+      "step": 125
+    },
+    {
+      "epoch": 0.11217449365679946,
+      "grad_norm": 2.3767588138580322,
+      "learning_rate": 0.00019297764858882514,
+      "loss": 2.1805,
+      "step": 126
+    },
+    {
+      "epoch": 0.11306476741598041,
+      "grad_norm": 1.9597545862197876,
+      "learning_rate": 0.00019286154021410173,
+      "loss": 2.5409,
+      "step": 127
+    },
+    {
+      "epoch": 0.11395504117516136,
+      "grad_norm": 2.3952715396881104,
+      "learning_rate": 0.00019274451533346615,
+      "loss": 2.2106,
+      "step": 128
+    },
+    {
+      "epoch": 0.11484531493434232,
+      "grad_norm": 2.271564245223999,
+      "learning_rate": 0.00019262657510190666,
+      "loss": 1.4799,
+      "step": 129
+    },
+    {
+      "epoch": 0.11573558869352325,
+      "grad_norm": 2.6225390434265137,
+      "learning_rate": 0.0001925077206834458,
+      "loss": 2.0251,
+      "step": 130
+    },
+    {
+      "epoch": 0.1166258624527042,
+      "grad_norm": 1.9409428834915161,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 2.1348,
+      "step": 131
+    },
+    {
+      "epoch": 0.11751613621188516,
+      "grad_norm": 3.0642130374908447,
+      "learning_rate": 0.0001922672739870115,
+      "loss": 2.6375,
+      "step": 132
+    },
+    {
+      "epoch": 0.1184064099710661,
+      "grad_norm": 2.3774688243865967,
+      "learning_rate": 0.00019214568408214985,
+      "loss": 2.6276,
+      "step": 133
+    },
+    {
+      "epoch": 0.11929668373024706,
+      "grad_norm": 1.9567856788635254,
+      "learning_rate": 0.00019202318473658705,
+      "loss": 2.6106,
+      "step": 134
+    },
+    {
+      "epoch": 0.120186957489428,
+      "grad_norm": 2.2974131107330322,
+      "learning_rate": 0.00019189977715934213,
+      "loss": 2.4285,
+      "step": 135
+    },
+    {
+      "epoch": 0.12107723124860895,
+      "grad_norm": 3.681556463241577,
+      "learning_rate": 0.00019177546256839812,
+      "loss": 2.104,
+      "step": 136
+    },
+    {
+      "epoch": 0.1219675050077899,
+      "grad_norm": 6.61801290512085,
+      "learning_rate": 0.0001916502421906898,
+      "loss": 1.8274,
+      "step": 137
+    },
+    {
+      "epoch": 0.12285777876697085,
+      "grad_norm": 2.095499277114868,
+      "learning_rate": 0.00019152411726209176,
+      "loss": 1.7257,
+      "step": 138
+    },
+    {
+      "epoch": 0.12374805252615179,
+      "grad_norm": 1.8911921977996826,
+      "learning_rate": 0.00019139708902740613,
+      "loss": 1.907,
+      "step": 139
+    },
+    {
+      "epoch": 0.12463832628533274,
+      "grad_norm": 1.6587978601455688,
+      "learning_rate": 0.0001912691587403503,
+      "loss": 1.6967,
+      "step": 140
+    },
+    {
+      "epoch": 0.1255286000445137,
+      "grad_norm": 2.2344250679016113,
+      "learning_rate": 0.00019114032766354453,
+      "loss": 1.841,
+      "step": 141
+    },
+    {
+      "epoch": 0.12641887380369463,
+      "grad_norm": 2.4435856342315674,
+      "learning_rate": 0.00019101059706849957,
+      "loss": 2.4631,
+      "step": 142
+    },
+    {
+      "epoch": 0.1273091475628756,
+      "grad_norm": 2.551354169845581,
+      "learning_rate": 0.00019087996823560402,
+      "loss": 2.413,
+      "step": 143
+    },
+    {
+      "epoch": 0.12819942132205653,
+      "grad_norm": 2.87556791305542,
+      "learning_rate": 0.0001907484424541117,
+      "loss": 1.9834,
+      "step": 144
+    },
+    {
+      "epoch": 0.1290896950812375,
+      "grad_norm": 2.5196003913879395,
+      "learning_rate": 0.00019061602102212898,
+      "loss": 2.9326,
+      "step": 145
+    },
+    {
+      "epoch": 0.12997996884041843,
+      "grad_norm": 2.4851393699645996,
+      "learning_rate": 0.00019048270524660196,
+      "loss": 1.8773,
+      "step": 146
+    },
+    {
+      "epoch": 0.13087024259959937,
+      "grad_norm": 2.0737171173095703,
+      "learning_rate": 0.0001903484964433035,
+      "loss": 1.9065,
+      "step": 147
+    },
+    {
+      "epoch": 0.13176051635878033,
+      "grad_norm": 2.818920612335205,
+      "learning_rate": 0.00019021339593682028,
+      "loss": 2.1904,
+      "step": 148
+    },
+    {
+      "epoch": 0.13265079011796127,
+      "grad_norm": 1.77272629737854,
+      "learning_rate": 0.00019007740506053983,
+      "loss": 1.9956,
+      "step": 149
+    },
+    {
+      "epoch": 0.1335410638771422,
+      "grad_norm": 1.917798638343811,
+      "learning_rate": 0.0001899405251566371,
+      "loss": 1.9516,
+      "step": 150
+    },
+    {
+      "epoch": 0.13443133763632317,
+      "grad_norm": 1.5387191772460938,
+      "learning_rate": 0.00018980275757606157,
+      "loss": 1.2293,
+      "step": 151
+    },
+    {
+      "epoch": 0.1353216113955041,
+      "grad_norm": 2.3466856479644775,
+      "learning_rate": 0.00018966410367852362,
+      "loss": 2.0018,
+      "step": 152
+    },
+    {
+      "epoch": 0.13621188515468508,
+      "grad_norm": 2.354393243789673,
+      "learning_rate": 0.00018952456483248119,
+      "loss": 1.9607,
+      "step": 153
+    },
+    {
+      "epoch": 0.137102158913866,
+      "grad_norm": 3.2391810417175293,
+      "learning_rate": 0.0001893841424151264,
+      "loss": 1.6369,
+      "step": 154
+    },
+    {
+      "epoch": 0.13799243267304695,
+      "grad_norm": 1.563175082206726,
+      "learning_rate": 0.0001892428378123718,
+      "loss": 1.7831,
+      "step": 155
+    },
+    {
+      "epoch": 0.13888270643222791,
+      "grad_norm": 2.1940622329711914,
+      "learning_rate": 0.0001891006524188368,
+      "loss": 2.1351,
+      "step": 156
+    },
+    {
+      "epoch": 0.13977298019140885,
+      "grad_norm": 2.7515769004821777,
+      "learning_rate": 0.00018895758763783383,
+      "loss": 2.5552,
+      "step": 157
+    },
+    {
+      "epoch": 0.14066325395058982,
+      "grad_norm": 2.5042002201080322,
+      "learning_rate": 0.00018881364488135448,
+      "loss": 1.6571,
+      "step": 158
+    },
+    {
+      "epoch": 0.14155352770977075,
+      "grad_norm": 1.2943757772445679,
+      "learning_rate": 0.00018866882557005567,
+      "loss": 1.3131,
+      "step": 159
+    },
+    {
+      "epoch": 0.1424438014689517,
+      "grad_norm": 1.860222339630127,
+      "learning_rate": 0.00018852313113324552,
+      "loss": 2.0416,
+      "step": 160
+    },
+    {
+      "epoch": 0.14333407522813266,
+      "grad_norm": 1.4477007389068604,
+      "learning_rate": 0.00018837656300886937,
+      "loss": 1.4943,
+      "step": 161
+    },
+    {
+      "epoch": 0.1442243489873136,
+      "grad_norm": 2.673553705215454,
+      "learning_rate": 0.00018822912264349534,
+      "loss": 2.4297,
+      "step": 162
+    },
+    {
+      "epoch": 0.14511462274649456,
+      "grad_norm": 2.126965284347534,
+      "learning_rate": 0.00018808081149230036,
+      "loss": 1.9287,
+      "step": 163
+    },
+    {
+      "epoch": 0.1460048965056755,
+      "grad_norm": 1.6320929527282715,
+      "learning_rate": 0.00018793163101905563,
+      "loss": 1.8308,
+      "step": 164
+    },
+    {
+      "epoch": 0.14689517026485643,
+      "grad_norm": 2.4692060947418213,
+      "learning_rate": 0.00018778158269611218,
+      "loss": 2.6697,
+      "step": 165
+    },
+    {
+      "epoch": 0.1477854440240374,
+      "grad_norm": 1.8720828294754028,
+      "learning_rate": 0.00018763066800438636,
+      "loss": 2.1752,
+      "step": 166
+    },
+    {
+      "epoch": 0.14867571778321834,
+      "grad_norm": 1.8736835718154907,
+      "learning_rate": 0.0001874788884333453,
+      "loss": 1.821,
+      "step": 167
+    },
+    {
+      "epoch": 0.1495659915423993,
+      "grad_norm": 1.8159123659133911,
+      "learning_rate": 0.00018732624548099204,
+      "loss": 1.7778,
+      "step": 168
+    },
+    {
+      "epoch": 0.15045626530158024,
+      "grad_norm": 3.0733249187469482,
+      "learning_rate": 0.0001871727406538509,
+      "loss": 2.8908,
+      "step": 169
+    },
+    {
+      "epoch": 0.15134653906076118,
+      "grad_norm": 1.6238402128219604,
+      "learning_rate": 0.0001870183754669526,
+      "loss": 1.6326,
+      "step": 170
+    },
+    {
+      "epoch": 0.15223681281994214,
+      "grad_norm": 2.544752597808838,
+      "learning_rate": 0.00018686315144381913,
+      "loss": 2.0834,
+      "step": 171
+    },
+    {
+      "epoch": 0.15312708657912308,
+      "grad_norm": 2.253525972366333,
+      "learning_rate": 0.000186707070116449,
+      "loss": 2.3574,
+      "step": 172
+    },
+    {
+      "epoch": 0.15401736033830402,
+      "grad_norm": 2.5005857944488525,
+      "learning_rate": 0.0001865501330253019,
+      "loss": 2.3706,
+      "step": 173
+    },
+    {
+      "epoch": 0.15490763409748498,
+      "grad_norm": 2.5505003929138184,
+      "learning_rate": 0.00018639234171928353,
+      "loss": 2.6693,
+      "step": 174
+    },
+    {
+      "epoch": 0.15579790785666592,
+      "grad_norm": 2.0035719871520996,
+      "learning_rate": 0.0001862336977557304,
+      "loss": 1.8442,
+      "step": 175
+    },
+    {
+      "epoch": 0.15668818161584688,
+      "grad_norm": 1.8089210987091064,
+      "learning_rate": 0.0001860742027003944,
+      "loss": 2.1121,
+      "step": 176
+    },
+    {
+      "epoch": 0.15757845537502782,
+      "grad_norm": 1.3150620460510254,
+      "learning_rate": 0.00018591385812742725,
+      "loss": 1.6413,
+      "step": 177
+    },
+    {
+      "epoch": 0.15846872913420876,
+      "grad_norm": 2.3389761447906494,
+      "learning_rate": 0.00018575266561936523,
+      "loss": 2.561,
+      "step": 178
+    },
+    {
+      "epoch": 0.15935900289338972,
+      "grad_norm": 2.1656057834625244,
+      "learning_rate": 0.00018559062676711332,
+      "loss": 2.0227,
+      "step": 179
+    },
+    {
+      "epoch": 0.16024927665257066,
+      "grad_norm": 1.8123457431793213,
+      "learning_rate": 0.0001854277431699295,
+      "loss": 1.8462,
+      "step": 180
+    },
+    {
+      "epoch": 0.16113955041175163,
+      "grad_norm": 2.1559298038482666,
+      "learning_rate": 0.00018526401643540922,
+      "loss": 2.5308,
+      "step": 181
+    },
+    {
+      "epoch": 0.16202982417093256,
+      "grad_norm": 1.6904340982437134,
+      "learning_rate": 0.00018509944817946922,
+      "loss": 1.6456,
+      "step": 182
+    },
+    {
+      "epoch": 0.1629200979301135,
+      "grad_norm": 1.3393988609313965,
+      "learning_rate": 0.00018493404002633166,
+      "loss": 1.4475,
+      "step": 183
+    },
+    {
+      "epoch": 0.16381037168929447,
+      "grad_norm": 2.157855749130249,
+      "learning_rate": 0.00018476779360850832,
+      "loss": 2.446,
+      "step": 184
+    },
+    {
+      "epoch": 0.1647006454484754,
+      "grad_norm": 2.064180612564087,
+      "learning_rate": 0.00018460071056678422,
+      "loss": 1.7074,
+      "step": 185
+    },
+    {
+      "epoch": 0.16559091920765637,
+      "grad_norm": 1.8045761585235596,
+      "learning_rate": 0.00018443279255020152,
+      "loss": 2.1376,
+      "step": 186
+    },
+    {
+      "epoch": 0.1664811929668373,
+      "grad_norm": 1.4348753690719604,
+      "learning_rate": 0.00018426404121604323,
+      "loss": 1.889,
+      "step": 187
+    },
+    {
+      "epoch": 0.16737146672601824,
+      "grad_norm": 2.1404173374176025,
+      "learning_rate": 0.00018409445822981693,
+      "loss": 2.1494,
+      "step": 188
+    },
+    {
+      "epoch": 0.1682617404851992,
+      "grad_norm": 2.041149139404297,
+      "learning_rate": 0.00018392404526523817,
+      "loss": 2.061,
+      "step": 189
+    },
+    {
+      "epoch": 0.16915201424438014,
+      "grad_norm": 1.5126264095306396,
+      "learning_rate": 0.0001837528040042142,
+      "loss": 2.1654,
+      "step": 190
+    },
+    {
+      "epoch": 0.17004228800356108,
+      "grad_norm": 2.7885916233062744,
+      "learning_rate": 0.00018358073613682706,
+      "loss": 2.3817,
+      "step": 191
+    },
+    {
+      "epoch": 0.17093256176274205,
+      "grad_norm": 1.7389510869979858,
+      "learning_rate": 0.00018340784336131713,
+      "loss": 1.721,
+      "step": 192
+    },
+    {
+      "epoch": 0.17182283552192298,
+      "grad_norm": 1.8442341089248657,
+      "learning_rate": 0.00018323412738406635,
+      "loss": 1.8906,
+      "step": 193
+    },
+    {
+      "epoch": 0.17271310928110395,
+      "grad_norm": 1.5204037427902222,
+      "learning_rate": 0.00018305958991958127,
+      "loss": 1.4487,
+      "step": 194
+    },
+    {
+      "epoch": 0.1736033830402849,
+      "grad_norm": 1.7395809888839722,
+      "learning_rate": 0.0001828842326904762,
+      "loss": 1.8568,
+      "step": 195
+    },
+    {
+      "epoch": 0.17449365679946582,
+      "grad_norm": 2.0284206867218018,
+      "learning_rate": 0.00018270805742745617,
+      "loss": 2.1218,
+      "step": 196
+    },
+    {
+      "epoch": 0.1753839305586468,
+      "grad_norm": 1.6896398067474365,
+      "learning_rate": 0.00018253106586929997,
+      "loss": 1.6777,
+      "step": 197
+    },
+    {
+      "epoch": 0.17627420431782773,
+      "grad_norm": 2.2234766483306885,
+      "learning_rate": 0.00018235325976284275,
+      "loss": 1.9342,
+      "step": 198
+    },
+    {
+      "epoch": 0.1771644780770087,
+      "grad_norm": 1.6226762533187866,
+      "learning_rate": 0.00018217464086295904,
+      "loss": 1.4723,
+      "step": 199
+    },
+    {
+      "epoch": 0.17805475183618963,
+      "grad_norm": 1.9177464246749878,
+      "learning_rate": 0.00018199521093254523,
+      "loss": 1.4232,
+      "step": 200
+    },
+    {
+      "epoch": 0.17894502559537057,
+      "grad_norm": 2.020766019821167,
+      "learning_rate": 0.00018181497174250236,
+      "loss": 1.5295,
+      "step": 201
+    },
+    {
+      "epoch": 0.17983529935455153,
+      "grad_norm": 2.385566473007202,
+      "learning_rate": 0.00018163392507171842,
+      "loss": 2.0807,
+      "step": 202
+    },
+    {
+      "epoch": 0.18072557311373247,
+      "grad_norm": 1.6531530618667603,
+      "learning_rate": 0.00018145207270705096,
+      "loss": 1.8632,
+      "step": 203
+    },
+    {
+      "epoch": 0.18161584687291343,
+      "grad_norm": 1.9366607666015625,
+      "learning_rate": 0.0001812694164433094,
+      "loss": 1.8869,
+      "step": 204
+    },
+    {
+      "epoch": 0.18250612063209437,
+      "grad_norm": 2.225327253341675,
+      "learning_rate": 0.00018108595808323736,
+      "loss": 2.2026,
+      "step": 205
+    },
+    {
+      "epoch": 0.1833963943912753,
+      "grad_norm": 1.5953576564788818,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 1.6126,
+      "step": 206
+    },
+    {
+      "epoch": 0.18428666815045627,
+      "grad_norm": 1.2869915962219238,
+      "learning_rate": 0.00018071664232464002,
+      "loss": 1.2606,
+      "step": 207
+    },
+    {
+      "epoch": 0.1851769419096372,
+      "grad_norm": 1.9665155410766602,
+      "learning_rate": 0.0001805307885711122,
+      "loss": 1.5137,
+      "step": 208
+    },
+    {
+      "epoch": 0.18606721566881815,
+      "grad_norm": 1.9578527212142944,
+      "learning_rate": 0.00018034414001121278,
+      "loss": 1.8762,
+      "step": 209
+    },
+    {
+      "epoch": 0.1869574894279991,
+      "grad_norm": 2.5774004459381104,
+      "learning_rate": 0.00018015669848708767,
+      "loss": 2.44,
+      "step": 210
+    },
+    {
+      "epoch": 0.18784776318718005,
+      "grad_norm": 1.7559341192245483,
+      "learning_rate": 0.00017996846584870908,
+      "loss": 1.869,
+      "step": 211
+    },
+    {
+      "epoch": 0.18873803694636102,
+      "grad_norm": 1.6659877300262451,
+      "learning_rate": 0.0001797794439538571,
+      "loss": 1.7385,
+      "step": 212
+    },
+    {
+      "epoch": 0.18962831070554195,
+      "grad_norm": 1.969826102256775,
+      "learning_rate": 0.0001795896346681016,
+      "loss": 1.9256,
+      "step": 213
+    },
+    {
+      "epoch": 0.1905185844647229,
+      "grad_norm": 2.6088500022888184,
+      "learning_rate": 0.00017939903986478355,
+      "loss": 2.6625,
+      "step": 214
+    },
+    {
+      "epoch": 0.19140885822390385,
+      "grad_norm": 1.7767618894577026,
+      "learning_rate": 0.00017920766142499672,
+      "loss": 1.4711,
+      "step": 215
+    },
+    {
+      "epoch": 0.1922991319830848,
+      "grad_norm": 2.5906081199645996,
+      "learning_rate": 0.00017901550123756906,
+      "loss": 2.676,
+      "step": 216
+    },
+    {
+      "epoch": 0.19318940574226576,
+      "grad_norm": 1.8418434858322144,
+      "learning_rate": 0.00017882256119904403,
+      "loss": 1.8141,
+      "step": 217
+    },
+    {
+      "epoch": 0.1940796795014467,
+      "grad_norm": 2.3214592933654785,
+      "learning_rate": 0.00017862884321366188,
+      "loss": 1.6772,
+      "step": 218
+    },
+    {
+      "epoch": 0.19496995326062763,
+      "grad_norm": 2.0468828678131104,
+      "learning_rate": 0.000178434349193341,
+      "loss": 2.4774,
+      "step": 219
+    },
+    {
+      "epoch": 0.1958602270198086,
+      "grad_norm": 2.031306743621826,
+      "learning_rate": 0.0001782390810576588,
+      "loss": 1.6958,
+      "step": 220
+    },
+    {
+      "epoch": 0.19675050077898953,
+      "grad_norm": 2.217170476913452,
+      "learning_rate": 0.000178043040733833,
+      "loss": 1.8302,
+      "step": 221
+    },
+    {
+      "epoch": 0.1976407745381705,
+      "grad_norm": 2.470906972885132,
+      "learning_rate": 0.00017784623015670238,
+      "loss": 2.2488,
+      "step": 222
+    },
+    {
+      "epoch": 0.19853104829735144,
+      "grad_norm": 2.1658363342285156,
+      "learning_rate": 0.00017764865126870786,
+      "loss": 1.5873,
+      "step": 223
+    },
+    {
+      "epoch": 0.19942132205653237,
+      "grad_norm": 2.0716309547424316,
+      "learning_rate": 0.00017745030601987337,
+      "loss": 1.815,
+      "step": 224
+    },
+    {
+      "epoch": 0.20031159581571334,
+      "grad_norm": 1.9759398698806763,
+      "learning_rate": 0.00017725119636778644,
+      "loss": 2.1319,
+      "step": 225
+    },
+    {
+      "epoch": 0.20120186957489428,
+      "grad_norm": 1.9106535911560059,
+      "learning_rate": 0.00017705132427757895,
+      "loss": 1.5344,
+      "step": 226
+    },
+    {
+      "epoch": 0.20209214333407524,
+      "grad_norm": 1.441969633102417,
+      "learning_rate": 0.00017685069172190766,
+      "loss": 1.6145,
+      "step": 227
+    },
+    {
+      "epoch": 0.20298241709325618,
+      "grad_norm": 2.0080626010894775,
+      "learning_rate": 0.00017664930068093498,
+      "loss": 2.3943,
+      "step": 228
+    },
+    {
+      "epoch": 0.20387269085243712,
+      "grad_norm": 1.463238000869751,
+      "learning_rate": 0.00017644715314230918,
+      "loss": 1.6539,
+      "step": 229
+    },
+    {
+      "epoch": 0.20476296461161808,
+      "grad_norm": 1.6935185194015503,
+      "learning_rate": 0.0001762442511011448,
+      "loss": 1.893,
+      "step": 230
+    },
+    {
+      "epoch": 0.20565323837079902,
+      "grad_norm": 2.5301573276519775,
+      "learning_rate": 0.0001760405965600031,
+      "loss": 2.2857,
+      "step": 231
+    },
+    {
+      "epoch": 0.20654351212997996,
+      "grad_norm": 1.8854233026504517,
+      "learning_rate": 0.0001758361915288722,
+      "loss": 2.3479,
+      "step": 232
+    },
+    {
+      "epoch": 0.20743378588916092,
+      "grad_norm": 2.3383989334106445,
+      "learning_rate": 0.0001756310380251472,
+      "loss": 2.4742,
+      "step": 233
+    },
+    {
+      "epoch": 0.20832405964834186,
+      "grad_norm": 1.7122273445129395,
+      "learning_rate": 0.00017542513807361037,
+      "loss": 1.4847,
+      "step": 234
+    },
+    {
+      "epoch": 0.20921433340752282,
+      "grad_norm": 1.4948415756225586,
+      "learning_rate": 0.00017521849370641114,
+      "loss": 2.002,
+      "step": 235
+    },
+    {
+      "epoch": 0.21010460716670376,
+      "grad_norm": 1.8573172092437744,
+      "learning_rate": 0.00017501110696304596,
+      "loss": 1.5925,
+      "step": 236
+    },
+    {
+      "epoch": 0.2109948809258847,
+      "grad_norm": 2.2749545574188232,
+      "learning_rate": 0.00017480297989033825,
+      "loss": 1.9138,
+      "step": 237
+    },
+    {
+      "epoch": 0.21188515468506566,
+      "grad_norm": 1.9408509731292725,
+      "learning_rate": 0.00017459411454241822,
+      "loss": 1.8457,
+      "step": 238
+    },
+    {
+      "epoch": 0.2127754284442466,
+      "grad_norm": 1.870306134223938,
+      "learning_rate": 0.00017438451298070252,
+      "loss": 1.6819,
+      "step": 239
+    },
+    {
+      "epoch": 0.21366570220342757,
+      "grad_norm": 1.6488808393478394,
+      "learning_rate": 0.00017417417727387394,
+      "loss": 1.7439,
+      "step": 240
+    },
+    {
+      "epoch": 0.2145559759626085,
+      "grad_norm": 1.684953212738037,
+      "learning_rate": 0.000173963109497861,
+      "loss": 1.7115,
+      "step": 241
+    },
+    {
+      "epoch": 0.21544624972178944,
+      "grad_norm": 1.9300771951675415,
+      "learning_rate": 0.0001737513117358174,
+      "loss": 1.7342,
+      "step": 242
+    },
+    {
+      "epoch": 0.2163365234809704,
+      "grad_norm": 2.416626214981079,
+      "learning_rate": 0.0001735387860781016,
+      "loss": 2.1706,
+      "step": 243
+    },
+    {
+      "epoch": 0.21722679724015134,
+      "grad_norm": 3.1392476558685303,
+      "learning_rate": 0.00017332553462225602,
+      "loss": 1.9876,
+      "step": 244
+    },
+    {
+      "epoch": 0.2181170709993323,
+      "grad_norm": 3.1887779235839844,
+      "learning_rate": 0.00017311155947298643,
+      "loss": 1.8074,
+      "step": 245
+    },
+    {
+      "epoch": 0.21900734475851324,
+      "grad_norm": 2.362192153930664,
+      "learning_rate": 0.00017289686274214118,
+      "loss": 2.3624,
+      "step": 246
+    },
+    {
+      "epoch": 0.21989761851769418,
+      "grad_norm": 2.7475128173828125,
+      "learning_rate": 0.0001726814465486903,
+      "loss": 2.0255,
+      "step": 247
+    },
+    {
+      "epoch": 0.22078789227687515,
+      "grad_norm": 2.4550669193267822,
+      "learning_rate": 0.0001724653130187047,
+      "loss": 1.9455,
+      "step": 248
+    },
+    {
+      "epoch": 0.22167816603605608,
+      "grad_norm": 3.648386001586914,
+      "learning_rate": 0.00017224846428533499,
+      "loss": 2.045,
+      "step": 249
+    },
+    {
+      "epoch": 0.22256843979523702,
+      "grad_norm": 2.9079108238220215,
+      "learning_rate": 0.0001720309024887907,
+      "loss": 1.7025,
+      "step": 250
+    },
+    {
+      "epoch": 0.223458713554418,
+      "grad_norm": 2.9583306312561035,
+      "learning_rate": 0.00017181262977631888,
+      "loss": 1.6872,
+      "step": 251
+    },
+    {
+      "epoch": 0.22434898731359892,
+      "grad_norm": 2.069707155227661,
+      "learning_rate": 0.00017159364830218312,
+      "loss": 2.0011,
+      "step": 252
+    },
+    {
+      "epoch": 0.2252392610727799,
+      "grad_norm": 2.955867290496826,
+      "learning_rate": 0.00017137396022764214,
+      "loss": 1.8701,
+      "step": 253
+    },
+    {
+      "epoch": 0.22612953483196083,
+      "grad_norm": 2.031825065612793,
+      "learning_rate": 0.00017115356772092857,
+      "loss": 1.8168,
+      "step": 254
+    },
+    {
+      "epoch": 0.22701980859114176,
+      "grad_norm": 2.6691067218780518,
+      "learning_rate": 0.0001709324729572274,
+      "loss": 2.2007,
+      "step": 255
+    },
+    {
+      "epoch": 0.22791008235032273,
+      "grad_norm": 2.2941246032714844,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 2.6734,
+      "step": 256
+    },
+    {
+      "epoch": 0.22880035610950367,
+      "grad_norm": 4.628593444824219,
+      "learning_rate": 0.00017048818539423615,
+      "loss": 2.1296,
+      "step": 257
+    },
+    {
+      "epoch": 0.22969062986868463,
+      "grad_norm": 1.7624280452728271,
+      "learning_rate": 0.00017026499697988493,
+      "loss": 2.0804,
+      "step": 258
+    },
+    {
+      "epoch": 0.23058090362786557,
+      "grad_norm": 4.2342705726623535,
+      "learning_rate": 0.00017004111507838064,
+      "loss": 1.8746,
+      "step": 259
+    },
+    {
+      "epoch": 0.2314711773870465,
+      "grad_norm": 5.327749252319336,
+      "learning_rate": 0.00016981654189934727,
+      "loss": 2.1781,
+      "step": 260
+    },
+    {
+      "epoch": 0.23236145114622747,
+      "grad_norm": 2.6609578132629395,
+      "learning_rate": 0.00016959127965923142,
+      "loss": 1.7924,
+      "step": 261
+    },
+    {
+      "epoch": 0.2332517249054084,
+      "grad_norm": 1.2489802837371826,
+      "learning_rate": 0.0001693653305812805,
+      "loss": 1.4193,
+      "step": 262
+    },
+    {
+      "epoch": 0.23414199866458937,
+      "grad_norm": 2.315696954727173,
+      "learning_rate": 0.00016913869689552064,
+      "loss": 2.1331,
+      "step": 263
+    },
+    {
+      "epoch": 0.2350322724237703,
+      "grad_norm": 2.0057482719421387,
+      "learning_rate": 0.00016891138083873487,
+      "loss": 1.317,
+      "step": 264
+    },
+    {
+      "epoch": 0.23592254618295125,
+      "grad_norm": 1.934035301208496,
+      "learning_rate": 0.00016868338465444085,
+      "loss": 2.0334,
+      "step": 265
+    },
+    {
+      "epoch": 0.2368128199421322,
+      "grad_norm": 1.3984287977218628,
+      "learning_rate": 0.00016845471059286887,
+      "loss": 1.3556,
+      "step": 266
+    },
+    {
+      "epoch": 0.23770309370131315,
+      "grad_norm": 3.7118077278137207,
+      "learning_rate": 0.00016822536091093965,
+      "loss": 2.4466,
+      "step": 267
+    },
+    {
+      "epoch": 0.23859336746049412,
+      "grad_norm": 3.144308090209961,
+      "learning_rate": 0.00016799533787224192,
+      "loss": 2.0156,
+      "step": 268
+    },
+    {
+      "epoch": 0.23948364121967505,
+      "grad_norm": 1.601457118988037,
+      "learning_rate": 0.00016776464374701025,
+      "loss": 1.6662,
+      "step": 269
+    },
+    {
+      "epoch": 0.240373914978856,
+      "grad_norm": 2.1832356452941895,
+      "learning_rate": 0.00016753328081210245,
+      "loss": 1.8462,
+      "step": 270
+    },
+    {
+      "epoch": 0.24126418873803696,
+      "grad_norm": 2.149434804916382,
+      "learning_rate": 0.00016730125135097735,
+      "loss": 2.2022,
+      "step": 271
+    },
+    {
+      "epoch": 0.2421544624972179,
+      "grad_norm": 2.7085752487182617,
+      "learning_rate": 0.000167068557653672,
+      "loss": 1.306,
+      "step": 272
+    },
+    {
+      "epoch": 0.24304473625639883,
+      "grad_norm": 2.049161434173584,
+      "learning_rate": 0.0001668352020167793,
+      "loss": 2.1773,
+      "step": 273
+    },
+    {
+      "epoch": 0.2439350100155798,
+      "grad_norm": 1.8990561962127686,
+      "learning_rate": 0.00016660118674342517,
+      "loss": 2.1278,
+      "step": 274
+    },
+    {
+      "epoch": 0.24482528377476073,
+      "grad_norm": 1.579715371131897,
+      "learning_rate": 0.00016636651414324587,
+      "loss": 1.5829,
+      "step": 275
+    },
+    {
+      "epoch": 0.2457155575339417,
+      "grad_norm": 2.2845969200134277,
+      "learning_rate": 0.00016613118653236518,
+      "loss": 1.7622,
+      "step": 276
+    },
+    {
+      "epoch": 0.24660583129312263,
+      "grad_norm": 1.9798493385314941,
+      "learning_rate": 0.0001658952062333717,
+      "loss": 2.2187,
+      "step": 277
+    },
+    {
+      "epoch": 0.24749610505230357,
+      "grad_norm": 2.739555835723877,
+      "learning_rate": 0.00016565857557529566,
+      "loss": 1.7902,
+      "step": 278
+    },
+    {
+      "epoch": 0.24838637881148454,
+      "grad_norm": 3.5172815322875977,
+      "learning_rate": 0.00016542129689358612,
+      "loss": 2.1033,
+      "step": 279
+    },
+    {
+      "epoch": 0.24927665257066547,
+      "grad_norm": 2.1256103515625,
+      "learning_rate": 0.0001651833725300879,
+      "loss": 2.1375,
+      "step": 280
+    },
+    {
+      "epoch": 0.25016692632984644,
+      "grad_norm": 1.4635076522827148,
+      "learning_rate": 0.00016494480483301836,
+      "loss": 1.6513,
+      "step": 281
+    },
+    {
+      "epoch": 0.2510572000890274,
+      "grad_norm": 2.5206878185272217,
+      "learning_rate": 0.00016470559615694446,
+      "loss": 1.5158,
+      "step": 282
+    },
+    {
+      "epoch": 0.2519474738482083,
+      "grad_norm": 1.6548582315444946,
+      "learning_rate": 0.00016446574886275913,
+      "loss": 2.1055,
+      "step": 283
+    },
+    {
+      "epoch": 0.25283774760738925,
+      "grad_norm": 2.238192558288574,
+      "learning_rate": 0.00016422526531765846,
+      "loss": 2.2413,
+      "step": 284
+    },
+    {
+      "epoch": 0.25372802136657024,
+      "grad_norm": 1.7222086191177368,
+      "learning_rate": 0.00016398414789511786,
+      "loss": 2.0776,
+      "step": 285
+    },
+    {
+      "epoch": 0.2546182951257512,
+      "grad_norm": 3.4774537086486816,
+      "learning_rate": 0.000163742398974869,
+      "loss": 2.3177,
+      "step": 286
+    },
+    {
+      "epoch": 0.2555085688849321,
+      "grad_norm": 1.5054153203964233,
+      "learning_rate": 0.00016350002094287609,
+      "loss": 1.397,
+      "step": 287
+    },
+    {
+      "epoch": 0.25639884264411306,
+      "grad_norm": 2.4916911125183105,
+      "learning_rate": 0.00016325701619131246,
+      "loss": 2.4185,
+      "step": 288
+    },
+    {
+      "epoch": 0.257289116403294,
+      "grad_norm": 1.975932240486145,
+      "learning_rate": 0.00016301338711853693,
+      "loss": 1.79,
+      "step": 289
+    },
+    {
+      "epoch": 0.258179390162475,
+      "grad_norm": 2.211235761642456,
+      "learning_rate": 0.00016276913612907007,
+      "loss": 2.0443,
+      "step": 290
+    },
+    {
+      "epoch": 0.2590696639216559,
+      "grad_norm": 2.4086029529571533,
+      "learning_rate": 0.00016252426563357055,
+      "loss": 1.9955,
+      "step": 291
+    },
+    {
+      "epoch": 0.25995993768083686,
+      "grad_norm": 1.6216076612472534,
+      "learning_rate": 0.00016227877804881127,
+      "loss": 1.4914,
+      "step": 292
+    },
+    {
+      "epoch": 0.2608502114400178,
+      "grad_norm": 1.9729924201965332,
+      "learning_rate": 0.00016203267579765563,
+      "loss": 2.484,
+      "step": 293
+    },
+    {
+      "epoch": 0.26174048519919874,
+      "grad_norm": 1.7074676752090454,
+      "learning_rate": 0.00016178596130903344,
+      "loss": 1.9975,
+      "step": 294
+    },
+    {
+      "epoch": 0.26263075895837973,
+      "grad_norm": 2.2856390476226807,
+      "learning_rate": 0.00016153863701791717,
+      "loss": 2.5118,
+      "step": 295
+    },
+    {
+      "epoch": 0.26352103271756067,
+      "grad_norm": 2.10971999168396,
+      "learning_rate": 0.00016129070536529766,
+      "loss": 1.9017,
+      "step": 296
+    },
+    {
+      "epoch": 0.2644113064767416,
+      "grad_norm": 1.7737220525741577,
+      "learning_rate": 0.00016104216879816026,
+      "loss": 1.5883,
+      "step": 297
+    },
+    {
+      "epoch": 0.26530158023592254,
+      "grad_norm": 2.5703797340393066,
+      "learning_rate": 0.00016079302976946055,
+      "loss": 2.4941,
+      "step": 298
+    },
+    {
+      "epoch": 0.2661918539951035,
+      "grad_norm": 2.0059597492218018,
+      "learning_rate": 0.00016054329073810015,
+      "loss": 2.5748,
+      "step": 299
+    },
+    {
+      "epoch": 0.2670821277542844,
+      "grad_norm": 1.8103333711624146,
+      "learning_rate": 0.00016029295416890248,
+      "loss": 1.7253,
+      "step": 300
+    },
+    {
+      "epoch": 0.2679724015134654,
+      "grad_norm": 1.659818410873413,
+      "learning_rate": 0.00016004202253258842,
+      "loss": 1.727,
+      "step": 301
+    },
+    {
+      "epoch": 0.26886267527264635,
+      "grad_norm": 1.6676276922225952,
+      "learning_rate": 0.0001597904983057519,
+      "loss": 1.8779,
+      "step": 302
+    },
+    {
+      "epoch": 0.2697529490318273,
+      "grad_norm": 2.9163947105407715,
+      "learning_rate": 0.00015953838397083552,
+      "loss": 1.7142,
+      "step": 303
+    },
+    {
+      "epoch": 0.2706432227910082,
+      "grad_norm": 1.6108711957931519,
+      "learning_rate": 0.00015928568201610595,
+      "loss": 1.7663,
+      "step": 304
+    },
+    {
+      "epoch": 0.27153349655018916,
+      "grad_norm": 1.8713996410369873,
+      "learning_rate": 0.00015903239493562948,
+      "loss": 1.8855,
+      "step": 305
+    },
+    {
+      "epoch": 0.27242377030937015,
+      "grad_norm": 1.5807499885559082,
+      "learning_rate": 0.00015877852522924732,
+      "loss": 2.17,
+      "step": 306
+    },
+    {
+      "epoch": 0.2733140440685511,
+      "grad_norm": 1.5979249477386475,
+      "learning_rate": 0.00015852407540255104,
+      "loss": 1.6459,
+      "step": 307
+    },
+    {
+      "epoch": 0.274204317827732,
+      "grad_norm": 2.3570306301116943,
+      "learning_rate": 0.00015826904796685762,
+      "loss": 2.2516,
+      "step": 308
+    },
+    {
+      "epoch": 0.27509459158691296,
+      "grad_norm": 2.0605814456939697,
+      "learning_rate": 0.00015801344543918495,
+      "loss": 1.4365,
+      "step": 309
+    },
+    {
+      "epoch": 0.2759848653460939,
+      "grad_norm": 2.146772861480713,
+      "learning_rate": 0.00015775727034222675,
+      "loss": 2.0198,
+      "step": 310
+    },
+    {
+      "epoch": 0.2768751391052749,
+      "grad_norm": 2.2532994747161865,
+      "learning_rate": 0.00015750052520432787,
+      "loss": 2.3515,
+      "step": 311
+    },
+    {
+      "epoch": 0.27776541286445583,
+      "grad_norm": 2.261044502258301,
+      "learning_rate": 0.0001572432125594591,
+      "loss": 1.9073,
+      "step": 312
+    },
+    {
+      "epoch": 0.27865568662363677,
+      "grad_norm": 1.7866798639297485,
+      "learning_rate": 0.00015698533494719238,
+      "loss": 1.4373,
+      "step": 313
+    },
+    {
+      "epoch": 0.2795459603828177,
+      "grad_norm": 1.5028505325317383,
+      "learning_rate": 0.00015672689491267567,
+      "loss": 1.7461,
+      "step": 314
+    },
+    {
+      "epoch": 0.28043623414199864,
+      "grad_norm": 2.3228518962860107,
+      "learning_rate": 0.00015646789500660773,
+      "loss": 2.063,
+      "step": 315
+    },
+    {
+      "epoch": 0.28132650790117963,
+      "grad_norm": 4.042271137237549,
+      "learning_rate": 0.00015620833778521307,
+      "loss": 1.6605,
+      "step": 316
+    },
+    {
+      "epoch": 0.28221678166036057,
+      "grad_norm": 2.26991868019104,
+      "learning_rate": 0.0001559482258102167,
+      "loss": 1.6581,
+      "step": 317
+    },
+    {
+      "epoch": 0.2831070554195415,
+      "grad_norm": 2.9728291034698486,
+      "learning_rate": 0.00015568756164881882,
+      "loss": 2.0517,
+      "step": 318
+    },
+    {
+      "epoch": 0.28399732917872245,
+      "grad_norm": 2.1857290267944336,
+      "learning_rate": 0.00015542634787366942,
+      "loss": 1.9245,
+      "step": 319
+    },
+    {
+      "epoch": 0.2848876029379034,
+      "grad_norm": 1.9066046476364136,
+      "learning_rate": 0.00015516458706284303,
+      "loss": 2.2897,
+      "step": 320
+    },
+    {
+      "epoch": 0.2857778766970844,
+      "grad_norm": 1.6206704378128052,
+      "learning_rate": 0.0001549022817998132,
+      "loss": 1.327,
+      "step": 321
+    },
+    {
+      "epoch": 0.2866681504562653,
+      "grad_norm": 2.0593979358673096,
+      "learning_rate": 0.00015463943467342693,
+      "loss": 1.8543,
+      "step": 322
+    },
+    {
+      "epoch": 0.28755842421544625,
+      "grad_norm": 1.762455940246582,
+      "learning_rate": 0.00015437604827787927,
+      "loss": 1.278,
+      "step": 323
+    },
+    {
+      "epoch": 0.2884486979746272,
+      "grad_norm": 2.4268031120300293,
+      "learning_rate": 0.00015411212521268758,
+      "loss": 2.4316,
+      "step": 324
+    },
+    {
+      "epoch": 0.2893389717338081,
+      "grad_norm": 1.6024727821350098,
+      "learning_rate": 0.00015384766808266602,
+      "loss": 1.9825,
+      "step": 325
+    },
+    {
+      "epoch": 0.2902292454929891,
+      "grad_norm": 2.4106545448303223,
+      "learning_rate": 0.00015358267949789966,
+      "loss": 2.4162,
+      "step": 326
+    },
+    {
+      "epoch": 0.29111951925217006,
+      "grad_norm": 1.7342814207077026,
+      "learning_rate": 0.00015331716207371888,
+      "loss": 2.1146,
+      "step": 327
+    },
+    {
+      "epoch": 0.292009793011351,
+      "grad_norm": 2.6861941814422607,
+      "learning_rate": 0.0001530511184306734,
+      "loss": 2.3119,
+      "step": 328
+    },
+    {
+      "epoch": 0.29290006677053193,
+      "grad_norm": 2.7844765186309814,
+      "learning_rate": 0.00015278455119450664,
+      "loss": 1.779,
+      "step": 329
+    },
+    {
+      "epoch": 0.29379034052971287,
+      "grad_norm": 2.8383467197418213,
+      "learning_rate": 0.0001525174629961296,
+      "loss": 2.0959,
+      "step": 330
+    },
+    {
+      "epoch": 0.29468061428889386,
+      "grad_norm": 1.988197922706604,
+      "learning_rate": 0.0001522498564715949,
+      "loss": 2.205,
+      "step": 331
+    },
+    {
+      "epoch": 0.2955708880480748,
+      "grad_norm": 2.0876247882843018,
+      "learning_rate": 0.00015198173426207094,
+      "loss": 1.6833,
+      "step": 332
+    },
+    {
+      "epoch": 0.29646116180725574,
+      "grad_norm": 1.5038071870803833,
+      "learning_rate": 0.00015171309901381572,
+      "loss": 1.5355,
+      "step": 333
+    },
+    {
+      "epoch": 0.2973514355664367,
+      "grad_norm": 1.8406671285629272,
+      "learning_rate": 0.00015144395337815064,
+      "loss": 1.6882,
+      "step": 334
+    },
+    {
+      "epoch": 0.2982417093256176,
+      "grad_norm": 1.7252678871154785,
+      "learning_rate": 0.00015117430001143452,
+      "loss": 1.8316,
+      "step": 335
+    },
+    {
+      "epoch": 0.2991319830847986,
+      "grad_norm": 2.3191328048706055,
+      "learning_rate": 0.00015090414157503714,
+      "loss": 1.7169,
+      "step": 336
+    },
+    {
+      "epoch": 0.30002225684397954,
+      "grad_norm": 1.5416702032089233,
+      "learning_rate": 0.00015063348073531324,
+      "loss": 1.7561,
+      "step": 337
+    },
+    {
+      "epoch": 0.3009125306031605,
+      "grad_norm": 2.3278818130493164,
+      "learning_rate": 0.0001503623201635761,
+      "loss": 2.284,
+      "step": 338
+    },
+    {
+      "epoch": 0.3018028043623414,
+      "grad_norm": 2.2224559783935547,
+      "learning_rate": 0.000150090662536071,
+      "loss": 2.1782,
+      "step": 339
+    },
+    {
+      "epoch": 0.30269307812152235,
+      "grad_norm": 2.172250986099243,
+      "learning_rate": 0.0001498185105339491,
+      "loss": 2.119,
+      "step": 340
+    },
+    {
+      "epoch": 0.3035833518807033,
+      "grad_norm": 1.8174207210540771,
+      "learning_rate": 0.00014954586684324078,
+      "loss": 1.6587,
+      "step": 341
+    },
+    {
+      "epoch": 0.3044736256398843,
+      "grad_norm": 1.9933828115463257,
+      "learning_rate": 0.00014927273415482915,
+      "loss": 1.8421,
+      "step": 342
+    },
+    {
+      "epoch": 0.3053638993990652,
+      "grad_norm": 1.7593411207199097,
+      "learning_rate": 0.00014899911516442365,
+      "loss": 2.4904,
+      "step": 343
+    },
+    {
+      "epoch": 0.30625417315824616,
+      "grad_norm": 2.25948429107666,
+      "learning_rate": 0.00014872501257253323,
+      "loss": 2.3422,
+      "step": 344
+    },
+    {
+      "epoch": 0.3071444469174271,
+      "grad_norm": 2.7912564277648926,
+      "learning_rate": 0.0001484504290844398,
+      "loss": 2.271,
+      "step": 345
+    },
+    {
+      "epoch": 0.30803472067660803,
+      "grad_norm": 1.8514056205749512,
+      "learning_rate": 0.00014817536741017152,
+      "loss": 2.3421,
+      "step": 346
+    },
+    {
+      "epoch": 0.308924994435789,
+      "grad_norm": 4.047842025756836,
+      "learning_rate": 0.00014789983026447612,
+      "loss": 1.4962,
+      "step": 347
+    },
+    {
+      "epoch": 0.30981526819496996,
+      "grad_norm": 2.038198471069336,
+      "learning_rate": 0.0001476238203667939,
+      "loss": 1.8679,
+      "step": 348
+    },
+    {
+      "epoch": 0.3107055419541509,
+      "grad_norm": 1.5603249073028564,
+      "learning_rate": 0.0001473473404412312,
+      "loss": 2.2399,
+      "step": 349
+    },
+    {
+      "epoch": 0.31159581571333184,
+      "grad_norm": 2.0062670707702637,
+      "learning_rate": 0.0001470703932165333,
+      "loss": 2.1111,
+      "step": 350
+    },
+    {
+      "epoch": 0.3124860894725128,
+      "grad_norm": 1.6404775381088257,
+      "learning_rate": 0.00014679298142605734,
+      "loss": 1.7317,
+      "step": 351
+    },
+    {
+      "epoch": 0.31337636323169377,
+      "grad_norm": 2.2055864334106445,
+      "learning_rate": 0.00014651510780774583,
+      "loss": 1.7962,
+      "step": 352
+    },
+    {
+      "epoch": 0.3142666369908747,
+      "grad_norm": 2.013460636138916,
+      "learning_rate": 0.00014623677510409918,
+      "loss": 2.1917,
+      "step": 353
+    },
+    {
+      "epoch": 0.31515691075005564,
+      "grad_norm": 1.7454619407653809,
+      "learning_rate": 0.00014595798606214882,
+      "loss": 1.8126,
+      "step": 354
+    },
+    {
+      "epoch": 0.3160471845092366,
+      "grad_norm": 2.3334221839904785,
+      "learning_rate": 0.00014567874343342997,
+      "loss": 1.5489,
+      "step": 355
+    },
+    {
+      "epoch": 0.3169374582684175,
+      "grad_norm": 2.6440775394439697,
+      "learning_rate": 0.00014539904997395468,
+      "loss": 2.3619,
+      "step": 356
+    },
+    {
+      "epoch": 0.3178277320275985,
+      "grad_norm": 2.4322640895843506,
+      "learning_rate": 0.00014511890844418453,
+      "loss": 2.1962,
+      "step": 357
+    },
+    {
+      "epoch": 0.31871800578677945,
+      "grad_norm": 2.2413289546966553,
+      "learning_rate": 0.00014483832160900326,
+      "loss": 1.9381,
+      "step": 358
+    },
+    {
+      "epoch": 0.3196082795459604,
+      "grad_norm": 2.011824369430542,
+      "learning_rate": 0.00014455729223768966,
+      "loss": 1.8238,
+      "step": 359
+    },
+    {
+      "epoch": 0.3204985533051413,
+      "grad_norm": 1.9649301767349243,
+      "learning_rate": 0.0001442758231038902,
+      "loss": 1.446,
+      "step": 360
+    },
+    {
+      "epoch": 0.32138882706432226,
+      "grad_norm": 2.035813093185425,
+      "learning_rate": 0.00014399391698559152,
+      "loss": 1.9748,
+      "step": 361
+    },
+    {
+      "epoch": 0.32227910082350325,
+      "grad_norm": 2.2650058269500732,
+      "learning_rate": 0.0001437115766650933,
+      "loss": 1.8778,
+      "step": 362
+    },
+    {
+      "epoch": 0.3231693745826842,
+      "grad_norm": 4.823574066162109,
+      "learning_rate": 0.00014342880492898048,
+      "loss": 1.3213,
+      "step": 363
+    },
+    {
+      "epoch": 0.3240596483418651,
+      "grad_norm": 1.6868720054626465,
+      "learning_rate": 0.0001431456045680959,
+      "loss": 1.8927,
+      "step": 364
+    },
+    {
+      "epoch": 0.32494992210104606,
+      "grad_norm": 2.2304537296295166,
+      "learning_rate": 0.00014286197837751286,
+      "loss": 2.1925,
+      "step": 365
+    },
+    {
+      "epoch": 0.325840195860227,
+      "grad_norm": 1.998257040977478,
+      "learning_rate": 0.00014257792915650728,
+      "loss": 1.9694,
+      "step": 366
+    },
+    {
+      "epoch": 0.326730469619408,
+      "grad_norm": 2.4832890033721924,
+      "learning_rate": 0.00014229345970853032,
+      "loss": 2.319,
+      "step": 367
+    },
+    {
+      "epoch": 0.32762074337858893,
+      "grad_norm": 1.5136913061141968,
+      "learning_rate": 0.00014200857284118066,
+      "loss": 1.3975,
+      "step": 368
+    },
+    {
+      "epoch": 0.32851101713776987,
+      "grad_norm": 1.8113616704940796,
+      "learning_rate": 0.00014172327136617656,
+      "loss": 1.6132,
+      "step": 369
+    },
+    {
+      "epoch": 0.3294012908969508,
+      "grad_norm": 3.2161953449249268,
+      "learning_rate": 0.00014143755809932845,
+      "loss": 1.7368,
+      "step": 370
+    },
+    {
+      "epoch": 0.33029156465613174,
+      "grad_norm": 2.979186534881592,
+      "learning_rate": 0.00014115143586051088,
+      "loss": 2.2553,
+      "step": 371
+    },
+    {
+      "epoch": 0.33118183841531273,
+      "grad_norm": 4.46711540222168,
+      "learning_rate": 0.00014086490747363493,
+      "loss": 1.6109,
+      "step": 372
+    },
+    {
+      "epoch": 0.33207211217449367,
+      "grad_norm": 1.9303728342056274,
+      "learning_rate": 0.00014057797576662,
+      "loss": 1.7995,
+      "step": 373
+    },
+    {
+      "epoch": 0.3329623859336746,
+      "grad_norm": 1.720313310623169,
+      "learning_rate": 0.00014029064357136628,
+      "loss": 1.5064,
+      "step": 374
+    },
+    {
+      "epoch": 0.33385265969285555,
+      "grad_norm": 1.9013950824737549,
+      "learning_rate": 0.00014000291372372647,
+      "loss": 1.8142,
+      "step": 375
+    },
+    {
+      "epoch": 0.3347429334520365,
+      "grad_norm": 1.8569226264953613,
+      "learning_rate": 0.00013971478906347806,
+      "loss": 1.8789,
+      "step": 376
+    },
+    {
+      "epoch": 0.3356332072112175,
+      "grad_norm": 2.768855333328247,
+      "learning_rate": 0.00013942627243429512,
+      "loss": 2.3762,
+      "step": 377
+    },
+    {
+      "epoch": 0.3365234809703984,
+      "grad_norm": 2.8080546855926514,
+      "learning_rate": 0.00013913736668372026,
+      "loss": 2.5786,
+      "step": 378
+    },
+    {
+      "epoch": 0.33741375472957935,
+      "grad_norm": 1.9482845067977905,
+      "learning_rate": 0.00013884807466313663,
+      "loss": 1.6996,
+      "step": 379
+    },
+    {
+      "epoch": 0.3383040284887603,
+      "grad_norm": 2.0869028568267822,
+      "learning_rate": 0.00013855839922773968,
+      "loss": 2.0259,
+      "step": 380
+    },
+    {
+      "epoch": 0.3391943022479412,
+      "grad_norm": 2.383391857147217,
+      "learning_rate": 0.000138268343236509,
+      "loss": 1.8103,
+      "step": 381
+    },
+    {
+      "epoch": 0.34008457600712216,
+      "grad_norm": 2.783890724182129,
+      "learning_rate": 0.00013797790955218014,
+      "loss": 1.7385,
+      "step": 382
+    },
+    {
+      "epoch": 0.34097484976630316,
+      "grad_norm": 1.6383908987045288,
+      "learning_rate": 0.00013768710104121627,
+      "loss": 1.5179,
+      "step": 383
+    },
+    {
+      "epoch": 0.3418651235254841,
+      "grad_norm": 2.065347194671631,
+      "learning_rate": 0.00013739592057378003,
+      "loss": 1.8636,
+      "step": 384
+    },
+    {
+      "epoch": 0.34275539728466503,
+      "grad_norm": 2.55138897895813,
+      "learning_rate": 0.0001371043710237051,
+      "loss": 2.1537,
+      "step": 385
+    },
+    {
+      "epoch": 0.34364567104384597,
+      "grad_norm": 2.131523609161377,
+      "learning_rate": 0.00013681245526846783,
+      "loss": 2.7836,
+      "step": 386
+    },
+    {
+      "epoch": 0.3445359448030269,
+      "grad_norm": 2.4860782623291016,
+      "learning_rate": 0.0001365201761891588,
+      "loss": 1.6104,
+      "step": 387
+    },
+    {
+      "epoch": 0.3454262185622079,
+      "grad_norm": 1.913320541381836,
+      "learning_rate": 0.00013622753667045457,
+      "loss": 1.6247,
+      "step": 388
+    },
+    {
+      "epoch": 0.34631649232138884,
+      "grad_norm": 3.11645245552063,
+      "learning_rate": 0.00013593453960058908,
+      "loss": 2.4805,
+      "step": 389
+    },
+    {
+      "epoch": 0.3472067660805698,
+      "grad_norm": 1.9025098085403442,
+      "learning_rate": 0.00013564118787132506,
+      "loss": 2.1268,
+      "step": 390
+    },
+    {
+      "epoch": 0.3480970398397507,
+      "grad_norm": 2.1510019302368164,
+      "learning_rate": 0.00013534748437792573,
+      "loss": 1.7887,
+      "step": 391
+    },
+    {
+      "epoch": 0.34898731359893165,
+      "grad_norm": 2.0571095943450928,
+      "learning_rate": 0.0001350534320191259,
+      "loss": 2.1384,
+      "step": 392
+    },
+    {
+      "epoch": 0.34987758735811264,
+      "grad_norm": 1.7569631338119507,
+      "learning_rate": 0.0001347590336971037,
+      "loss": 1.7031,
+      "step": 393
+    },
+    {
+      "epoch": 0.3507678611172936,
+      "grad_norm": 1.738153338432312,
+      "learning_rate": 0.0001344642923174517,
+      "loss": 1.9589,
+      "step": 394
+    },
+    {
+      "epoch": 0.3516581348764745,
+      "grad_norm": 2.1747841835021973,
+      "learning_rate": 0.00013416921078914835,
+      "loss": 2.0953,
+      "step": 395
+    },
+    {
+      "epoch": 0.35254840863565545,
+      "grad_norm": 1.6244169473648071,
+      "learning_rate": 0.00013387379202452917,
+      "loss": 1.2683,
+      "step": 396
+    },
+    {
+      "epoch": 0.3534386823948364,
+      "grad_norm": 1.4268521070480347,
+      "learning_rate": 0.00013357803893925807,
+      "loss": 1.7831,
+      "step": 397
+    },
+    {
+      "epoch": 0.3543289561540174,
+      "grad_norm": 2.6630098819732666,
+      "learning_rate": 0.00013328195445229868,
+      "loss": 2.2224,
+      "step": 398
+    },
+    {
+      "epoch": 0.3552192299131983,
+      "grad_norm": 1.8395928144454956,
+      "learning_rate": 0.00013298554148588528,
+      "loss": 1.7026,
+      "step": 399
+    },
+    {
+      "epoch": 0.35610950367237926,
+      "grad_norm": 1.6070834398269653,
+      "learning_rate": 0.00013268880296549425,
+      "loss": 1.6763,
+      "step": 400
+    },
+    {
+      "epoch": 0.3569997774315602,
+      "grad_norm": 2.3490657806396484,
+      "learning_rate": 0.00013239174181981495,
+      "loss": 1.6599,
+      "step": 401
+    },
+    {
+      "epoch": 0.35789005119074113,
+      "grad_norm": 3.1510345935821533,
+      "learning_rate": 0.00013209436098072095,
+      "loss": 1.247,
+      "step": 402
+    },
+    {
+      "epoch": 0.3587803249499221,
+      "grad_norm": 1.871864676475525,
+      "learning_rate": 0.00013179666338324108,
+      "loss": 2.056,
+      "step": 403
+    },
+    {
+      "epoch": 0.35967059870910306,
+      "grad_norm": 2.7115745544433594,
+      "learning_rate": 0.0001314986519655305,
+      "loss": 1.7978,
+      "step": 404
+    },
+    {
+      "epoch": 0.360560872468284,
+      "grad_norm": 1.72799551486969,
+      "learning_rate": 0.0001312003296688415,
+      "loss": 1.8869,
+      "step": 405
+    },
+    {
+      "epoch": 0.36145114622746494,
+      "grad_norm": 1.9090784788131714,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 2.0729,
+      "step": 406
+    },
+    {
+      "epoch": 0.3623414199866459,
+      "grad_norm": 1.7954814434051514,
+      "learning_rate": 0.0001306027642188501,
+      "loss": 1.6842,
+      "step": 407
+    },
+    {
+      "epoch": 0.36323169374582687,
+      "grad_norm": 2.4236233234405518,
+      "learning_rate": 0.00013030352696327742,
+      "loss": 1.6774,
+      "step": 408
+    },
+    {
+      "epoch": 0.3641219675050078,
+      "grad_norm": 1.9163222312927246,
+      "learning_rate": 0.00013000399062412763,
+      "loss": 1.8087,
+      "step": 409
+    },
+    {
+      "epoch": 0.36501224126418874,
+      "grad_norm": 1.9570804834365845,
+      "learning_rate": 0.0001297041581577035,
+      "loss": 2.2163,
+      "step": 410
+    },
+    {
+      "epoch": 0.3659025150233697,
+      "grad_norm": 1.8605859279632568,
+      "learning_rate": 0.0001294040325232304,
+      "loss": 1.7177,
+      "step": 411
+    },
+    {
+      "epoch": 0.3667927887825506,
+      "grad_norm": 1.9943171739578247,
+      "learning_rate": 0.00012910361668282719,
+      "loss": 2.0689,
+      "step": 412
+    },
+    {
+      "epoch": 0.3676830625417316,
+      "grad_norm": 1.6847420930862427,
+      "learning_rate": 0.00012880291360147693,
+      "loss": 1.3597,
+      "step": 413
+    },
+    {
+      "epoch": 0.36857333630091255,
+      "grad_norm": 2.013718366622925,
+      "learning_rate": 0.0001285019262469976,
+      "loss": 2.3517,
+      "step": 414
+    },
+    {
+      "epoch": 0.3694636100600935,
+      "grad_norm": 1.6312776803970337,
+      "learning_rate": 0.00012820065759001293,
+      "loss": 1.4005,
+      "step": 415
+    },
+    {
+      "epoch": 0.3703538838192744,
+      "grad_norm": 2.1821882724761963,
+      "learning_rate": 0.00012789911060392294,
+      "loss": 1.9826,
+      "step": 416
+    },
+    {
+      "epoch": 0.37124415757845536,
+      "grad_norm": 2.313922166824341,
+      "learning_rate": 0.0001275972882648746,
+      "loss": 2.01,
+      "step": 417
+    },
+    {
+      "epoch": 0.3721344313376363,
+      "grad_norm": 3.776573419570923,
+      "learning_rate": 0.00012729519355173254,
+      "loss": 1.6818,
+      "step": 418
+    },
+    {
+      "epoch": 0.3730247050968173,
+      "grad_norm": 2.0204479694366455,
+      "learning_rate": 0.00012699282944604967,
+      "loss": 1.6056,
+      "step": 419
+    },
+    {
+      "epoch": 0.3739149788559982,
+      "grad_norm": 1.9438761472702026,
+      "learning_rate": 0.00012669019893203759,
+      "loss": 2.2081,
+      "step": 420
+    },
+    {
+      "epoch": 0.37480525261517916,
+      "grad_norm": 1.4874850511550903,
+      "learning_rate": 0.0001263873049965373,
+      "loss": 1.872,
+      "step": 421
+    },
+    {
+      "epoch": 0.3756955263743601,
+      "grad_norm": 2.692265272140503,
+      "learning_rate": 0.00012608415062898972,
+      "loss": 2.3535,
+      "step": 422
+    },
+    {
+      "epoch": 0.37658580013354104,
+      "grad_norm": 2.4645423889160156,
+      "learning_rate": 0.000125780738821406,
+      "loss": 2.0317,
+      "step": 423
+    },
+    {
+      "epoch": 0.37747607389272203,
+      "grad_norm": 1.6828348636627197,
+      "learning_rate": 0.00012547707256833823,
+      "loss": 1.5422,
+      "step": 424
+    },
+    {
+      "epoch": 0.37836634765190297,
+      "grad_norm": 2.2978315353393555,
+      "learning_rate": 0.00012517315486684972,
+      "loss": 1.5563,
+      "step": 425
+    },
+    {
+      "epoch": 0.3792566214110839,
+      "grad_norm": 2.2748594284057617,
+      "learning_rate": 0.0001248689887164855,
+      "loss": 2.0274,
+      "step": 426
+    },
+    {
+      "epoch": 0.38014689517026484,
+      "grad_norm": 1.9515248537063599,
+      "learning_rate": 0.00012456457711924266,
+      "loss": 1.569,
+      "step": 427
+    },
+    {
+      "epoch": 0.3810371689294458,
+      "grad_norm": 2.134955644607544,
+      "learning_rate": 0.00012425992307954075,
+      "loss": 2.2598,
+      "step": 428
+    },
+    {
+      "epoch": 0.3819274426886268,
+      "grad_norm": 4.138272285461426,
+      "learning_rate": 0.0001239550296041922,
+      "loss": 2.3726,
+      "step": 429
+    },
+    {
+      "epoch": 0.3828177164478077,
+      "grad_norm": 2.6821060180664062,
+      "learning_rate": 0.00012364989970237248,
+      "loss": 1.8071,
+      "step": 430
+    },
+    {
+      "epoch": 0.38370799020698865,
+      "grad_norm": 2.3056938648223877,
+      "learning_rate": 0.00012334453638559057,
+      "loss": 2.4385,
+      "step": 431
+    },
+    {
+      "epoch": 0.3845982639661696,
+      "grad_norm": 2.1225337982177734,
+      "learning_rate": 0.00012303894266765908,
+      "loss": 2.2096,
+      "step": 432
+    },
+    {
+      "epoch": 0.3854885377253505,
+      "grad_norm": 2.1250829696655273,
+      "learning_rate": 0.00012273312156466464,
+      "loss": 2.1836,
+      "step": 433
+    },
+    {
+      "epoch": 0.3863788114845315,
+      "grad_norm": 2.2559192180633545,
+      "learning_rate": 0.00012242707609493814,
+      "loss": 2.0049,
+      "step": 434
+    },
+    {
+      "epoch": 0.38726908524371245,
+      "grad_norm": 1.6558101177215576,
+      "learning_rate": 0.00012212080927902474,
+      "loss": 2.542,
+      "step": 435
+    },
+    {
+      "epoch": 0.3881593590028934,
+      "grad_norm": 2.9783380031585693,
+      "learning_rate": 0.00012181432413965428,
+      "loss": 2.2321,
+      "step": 436
+    },
+    {
+      "epoch": 0.3890496327620743,
+      "grad_norm": 2.465571165084839,
+      "learning_rate": 0.00012150762370171136,
+      "loss": 2.2325,
+      "step": 437
+    },
+    {
+      "epoch": 0.38993990652125526,
+      "grad_norm": 2.5568687915802,
+      "learning_rate": 0.00012120071099220549,
+      "loss": 1.9866,
+      "step": 438
+    },
+    {
+      "epoch": 0.39083018028043626,
+      "grad_norm": 1.9345122575759888,
+      "learning_rate": 0.00012089358904024117,
+      "loss": 1.977,
+      "step": 439
+    },
+    {
+      "epoch": 0.3917204540396172,
+      "grad_norm": 1.9170026779174805,
+      "learning_rate": 0.00012058626087698814,
+      "loss": 1.9797,
+      "step": 440
+    },
+    {
+      "epoch": 0.39261072779879813,
+      "grad_norm": 2.2700252532958984,
+      "learning_rate": 0.00012027872953565125,
+      "loss": 1.4902,
+      "step": 441
+    },
+    {
+      "epoch": 0.39350100155797907,
+      "grad_norm": 2.4803783893585205,
+      "learning_rate": 0.00011997099805144069,
+      "loss": 2.3046,
+      "step": 442
+    },
+    {
+      "epoch": 0.39439127531716,
+      "grad_norm": 1.6946451663970947,
+      "learning_rate": 0.000119663069461542,
+      "loss": 1.6621,
+      "step": 443
+    },
+    {
+      "epoch": 0.395281549076341,
+      "grad_norm": 1.673259973526001,
+      "learning_rate": 0.00011935494680508606,
+      "loss": 1.9749,
+      "step": 444
+    },
+    {
+      "epoch": 0.39617182283552194,
+      "grad_norm": 1.9574657678604126,
+      "learning_rate": 0.00011904663312311901,
+      "loss": 2.1803,
+      "step": 445
+    },
+    {
+      "epoch": 0.3970620965947029,
+      "grad_norm": 1.7250328063964844,
+      "learning_rate": 0.00011873813145857249,
+      "loss": 1.8238,
+      "step": 446
+    },
+    {
+      "epoch": 0.3979523703538838,
+      "grad_norm": 1.4955158233642578,
+      "learning_rate": 0.00011842944485623335,
+      "loss": 1.8026,
+      "step": 447
+    },
+    {
+      "epoch": 0.39884264411306475,
+      "grad_norm": 1.5842695236206055,
+      "learning_rate": 0.00011812057636271374,
+      "loss": 1.4951,
+      "step": 448
+    },
+    {
+      "epoch": 0.39973291787224574,
+      "grad_norm": 3.3510076999664307,
+      "learning_rate": 0.000117811529026421,
+      "loss": 2.1284,
+      "step": 449
+    },
+    {
+      "epoch": 0.4006231916314267,
+      "grad_norm": 1.7413403987884521,
+      "learning_rate": 0.00011750230589752762,
+      "loss": 2.1252,
+      "step": 450
+    },
+    {
+      "epoch": 0.4015134653906076,
+      "grad_norm": 1.716800332069397,
+      "learning_rate": 0.00011719291002794096,
+      "loss": 1.9248,
+      "step": 451
+    },
+    {
+      "epoch": 0.40240373914978855,
+      "grad_norm": 1.9150794744491577,
+      "learning_rate": 0.00011688334447127338,
+      "loss": 1.9823,
+      "step": 452
+    },
+    {
+      "epoch": 0.4032940129089695,
+      "grad_norm": 2.7563483715057373,
+      "learning_rate": 0.00011657361228281199,
+      "loss": 2.0374,
+      "step": 453
+    },
+    {
+      "epoch": 0.4041842866681505,
+      "grad_norm": 3.253457546234131,
+      "learning_rate": 0.00011626371651948838,
+      "loss": 2.2565,
+      "step": 454
+    },
+    {
+      "epoch": 0.4050745604273314,
+      "grad_norm": 1.7900830507278442,
+      "learning_rate": 0.00011595366023984864,
+      "loss": 1.7175,
+      "step": 455
+    },
+    {
+      "epoch": 0.40596483418651236,
+      "grad_norm": 1.7450424432754517,
+      "learning_rate": 0.0001156434465040231,
+      "loss": 1.8861,
+      "step": 456
+    },
+    {
+      "epoch": 0.4068551079456933,
+      "grad_norm": 3.2053165435791016,
+      "learning_rate": 0.00011533307837369607,
+      "loss": 1.8854,
+      "step": 457
+    },
+    {
+      "epoch": 0.40774538170487423,
+      "grad_norm": 4.137878894805908,
+      "learning_rate": 0.00011502255891207572,
+      "loss": 1.6806,
+      "step": 458
+    },
+    {
+      "epoch": 0.40863565546405517,
+      "grad_norm": 2.7607953548431396,
+      "learning_rate": 0.00011471189118386375,
+      "loss": 1.9632,
+      "step": 459
+    },
+    {
+      "epoch": 0.40952592922323616,
+      "grad_norm": 1.9842784404754639,
+      "learning_rate": 0.00011440107825522521,
+      "loss": 1.8746,
+      "step": 460
+    },
+    {
+      "epoch": 0.4104162029824171,
+      "grad_norm": 2.3230137825012207,
+      "learning_rate": 0.00011409012319375827,
+      "loss": 1.9887,
+      "step": 461
+    },
+    {
+      "epoch": 0.41130647674159804,
+      "grad_norm": 1.3049930334091187,
+      "learning_rate": 0.0001137790290684638,
+      "loss": 1.2191,
+      "step": 462
+    },
+    {
+      "epoch": 0.412196750500779,
+      "grad_norm": 2.8393821716308594,
+      "learning_rate": 0.00011346779894971527,
+      "loss": 2.3616,
+      "step": 463
+    },
+    {
+      "epoch": 0.4130870242599599,
+      "grad_norm": 1.8259540796279907,
+      "learning_rate": 0.00011315643590922827,
+      "loss": 1.755,
+      "step": 464
+    },
+    {
+      "epoch": 0.4139772980191409,
+      "grad_norm": 1.8513232469558716,
+      "learning_rate": 0.0001128449430200303,
+      "loss": 1.7911,
+      "step": 465
+    },
+    {
+      "epoch": 0.41486757177832184,
+      "grad_norm": 1.5856389999389648,
+      "learning_rate": 0.00011253332335643043,
+      "loss": 1.0047,
+      "step": 466
+    },
+    {
+      "epoch": 0.4157578455375028,
+      "grad_norm": 1.9679127931594849,
+      "learning_rate": 0.00011222157999398895,
+      "loss": 1.6357,
+      "step": 467
+    },
+    {
+      "epoch": 0.4166481192966837,
+      "grad_norm": 1.9347325563430786,
+      "learning_rate": 0.00011190971600948699,
+      "loss": 1.8921,
+      "step": 468
+    },
+    {
+      "epoch": 0.41753839305586465,
+      "grad_norm": 1.6752451658248901,
+      "learning_rate": 0.00011159773448089614,
+      "loss": 2.0727,
+      "step": 469
+    },
+    {
+      "epoch": 0.41842866681504565,
+      "grad_norm": 1.5543407201766968,
+      "learning_rate": 0.00011128563848734816,
+      "loss": 2.0131,
+      "step": 470
+    },
+    {
+      "epoch": 0.4193189405742266,
+      "grad_norm": 1.8420300483703613,
+      "learning_rate": 0.00011097343110910452,
+      "loss": 1.9657,
+      "step": 471
+    },
+    {
+      "epoch": 0.4202092143334075,
+      "grad_norm": 1.8071311712265015,
+      "learning_rate": 0.000110661115427526,
+      "loss": 1.5397,
+      "step": 472
+    },
+    {
+      "epoch": 0.42109948809258846,
+      "grad_norm": 2.43044114112854,
+      "learning_rate": 0.00011034869452504226,
+      "loss": 2.0463,
+      "step": 473
+    },
+    {
+      "epoch": 0.4219897618517694,
+      "grad_norm": 1.7965081930160522,
+      "learning_rate": 0.00011003617148512149,
+      "loss": 1.9827,
+      "step": 474
+    },
+    {
+      "epoch": 0.4228800356109504,
+      "grad_norm": 1.98358952999115,
+      "learning_rate": 0.00010972354939223996,
+      "loss": 1.989,
+      "step": 475
+    },
+    {
+      "epoch": 0.4237703093701313,
+      "grad_norm": 1.868479609489441,
+      "learning_rate": 0.00010941083133185146,
+      "loss": 2.0278,
+      "step": 476
+    },
+    {
+      "epoch": 0.42466058312931226,
+      "grad_norm": 1.813387393951416,
+      "learning_rate": 0.00010909802039035701,
+      "loss": 1.6557,
+      "step": 477
+    },
+    {
+      "epoch": 0.4255508568884932,
+      "grad_norm": 5.017477989196777,
+      "learning_rate": 0.00010878511965507434,
+      "loss": 2.2248,
+      "step": 478
+    },
+    {
+      "epoch": 0.42644113064767414,
+      "grad_norm": 1.644929051399231,
+      "learning_rate": 0.00010847213221420736,
+      "loss": 1.8849,
+      "step": 479
+    },
+    {
+      "epoch": 0.42733140440685513,
+      "grad_norm": 1.4300436973571777,
+      "learning_rate": 0.00010815906115681578,
+      "loss": 1.2436,
+      "step": 480
+    },
+    {
+      "epoch": 0.42822167816603607,
+      "grad_norm": 1.765423059463501,
+      "learning_rate": 0.0001078459095727845,
+      "loss": 1.8819,
+      "step": 481
+    },
+    {
+      "epoch": 0.429111951925217,
+      "grad_norm": 2.2176973819732666,
+      "learning_rate": 0.00010753268055279329,
+      "loss": 1.7296,
+      "step": 482
+    },
+    {
+      "epoch": 0.43000222568439794,
+      "grad_norm": 2.369621515274048,
+      "learning_rate": 0.0001072193771882861,
+      "loss": 2.3114,
+      "step": 483
+    },
+    {
+      "epoch": 0.4308924994435789,
+      "grad_norm": 2.4713504314422607,
+      "learning_rate": 0.00010690600257144061,
+      "loss": 2.204,
+      "step": 484
+    },
+    {
+      "epoch": 0.4317827732027599,
+      "grad_norm": 2.2376275062561035,
+      "learning_rate": 0.0001065925597951378,
+      "loss": 1.7342,
+      "step": 485
+    },
+    {
+      "epoch": 0.4326730469619408,
+      "grad_norm": 1.8253062963485718,
+      "learning_rate": 0.00010627905195293135,
+      "loss": 1.1762,
+      "step": 486
+    },
+    {
+      "epoch": 0.43356332072112175,
+      "grad_norm": 3.376502513885498,
+      "learning_rate": 0.00010596548213901708,
+      "loss": 2.3694,
+      "step": 487
+    },
+    {
+      "epoch": 0.4344535944803027,
+      "grad_norm": 1.9548606872558594,
+      "learning_rate": 0.00010565185344820247,
+      "loss": 2.0768,
+      "step": 488
+    },
+    {
+      "epoch": 0.4353438682394836,
+      "grad_norm": 1.488417148590088,
+      "learning_rate": 0.00010533816897587606,
+      "loss": 1.7886,
+      "step": 489
+    },
+    {
+      "epoch": 0.4362341419986646,
+      "grad_norm": 1.4256919622421265,
+      "learning_rate": 0.00010502443181797697,
+      "loss": 1.6104,
+      "step": 490
+    },
+    {
+      "epoch": 0.43712441575784555,
+      "grad_norm": 2.147977828979492,
+      "learning_rate": 0.00010471064507096426,
+      "loss": 1.8361,
+      "step": 491
+    },
+    {
+      "epoch": 0.4380146895170265,
+      "grad_norm": 1.7447888851165771,
+      "learning_rate": 0.0001043968118317865,
+      "loss": 1.9855,
+      "step": 492
+    },
+    {
+      "epoch": 0.4389049632762074,
+      "grad_norm": 1.9062583446502686,
+      "learning_rate": 0.00010408293519785101,
+      "loss": 2.5487,
+      "step": 493
+    },
+    {
+      "epoch": 0.43979523703538836,
+      "grad_norm": 2.2021713256835938,
+      "learning_rate": 0.00010376901826699348,
+      "loss": 2.1764,
+      "step": 494
+    },
+    {
+      "epoch": 0.44068551079456936,
+      "grad_norm": 1.2943754196166992,
+      "learning_rate": 0.00010345506413744726,
+      "loss": 1.6313,
+      "step": 495
+    },
+    {
+      "epoch": 0.4415757845537503,
+      "grad_norm": 1.3317290544509888,
+      "learning_rate": 0.00010314107590781284,
+      "loss": 1.5996,
+      "step": 496
+    },
+    {
+      "epoch": 0.44246605831293123,
+      "grad_norm": 2.0627801418304443,
+      "learning_rate": 0.00010282705667702734,
+      "loss": 1.8221,
+      "step": 497
+    },
+    {
+      "epoch": 0.44335633207211217,
+      "grad_norm": 2.628443717956543,
+      "learning_rate": 0.00010251300954433376,
+      "loss": 1.2766,
+      "step": 498
+    },
+    {
+      "epoch": 0.4442466058312931,
+      "grad_norm": 2.0734922885894775,
+      "learning_rate": 0.00010219893760925052,
+      "loss": 2.1783,
+      "step": 499
+    },
+    {
+      "epoch": 0.44513687959047404,
+      "grad_norm": 1.527775764465332,
+      "learning_rate": 0.00010188484397154084,
+      "loss": 1.733,
+      "step": 500
+    },
+    {
+      "epoch": 0.44602715334965504,
+      "grad_norm": 2.605490207672119,
+      "learning_rate": 0.00010157073173118208,
+      "loss": 2.1727,
+      "step": 501
+    },
+    {
+      "epoch": 0.446917427108836,
+      "grad_norm": 2.056980609893799,
+      "learning_rate": 0.00010125660398833528,
+      "loss": 2.4628,
+      "step": 502
+    },
+    {
+      "epoch": 0.4478077008680169,
+      "grad_norm": 3.1581008434295654,
+      "learning_rate": 0.00010094246384331442,
+      "loss": 2.2271,
+      "step": 503
+    },
+    {
+      "epoch": 0.44869797462719785,
+      "grad_norm": 1.937180519104004,
+      "learning_rate": 0.00010062831439655591,
+      "loss": 2.0092,
+      "step": 504
+    },
+    {
+      "epoch": 0.4495882483863788,
+      "grad_norm": 2.2551982402801514,
+      "learning_rate": 0.00010031415874858797,
+      "loss": 1.9285,
+      "step": 505
+    },
+    {
+      "epoch": 0.4504785221455598,
+      "grad_norm": 3.13543438911438,
+      "learning_rate": 0.0001,
+      "loss": 1.8066,
+      "step": 506
+    },
+    {
+      "epoch": 0.4513687959047407,
+      "grad_norm": 2.2084224224090576,
+      "learning_rate": 9.968584125141204e-05,
+      "loss": 2.1068,
+      "step": 507
+    },
+    {
+      "epoch": 0.45225906966392165,
+      "grad_norm": 2.102102518081665,
+      "learning_rate": 9.937168560344412e-05,
+      "loss": 1.8429,
+      "step": 508
+    },
+    {
+      "epoch": 0.4531493434231026,
+      "grad_norm": 1.972419023513794,
+      "learning_rate": 9.90575361566856e-05,
+      "loss": 1.8751,
+      "step": 509
+    },
+    {
+      "epoch": 0.45403961718228353,
+      "grad_norm": 1.8030009269714355,
+      "learning_rate": 9.874339601166473e-05,
+      "loss": 1.9289,
+      "step": 510
+    },
+    {
+      "epoch": 0.4549298909414645,
+      "grad_norm": 2.8846678733825684,
+      "learning_rate": 9.842926826881796e-05,
+      "loss": 2.224,
+      "step": 511
+    },
+    {
+      "epoch": 0.45582016470064546,
+      "grad_norm": 1.920257329940796,
+      "learning_rate": 9.81151560284592e-05,
+      "loss": 1.9377,
+      "step": 512
+    },
+    {
+      "epoch": 0.4567104384598264,
+      "grad_norm": 1.8159655332565308,
+      "learning_rate": 9.78010623907495e-05,
+      "loss": 1.6231,
+      "step": 513
+    },
+    {
+      "epoch": 0.45760071221900733,
+      "grad_norm": 2.1245462894439697,
+      "learning_rate": 9.748699045566626e-05,
+      "loss": 2.3452,
+      "step": 514
+    },
+    {
+      "epoch": 0.45849098597818827,
+      "grad_norm": 1.6165993213653564,
+      "learning_rate": 9.717294332297268e-05,
+      "loss": 1.4252,
+      "step": 515
+    },
+    {
+      "epoch": 0.45938125973736926,
+      "grad_norm": 1.8090025186538696,
+      "learning_rate": 9.685892409218717e-05,
+      "loss": 1.552,
+      "step": 516
+    },
+    {
+      "epoch": 0.4602715334965502,
+      "grad_norm": 1.4432064294815063,
+      "learning_rate": 9.654493586255278e-05,
+      "loss": 1.4914,
+      "step": 517
+    },
+    {
+      "epoch": 0.46116180725573114,
+      "grad_norm": 2.572908878326416,
+      "learning_rate": 9.623098173300654e-05,
+      "loss": 2.4062,
+      "step": 518
+    },
+    {
+      "epoch": 0.4620520810149121,
+      "grad_norm": 2.2798383235931396,
+      "learning_rate": 9.591706480214901e-05,
+      "loss": 1.9997,
+      "step": 519
+    },
+    {
+      "epoch": 0.462942354774093,
+      "grad_norm": 1.5268291234970093,
+      "learning_rate": 9.560318816821353e-05,
+      "loss": 1.6246,
+      "step": 520
+    },
+    {
+      "epoch": 0.463832628533274,
+      "grad_norm": 1.6568892002105713,
+      "learning_rate": 9.528935492903575e-05,
+      "loss": 1.4726,
+      "step": 521
+    },
+    {
+      "epoch": 0.46472290229245494,
+      "grad_norm": 1.6617227792739868,
+      "learning_rate": 9.497556818202306e-05,
+      "loss": 2.3641,
+      "step": 522
+    },
+    {
+      "epoch": 0.4656131760516359,
+      "grad_norm": 1.6580737829208374,
+      "learning_rate": 9.466183102412395e-05,
+      "loss": 1.4021,
+      "step": 523
+    },
+    {
+      "epoch": 0.4665034498108168,
+      "grad_norm": 2.1587393283843994,
+      "learning_rate": 9.434814655179755e-05,
+      "loss": 2.0219,
+      "step": 524
+    },
+    {
+      "epoch": 0.46739372356999775,
+      "grad_norm": 1.6377232074737549,
+      "learning_rate": 9.403451786098294e-05,
+      "loss": 1.7731,
+      "step": 525
+    },
+    {
+      "epoch": 0.46828399732917875,
+      "grad_norm": 1.7926323413848877,
+      "learning_rate": 9.372094804706867e-05,
+      "loss": 1.8779,
+      "step": 526
+    },
+    {
+      "epoch": 0.4691742710883597,
+      "grad_norm": 1.9899790287017822,
+      "learning_rate": 9.340744020486222e-05,
+      "loss": 1.7574,
+      "step": 527
+    },
+    {
+      "epoch": 0.4700645448475406,
+      "grad_norm": 3.482309341430664,
+      "learning_rate": 9.309399742855942e-05,
+      "loss": 2.1806,
+      "step": 528
+    },
+    {
+      "epoch": 0.47095481860672156,
+      "grad_norm": 1.9405080080032349,
+      "learning_rate": 9.278062281171393e-05,
+      "loss": 1.7881,
+      "step": 529
+    },
+    {
+      "epoch": 0.4718450923659025,
+      "grad_norm": 3.1491637229919434,
+      "learning_rate": 9.246731944720675e-05,
+      "loss": 2.3996,
+      "step": 530
+    },
+    {
+      "epoch": 0.4727353661250835,
+      "grad_norm": 2.0104711055755615,
+      "learning_rate": 9.215409042721552e-05,
+      "loss": 2.3646,
+      "step": 531
+    },
+    {
+      "epoch": 0.4736256398842644,
+      "grad_norm": 1.6170440912246704,
+      "learning_rate": 9.184093884318425e-05,
+      "loss": 1.6469,
+      "step": 532
+    },
+    {
+      "epoch": 0.47451591364344536,
+      "grad_norm": 2.090585947036743,
+      "learning_rate": 9.152786778579267e-05,
+      "loss": 1.8581,
+      "step": 533
+    },
+    {
+      "epoch": 0.4754061874026263,
+      "grad_norm": 2.355839490890503,
+      "learning_rate": 9.121488034492569e-05,
+      "loss": 1.8487,
+      "step": 534
+    },
+    {
+      "epoch": 0.47629646116180724,
+      "grad_norm": 2.266519546508789,
+      "learning_rate": 9.090197960964301e-05,
+      "loss": 2.0358,
+      "step": 535
+    },
+    {
+      "epoch": 0.47718673492098823,
+      "grad_norm": 2.3160934448242188,
+      "learning_rate": 9.058916866814858e-05,
+      "loss": 2.1925,
+      "step": 536
+    },
+    {
+      "epoch": 0.47807700868016917,
+      "grad_norm": 1.7758020162582397,
+      "learning_rate": 9.027645060776006e-05,
+      "loss": 2.1548,
+      "step": 537
+    },
+    {
+      "epoch": 0.4789672824393501,
+      "grad_norm": 2.3746981620788574,
+      "learning_rate": 8.99638285148785e-05,
+      "loss": 2.144,
+      "step": 538
+    },
+    {
+      "epoch": 0.47985755619853104,
+      "grad_norm": 2.0492055416107178,
+      "learning_rate": 8.965130547495776e-05,
+      "loss": 1.78,
+      "step": 539
+    },
+    {
+      "epoch": 0.480747829957712,
+      "grad_norm": 1.5570842027664185,
+      "learning_rate": 8.933888457247402e-05,
+      "loss": 1.9701,
+      "step": 540
+    },
+    {
+      "epoch": 0.4816381037168929,
+      "grad_norm": 1.4735817909240723,
+      "learning_rate": 8.902656889089548e-05,
+      "loss": 1.3506,
+      "step": 541
+    },
+    {
+      "epoch": 0.4825283774760739,
+      "grad_norm": 1.8302412033081055,
+      "learning_rate": 8.871436151265184e-05,
+      "loss": 1.6521,
+      "step": 542
+    },
+    {
+      "epoch": 0.48341865123525485,
+      "grad_norm": 1.834014654159546,
+      "learning_rate": 8.840226551910387e-05,
+      "loss": 1.7078,
+      "step": 543
+    },
+    {
+      "epoch": 0.4843089249944358,
+      "grad_norm": 1.8256818056106567,
+      "learning_rate": 8.809028399051302e-05,
+      "loss": 1.982,
+      "step": 544
+    },
+    {
+      "epoch": 0.4851991987536167,
+      "grad_norm": 3.1654930114746094,
+      "learning_rate": 8.777842000601105e-05,
+      "loss": 1.8923,
+      "step": 545
+    },
+    {
+      "epoch": 0.48608947251279766,
+      "grad_norm": 2.625009059906006,
+      "learning_rate": 8.746667664356956e-05,
+      "loss": 1.5361,
+      "step": 546
+    },
+    {
+      "epoch": 0.48697974627197865,
+      "grad_norm": 1.5053600072860718,
+      "learning_rate": 8.715505697996971e-05,
+      "loss": 1.7811,
+      "step": 547
+    },
+    {
+      "epoch": 0.4878700200311596,
+      "grad_norm": 1.6677234172821045,
+      "learning_rate": 8.684356409077176e-05,
+      "loss": 1.6811,
+      "step": 548
+    },
+    {
+      "epoch": 0.4887602937903405,
+      "grad_norm": 1.4328875541687012,
+      "learning_rate": 8.653220105028474e-05,
+      "loss": 1.4466,
+      "step": 549
+    },
+    {
+      "epoch": 0.48965056754952146,
+      "grad_norm": 6.340122222900391,
+      "learning_rate": 8.62209709315362e-05,
+      "loss": 2.122,
+      "step": 550
+    },
+    {
+      "epoch": 0.4905408413087024,
+      "grad_norm": 2.1618196964263916,
+      "learning_rate": 8.590987680624174e-05,
+      "loss": 1.6239,
+      "step": 551
+    },
+    {
+      "epoch": 0.4914311150678834,
+      "grad_norm": 1.8359202146530151,
+      "learning_rate": 8.559892174477479e-05,
+      "loss": 1.9783,
+      "step": 552
+    },
+    {
+      "epoch": 0.49232138882706433,
+      "grad_norm": 1.959260106086731,
+      "learning_rate": 8.528810881613626e-05,
+      "loss": 1.9082,
+      "step": 553
+    },
+    {
+      "epoch": 0.49321166258624527,
+      "grad_norm": 1.870314121246338,
+      "learning_rate": 8.497744108792429e-05,
+      "loss": 1.786,
+      "step": 554
+    },
+    {
+      "epoch": 0.4941019363454262,
+      "grad_norm": 2.359234571456909,
+      "learning_rate": 8.466692162630392e-05,
+      "loss": 2.0353,
+      "step": 555
+    },
+    {
+      "epoch": 0.49499221010460714,
+      "grad_norm": 1.3800303936004639,
+      "learning_rate": 8.435655349597689e-05,
+      "loss": 1.3332,
+      "step": 556
+    },
+    {
+      "epoch": 0.49588248386378814,
+      "grad_norm": 2.568152904510498,
+      "learning_rate": 8.404633976015134e-05,
+      "loss": 2.4055,
+      "step": 557
+    },
+    {
+      "epoch": 0.4967727576229691,
+      "grad_norm": 1.5831269025802612,
+      "learning_rate": 8.373628348051165e-05,
+      "loss": 1.443,
+      "step": 558
+    },
+    {
+      "epoch": 0.49766303138215,
+      "grad_norm": 1.4538564682006836,
+      "learning_rate": 8.342638771718802e-05,
+      "loss": 1.3316,
+      "step": 559
+    },
+    {
+      "epoch": 0.49855330514133095,
+      "grad_norm": 1.7724665403366089,
+      "learning_rate": 8.311665552872662e-05,
+      "loss": 1.5942,
+      "step": 560
+    },
+    {
+      "epoch": 0.4994435789005119,
+      "grad_norm": 1.7002063989639282,
+      "learning_rate": 8.280708997205904e-05,
+      "loss": 1.877,
+      "step": 561
+    },
+    {
+      "epoch": 0.5003338526596929,
+      "grad_norm": 2.1765499114990234,
+      "learning_rate": 8.249769410247239e-05,
+      "loss": 2.2905,
+      "step": 562
+    },
+    {
+      "epoch": 0.5012241264188738,
+      "grad_norm": 2.2722952365875244,
+      "learning_rate": 8.218847097357898e-05,
+      "loss": 1.4615,
+      "step": 563
+    },
+    {
+      "epoch": 0.5021144001780548,
+      "grad_norm": 1.7580784559249878,
+      "learning_rate": 8.187942363728625e-05,
+      "loss": 2.1017,
+      "step": 564
+    },
+    {
+      "epoch": 0.5030046739372357,
+      "grad_norm": 2.0220673084259033,
+      "learning_rate": 8.157055514376666e-05,
+      "loss": 2.2579,
+      "step": 565
+    },
+    {
+      "epoch": 0.5038949476964166,
+      "grad_norm": 1.9967743158340454,
+      "learning_rate": 8.126186854142752e-05,
+      "loss": 1.6888,
+      "step": 566
+    },
+    {
+      "epoch": 0.5047852214555976,
+      "grad_norm": 1.4775068759918213,
+      "learning_rate": 8.095336687688102e-05,
+      "loss": 1.3643,
+      "step": 567
+    },
+    {
+      "epoch": 0.5056754952147785,
+      "grad_norm": 5.734138011932373,
+      "learning_rate": 8.064505319491398e-05,
+      "loss": 2.1258,
+      "step": 568
+    },
+    {
+      "epoch": 0.5065657689739594,
+      "grad_norm": 1.27615487575531,
+      "learning_rate": 8.033693053845801e-05,
+      "loss": 1.5468,
+      "step": 569
+    },
+    {
+      "epoch": 0.5074560427331405,
+      "grad_norm": 1.7572683095932007,
+      "learning_rate": 8.002900194855932e-05,
+      "loss": 1.7513,
+      "step": 570
+    },
+    {
+      "epoch": 0.5083463164923214,
+      "grad_norm": 1.6680421829223633,
+      "learning_rate": 7.972127046434878e-05,
+      "loss": 1.6567,
+      "step": 571
+    },
+    {
+      "epoch": 0.5092365902515024,
+      "grad_norm": 1.8064501285552979,
+      "learning_rate": 7.941373912301189e-05,
+      "loss": 1.6988,
+      "step": 572
+    },
+    {
+      "epoch": 0.5101268640106833,
+      "grad_norm": 1.894484519958496,
+      "learning_rate": 7.910641095975886e-05,
+      "loss": 1.65,
+      "step": 573
+    },
+    {
+      "epoch": 0.5110171377698642,
+      "grad_norm": 2.205246686935425,
+      "learning_rate": 7.879928900779456e-05,
+      "loss": 2.2756,
+      "step": 574
+    },
+    {
+      "epoch": 0.5119074115290452,
+      "grad_norm": 2.020557403564453,
+      "learning_rate": 7.849237629828869e-05,
+      "loss": 1.8741,
+      "step": 575
+    },
+    {
+      "epoch": 0.5127976852882261,
+      "grad_norm": 1.7418237924575806,
+      "learning_rate": 7.818567586034577e-05,
+      "loss": 1.8595,
+      "step": 576
+    },
+    {
+      "epoch": 0.513687959047407,
+      "grad_norm": 2.7566726207733154,
+      "learning_rate": 7.787919072097531e-05,
+      "loss": 2.1259,
+      "step": 577
+    },
+    {
+      "epoch": 0.514578232806588,
+      "grad_norm": 2.0560312271118164,
+      "learning_rate": 7.75729239050619e-05,
+      "loss": 1.5462,
+      "step": 578
+    },
+    {
+      "epoch": 0.5154685065657689,
+      "grad_norm": 3.745788335800171,
+      "learning_rate": 7.726687843533538e-05,
+      "loss": 2.482,
+      "step": 579
+    },
+    {
+      "epoch": 0.51635878032495,
+      "grad_norm": 3.1230719089508057,
+      "learning_rate": 7.696105733234098e-05,
+      "loss": 1.6565,
+      "step": 580
+    },
+    {
+      "epoch": 0.5172490540841309,
+      "grad_norm": 5.474112033843994,
+      "learning_rate": 7.66554636144095e-05,
+      "loss": 1.6507,
+      "step": 581
+    },
+    {
+      "epoch": 0.5181393278433118,
+      "grad_norm": 1.4968229532241821,
+      "learning_rate": 7.635010029762756e-05,
+      "loss": 1.5576,
+      "step": 582
+    },
+    {
+      "epoch": 0.5190296016024928,
+      "grad_norm": 1.8227341175079346,
+      "learning_rate": 7.604497039580785e-05,
+      "loss": 2.0556,
+      "step": 583
+    },
+    {
+      "epoch": 0.5199198753616737,
+      "grad_norm": 1.888366937637329,
+      "learning_rate": 7.574007692045928e-05,
+      "loss": 2.0041,
+      "step": 584
+    },
+    {
+      "epoch": 0.5208101491208547,
+      "grad_norm": 1.5480713844299316,
+      "learning_rate": 7.543542288075739e-05,
+      "loss": 1.6337,
+      "step": 585
+    },
+    {
+      "epoch": 0.5217004228800356,
+      "grad_norm": 2.2726070880889893,
+      "learning_rate": 7.513101128351454e-05,
+      "loss": 1.7892,
+      "step": 586
+    },
+    {
+      "epoch": 0.5225906966392165,
+      "grad_norm": 2.1441900730133057,
+      "learning_rate": 7.48268451331503e-05,
+      "loss": 2.2337,
+      "step": 587
+    },
+    {
+      "epoch": 0.5234809703983975,
+      "grad_norm": 2.6022322177886963,
+      "learning_rate": 7.45229274316618e-05,
+      "loss": 1.742,
+      "step": 588
+    },
+    {
+      "epoch": 0.5243712441575784,
+      "grad_norm": 2.375544548034668,
+      "learning_rate": 7.421926117859403e-05,
+      "loss": 1.8181,
+      "step": 589
+    },
+    {
+      "epoch": 0.5252615179167595,
+      "grad_norm": 1.6946896314620972,
+      "learning_rate": 7.391584937101033e-05,
+      "loss": 1.84,
+      "step": 590
+    },
+    {
+      "epoch": 0.5261517916759404,
+      "grad_norm": 1.648368239402771,
+      "learning_rate": 7.361269500346274e-05,
+      "loss": 1.9039,
+      "step": 591
+    },
+    {
+      "epoch": 0.5270420654351213,
+      "grad_norm": 2.288290500640869,
+      "learning_rate": 7.330980106796246e-05,
+      "loss": 1.7733,
+      "step": 592
+    },
+    {
+      "epoch": 0.5279323391943023,
+      "grad_norm": 1.911532998085022,
+      "learning_rate": 7.300717055395039e-05,
+      "loss": 2.1929,
+      "step": 593
+    },
+    {
+      "epoch": 0.5288226129534832,
+      "grad_norm": 2.3173465728759766,
+      "learning_rate": 7.270480644826749e-05,
+      "loss": 1.7691,
+      "step": 594
+    },
+    {
+      "epoch": 0.5297128867126641,
+      "grad_norm": 2.155731439590454,
+      "learning_rate": 7.240271173512546e-05,
+      "loss": 2.3191,
+      "step": 595
+    },
+    {
+      "epoch": 0.5306031604718451,
+      "grad_norm": 2.021113634109497,
+      "learning_rate": 7.210088939607708e-05,
+      "loss": 2.4124,
+      "step": 596
+    },
+    {
+      "epoch": 0.531493434231026,
+      "grad_norm": 1.988287091255188,
+      "learning_rate": 7.179934240998706e-05,
+      "loss": 1.3772,
+      "step": 597
+    },
+    {
+      "epoch": 0.532383707990207,
+      "grad_norm": 1.78477942943573,
+      "learning_rate": 7.149807375300239e-05,
+      "loss": 1.9691,
+      "step": 598
+    },
+    {
+      "epoch": 0.5332739817493879,
+      "grad_norm": 2.129854679107666,
+      "learning_rate": 7.119708639852312e-05,
+      "loss": 1.5694,
+      "step": 599
+    },
+    {
+      "epoch": 0.5341642555085688,
+      "grad_norm": 1.9304897785186768,
+      "learning_rate": 7.089638331717284e-05,
+      "loss": 2.1899,
+      "step": 600
+    },
+    {
+      "epoch": 0.5350545292677499,
+      "grad_norm": 1.955014944076538,
+      "learning_rate": 7.059596747676962e-05,
+      "loss": 1.9041,
+      "step": 601
+    },
+    {
+      "epoch": 0.5359448030269308,
+      "grad_norm": 1.9473272562026978,
+      "learning_rate": 7.029584184229653e-05,
+      "loss": 2.2461,
+      "step": 602
+    },
+    {
+      "epoch": 0.5368350767861118,
+      "grad_norm": 3.0118231773376465,
+      "learning_rate": 6.999600937587239e-05,
+      "loss": 1.8597,
+      "step": 603
+    },
+    {
+      "epoch": 0.5377253505452927,
+      "grad_norm": 1.5499767065048218,
+      "learning_rate": 6.969647303672262e-05,
+      "loss": 1.2288,
+      "step": 604
+    },
+    {
+      "epoch": 0.5386156243044736,
+      "grad_norm": 2.489497661590576,
+      "learning_rate": 6.939723578114993e-05,
+      "loss": 2.0863,
+      "step": 605
+    },
+    {
+      "epoch": 0.5395058980636546,
+      "grad_norm": 1.4985642433166504,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 1.6077,
+      "step": 606
+    },
+    {
+      "epoch": 0.5403961718228355,
+      "grad_norm": 2.099515676498413,
+      "learning_rate": 6.879967033115853e-05,
+      "loss": 1.8709,
+      "step": 607
+    },
+    {
+      "epoch": 0.5412864455820164,
+      "grad_norm": 1.7907514572143555,
+      "learning_rate": 6.850134803446954e-05,
+      "loss": 1.7509,
+      "step": 608
+    },
+    {
+      "epoch": 0.5421767193411974,
+      "grad_norm": 2.593832492828369,
+      "learning_rate": 6.820333661675893e-05,
+      "loss": 1.6985,
+      "step": 609
+    },
+    {
+      "epoch": 0.5430669931003783,
+      "grad_norm": 1.652707815170288,
+      "learning_rate": 6.790563901927907e-05,
+      "loss": 1.7336,
+      "step": 610
+    },
+    {
+      "epoch": 0.5439572668595594,
+      "grad_norm": 2.1123557090759277,
+      "learning_rate": 6.760825818018508e-05,
+      "loss": 2.0235,
+      "step": 611
+    },
+    {
+      "epoch": 0.5448475406187403,
+      "grad_norm": 1.8303403854370117,
+      "learning_rate": 6.731119703450577e-05,
+      "loss": 1.988,
+      "step": 612
+    },
+    {
+      "epoch": 0.5457378143779212,
+      "grad_norm": 1.746991753578186,
+      "learning_rate": 6.701445851411472e-05,
+      "loss": 1.7563,
+      "step": 613
+    },
+    {
+      "epoch": 0.5466280881371022,
+      "grad_norm": 1.520955204963684,
+      "learning_rate": 6.671804554770135e-05,
+      "loss": 1.2435,
+      "step": 614
+    },
+    {
+      "epoch": 0.5475183618962831,
+      "grad_norm": 2.8054568767547607,
+      "learning_rate": 6.642196106074194e-05,
+      "loss": 1.6526,
+      "step": 615
+    },
+    {
+      "epoch": 0.548408635655464,
+      "grad_norm": 3.158360004425049,
+      "learning_rate": 6.612620797547087e-05,
+      "loss": 2.4104,
+      "step": 616
+    },
+    {
+      "epoch": 0.549298909414645,
+      "grad_norm": 2.212456226348877,
+      "learning_rate": 6.583078921085167e-05,
+      "loss": 2.2195,
+      "step": 617
+    },
+    {
+      "epoch": 0.5501891831738259,
+      "grad_norm": 2.032042980194092,
+      "learning_rate": 6.55357076825483e-05,
+      "loss": 1.5883,
+      "step": 618
+    },
+    {
+      "epoch": 0.5510794569330069,
+      "grad_norm": 2.144111394882202,
+      "learning_rate": 6.52409663028963e-05,
+      "loss": 2.2491,
+      "step": 619
+    },
+    {
+      "epoch": 0.5519697306921878,
+      "grad_norm": 1.9133466482162476,
+      "learning_rate": 6.494656798087412e-05,
+      "loss": 1.5789,
+      "step": 620
+    },
+    {
+      "epoch": 0.5528600044513688,
+      "grad_norm": 1.6118453741073608,
+      "learning_rate": 6.465251562207431e-05,
+      "loss": 1.7489,
+      "step": 621
+    },
+    {
+      "epoch": 0.5537502782105498,
+      "grad_norm": 2.499858856201172,
+      "learning_rate": 6.435881212867493e-05,
+      "loss": 2.5306,
+      "step": 622
+    },
+    {
+      "epoch": 0.5546405519697307,
+      "grad_norm": 2.0919814109802246,
+      "learning_rate": 6.406546039941094e-05,
+      "loss": 1.71,
+      "step": 623
+    },
+    {
+      "epoch": 0.5555308257289117,
+      "grad_norm": 1.7591675519943237,
+      "learning_rate": 6.377246332954544e-05,
+      "loss": 1.8629,
+      "step": 624
+    },
+    {
+      "epoch": 0.5564210994880926,
+      "grad_norm": 1.4329679012298584,
+      "learning_rate": 6.347982381084123e-05,
+      "loss": 1.6061,
+      "step": 625
+    },
+    {
+      "epoch": 0.5573113732472735,
+      "grad_norm": 2.896592140197754,
+      "learning_rate": 6.318754473153221e-05,
+      "loss": 1.7499,
+      "step": 626
+    },
+    {
+      "epoch": 0.5582016470064545,
+      "grad_norm": 3.4776594638824463,
+      "learning_rate": 6.289562897629492e-05,
+      "loss": 1.8892,
+      "step": 627
+    },
+    {
+      "epoch": 0.5590919207656354,
+      "grad_norm": 2.0120224952697754,
+      "learning_rate": 6.260407942621998e-05,
+      "loss": 2.1044,
+      "step": 628
+    },
+    {
+      "epoch": 0.5599821945248163,
+      "grad_norm": 1.950958251953125,
+      "learning_rate": 6.231289895878375e-05,
+      "loss": 1.7997,
+      "step": 629
+    },
+    {
+      "epoch": 0.5608724682839973,
+      "grad_norm": 2.242223024368286,
+      "learning_rate": 6.20220904478199e-05,
+      "loss": 1.9631,
+      "step": 630
+    },
+    {
+      "epoch": 0.5617627420431783,
+      "grad_norm": 3.0490360260009766,
+      "learning_rate": 6.173165676349103e-05,
+      "loss": 1.8321,
+      "step": 631
+    },
+    {
+      "epoch": 0.5626530158023593,
+      "grad_norm": 1.520606279373169,
+      "learning_rate": 6.144160077226036e-05,
+      "loss": 1.4851,
+      "step": 632
+    },
+    {
+      "epoch": 0.5635432895615402,
+      "grad_norm": 1.9609206914901733,
+      "learning_rate": 6.11519253368634e-05,
+      "loss": 1.9231,
+      "step": 633
+    },
+    {
+      "epoch": 0.5644335633207211,
+      "grad_norm": 1.3588906526565552,
+      "learning_rate": 6.086263331627976e-05,
+      "loss": 1.6969,
+      "step": 634
+    },
+    {
+      "epoch": 0.5653238370799021,
+      "grad_norm": 1.5578280687332153,
+      "learning_rate": 6.05737275657049e-05,
+      "loss": 2.2163,
+      "step": 635
+    },
+    {
+      "epoch": 0.566214110839083,
+      "grad_norm": 1.2065187692642212,
+      "learning_rate": 6.0285210936521955e-05,
+      "loss": 1.6729,
+      "step": 636
+    },
+    {
+      "epoch": 0.567104384598264,
+      "grad_norm": 1.831241488456726,
+      "learning_rate": 5.999708627627354e-05,
+      "loss": 1.762,
+      "step": 637
+    },
+    {
+      "epoch": 0.5679946583574449,
+      "grad_norm": 1.7721295356750488,
+      "learning_rate": 5.9709356428633746e-05,
+      "loss": 2.003,
+      "step": 638
+    },
+    {
+      "epoch": 0.5688849321166258,
+      "grad_norm": 1.7327338457107544,
+      "learning_rate": 5.9422024233380013e-05,
+      "loss": 2.7199,
+      "step": 639
+    },
+    {
+      "epoch": 0.5697752058758068,
+      "grad_norm": 2.1094236373901367,
+      "learning_rate": 5.913509252636511e-05,
+      "loss": 2.502,
+      "step": 640
+    },
+    {
+      "epoch": 0.5706654796349877,
+      "grad_norm": 1.6638685464859009,
+      "learning_rate": 5.884856413948913e-05,
+      "loss": 1.7315,
+      "step": 641
+    },
+    {
+      "epoch": 0.5715557533941688,
+      "grad_norm": 1.3953161239624023,
+      "learning_rate": 5.856244190067159e-05,
+      "loss": 1.3959,
+      "step": 642
+    },
+    {
+      "epoch": 0.5724460271533497,
+      "grad_norm": 1.945706844329834,
+      "learning_rate": 5.82767286338235e-05,
+      "loss": 2.1033,
+      "step": 643
+    },
+    {
+      "epoch": 0.5733363009125306,
+      "grad_norm": 1.9581321477890015,
+      "learning_rate": 5.799142715881938e-05,
+      "loss": 1.8766,
+      "step": 644
+    },
+    {
+      "epoch": 0.5742265746717116,
+      "grad_norm": 1.7764853239059448,
+      "learning_rate": 5.770654029146969e-05,
+      "loss": 1.5566,
+      "step": 645
+    },
+    {
+      "epoch": 0.5751168484308925,
+      "grad_norm": 2.052767038345337,
+      "learning_rate": 5.7422070843492734e-05,
+      "loss": 2.0427,
+      "step": 646
+    },
+    {
+      "epoch": 0.5760071221900734,
+      "grad_norm": 1.8176392316818237,
+      "learning_rate": 5.713802162248718e-05,
+      "loss": 1.5934,
+      "step": 647
+    },
+    {
+      "epoch": 0.5768973959492544,
+      "grad_norm": 1.4187897443771362,
+      "learning_rate": 5.6854395431904094e-05,
+      "loss": 1.5727,
+      "step": 648
+    },
+    {
+      "epoch": 0.5777876697084353,
+      "grad_norm": 2.4474024772644043,
+      "learning_rate": 5.657119507101954e-05,
+      "loss": 2.5993,
+      "step": 649
+    },
+    {
+      "epoch": 0.5786779434676163,
+      "grad_norm": 1.395693063735962,
+      "learning_rate": 5.6288423334906735e-05,
+      "loss": 1.3948,
+      "step": 650
+    },
+    {
+      "epoch": 0.5795682172267972,
+      "grad_norm": 1.5279170274734497,
+      "learning_rate": 5.6006083014408484e-05,
+      "loss": 1.7575,
+      "step": 651
+    },
+    {
+      "epoch": 0.5804584909859782,
+      "grad_norm": 1.3539286851882935,
+      "learning_rate": 5.572417689610987e-05,
+      "loss": 1.5935,
+      "step": 652
+    },
+    {
+      "epoch": 0.5813487647451592,
+      "grad_norm": 2.212347984313965,
+      "learning_rate": 5.544270776231038e-05,
+      "loss": 1.92,
+      "step": 653
+    },
+    {
+      "epoch": 0.5822390385043401,
+      "grad_norm": 1.814328670501709,
+      "learning_rate": 5.5161678390996796e-05,
+      "loss": 1.9437,
+      "step": 654
+    },
+    {
+      "epoch": 0.583129312263521,
+      "grad_norm": 1.3090717792510986,
+      "learning_rate": 5.488109155581549e-05,
+      "loss": 1.2791,
+      "step": 655
+    },
+    {
+      "epoch": 0.584019586022702,
+      "grad_norm": 1.4268810749053955,
+      "learning_rate": 5.4600950026045326e-05,
+      "loss": 1.5508,
+      "step": 656
+    },
+    {
+      "epoch": 0.5849098597818829,
+      "grad_norm": 1.5431543588638306,
+      "learning_rate": 5.4321256566570036e-05,
+      "loss": 1.8103,
+      "step": 657
+    },
+    {
+      "epoch": 0.5858001335410639,
+      "grad_norm": 1.4491729736328125,
+      "learning_rate": 5.404201393785122e-05,
+      "loss": 1.6786,
+      "step": 658
+    },
+    {
+      "epoch": 0.5866904073002448,
+      "grad_norm": 1.7064595222473145,
+      "learning_rate": 5.3763224895900846e-05,
+      "loss": 1.7031,
+      "step": 659
+    },
+    {
+      "epoch": 0.5875806810594257,
+      "grad_norm": 1.3737152814865112,
+      "learning_rate": 5.348489219225416e-05,
+      "loss": 1.98,
+      "step": 660
+    },
+    {
+      "epoch": 0.5884709548186067,
+      "grad_norm": 1.2751985788345337,
+      "learning_rate": 5.320701857394268e-05,
+      "loss": 1.4149,
+      "step": 661
+    },
+    {
+      "epoch": 0.5893612285777877,
+      "grad_norm": 1.9529341459274292,
+      "learning_rate": 5.292960678346675e-05,
+      "loss": 2.1125,
+      "step": 662
+    },
+    {
+      "epoch": 0.5902515023369687,
+      "grad_norm": 1.7817922830581665,
+      "learning_rate": 5.265265955876879e-05,
+      "loss": 1.4712,
+      "step": 663
+    },
+    {
+      "epoch": 0.5911417760961496,
+      "grad_norm": 1.8877997398376465,
+      "learning_rate": 5.237617963320608e-05,
+      "loss": 1.5736,
+      "step": 664
+    },
+    {
+      "epoch": 0.5920320498553305,
+      "grad_norm": 2.1079890727996826,
+      "learning_rate": 5.210016973552391e-05,
+      "loss": 2.2376,
+      "step": 665
+    },
+    {
+      "epoch": 0.5929223236145115,
+      "grad_norm": 1.34657883644104,
+      "learning_rate": 5.182463258982846e-05,
+      "loss": 1.539,
+      "step": 666
+    },
+    {
+      "epoch": 0.5938125973736924,
+      "grad_norm": 1.7172969579696655,
+      "learning_rate": 5.1549570915560206e-05,
+      "loss": 1.8295,
+      "step": 667
+    },
+    {
+      "epoch": 0.5947028711328733,
+      "grad_norm": 1.6305186748504639,
+      "learning_rate": 5.127498742746675e-05,
+      "loss": 1.8357,
+      "step": 668
+    },
+    {
+      "epoch": 0.5955931448920543,
+      "grad_norm": 1.9031035900115967,
+      "learning_rate": 5.100088483557634e-05,
+      "loss": 2.1622,
+      "step": 669
+    },
+    {
+      "epoch": 0.5964834186512352,
+      "grad_norm": 1.91080904006958,
+      "learning_rate": 5.072726584517086e-05,
+      "loss": 1.9162,
+      "step": 670
+    },
+    {
+      "epoch": 0.5973736924104162,
+      "grad_norm": 1.9471015930175781,
+      "learning_rate": 5.045413315675924e-05,
+      "loss": 1.6229,
+      "step": 671
+    },
+    {
+      "epoch": 0.5982639661695972,
+      "grad_norm": 2.020172595977783,
+      "learning_rate": 5.018148946605092e-05,
+      "loss": 2.0195,
+      "step": 672
+    },
+    {
+      "epoch": 0.5991542399287781,
+      "grad_norm": 1.5960651636123657,
+      "learning_rate": 4.990933746392899e-05,
+      "loss": 1.8776,
+      "step": 673
+    },
+    {
+      "epoch": 0.6000445136879591,
+      "grad_norm": 1.7368475198745728,
+      "learning_rate": 4.9637679836423924e-05,
+      "loss": 2.228,
+      "step": 674
+    },
+    {
+      "epoch": 0.60093478744714,
+      "grad_norm": 1.6208680868148804,
+      "learning_rate": 4.9366519264686725e-05,
+      "loss": 1.2907,
+      "step": 675
+    },
+    {
+      "epoch": 0.601825061206321,
+      "grad_norm": 2.604660749435425,
+      "learning_rate": 4.909585842496287e-05,
+      "loss": 1.7746,
+      "step": 676
+    },
+    {
+      "epoch": 0.6027153349655019,
+      "grad_norm": 2.447293519973755,
+      "learning_rate": 4.8825699988565485e-05,
+      "loss": 2.4796,
+      "step": 677
+    },
+    {
+      "epoch": 0.6036056087246828,
+      "grad_norm": 1.7362377643585205,
+      "learning_rate": 4.8556046621849346e-05,
+      "loss": 1.8023,
+      "step": 678
+    },
+    {
+      "epoch": 0.6044958824838638,
+      "grad_norm": 1.647889256477356,
+      "learning_rate": 4.828690098618429e-05,
+      "loss": 2.2236,
+      "step": 679
+    },
+    {
+      "epoch": 0.6053861562430447,
+      "grad_norm": 1.913087010383606,
+      "learning_rate": 4.8018265737929044e-05,
+      "loss": 2.0162,
+      "step": 680
+    },
+    {
+      "epoch": 0.6062764300022256,
+      "grad_norm": 1.737635612487793,
+      "learning_rate": 4.7750143528405126e-05,
+      "loss": 1.5109,
+      "step": 681
+    },
+    {
+      "epoch": 0.6071667037614066,
+      "grad_norm": 1.8633228540420532,
+      "learning_rate": 4.748253700387042e-05,
+      "loss": 1.7251,
+      "step": 682
+    },
+    {
+      "epoch": 0.6080569775205876,
+      "grad_norm": 1.7534853219985962,
+      "learning_rate": 4.721544880549337e-05,
+      "loss": 1.5481,
+      "step": 683
+    },
+    {
+      "epoch": 0.6089472512797686,
+      "grad_norm": 1.9271601438522339,
+      "learning_rate": 4.694888156932658e-05,
+      "loss": 2.0399,
+      "step": 684
+    },
+    {
+      "epoch": 0.6098375250389495,
+      "grad_norm": 2.470161199569702,
+      "learning_rate": 4.668283792628114e-05,
+      "loss": 1.8926,
+      "step": 685
+    },
+    {
+      "epoch": 0.6107277987981304,
+      "grad_norm": 2.124586582183838,
+      "learning_rate": 4.6417320502100316e-05,
+      "loss": 1.98,
+      "step": 686
+    },
+    {
+      "epoch": 0.6116180725573114,
+      "grad_norm": 2.7901787757873535,
+      "learning_rate": 4.615233191733398e-05,
+      "loss": 2.2678,
+      "step": 687
+    },
+    {
+      "epoch": 0.6125083463164923,
+      "grad_norm": 2.0822958946228027,
+      "learning_rate": 4.588787478731242e-05,
+      "loss": 2.0991,
+      "step": 688
+    },
+    {
+      "epoch": 0.6133986200756733,
+      "grad_norm": 1.843925952911377,
+      "learning_rate": 4.5623951722120736e-05,
+      "loss": 1.395,
+      "step": 689
+    },
+    {
+      "epoch": 0.6142888938348542,
+      "grad_norm": 1.8190912008285522,
+      "learning_rate": 4.5360565326573104e-05,
+      "loss": 1.9312,
+      "step": 690
+    },
+    {
+      "epoch": 0.6151791675940351,
+      "grad_norm": 3.005737781524658,
+      "learning_rate": 4.5097718200186814e-05,
+      "loss": 2.49,
+      "step": 691
+    },
+    {
+      "epoch": 0.6160694413532161,
+      "grad_norm": 2.127828598022461,
+      "learning_rate": 4.483541293715698e-05,
+      "loss": 2.263,
+      "step": 692
+    },
+    {
+      "epoch": 0.6169597151123971,
+      "grad_norm": 1.2479519844055176,
+      "learning_rate": 4.457365212633058e-05,
+      "loss": 1.4489,
+      "step": 693
+    },
+    {
+      "epoch": 0.617849988871578,
+      "grad_norm": 1.4654552936553955,
+      "learning_rate": 4.431243835118124e-05,
+      "loss": 1.4409,
+      "step": 694
+    },
+    {
+      "epoch": 0.618740262630759,
+      "grad_norm": 2.2219717502593994,
+      "learning_rate": 4.4051774189783315e-05,
+      "loss": 1.7917,
+      "step": 695
+    },
+    {
+      "epoch": 0.6196305363899399,
+      "grad_norm": 1.9172980785369873,
+      "learning_rate": 4.379166221478697e-05,
+      "loss": 2.211,
+      "step": 696
+    },
+    {
+      "epoch": 0.6205208101491209,
+      "grad_norm": 1.792610764503479,
+      "learning_rate": 4.3532104993392306e-05,
+      "loss": 1.6492,
+      "step": 697
+    },
+    {
+      "epoch": 0.6214110839083018,
+      "grad_norm": 2.376288414001465,
+      "learning_rate": 4.327310508732437e-05,
+      "loss": 2.417,
+      "step": 698
+    },
+    {
+      "epoch": 0.6223013576674827,
+      "grad_norm": 3.1200907230377197,
+      "learning_rate": 4.301466505280762e-05,
+      "loss": 2.2469,
+      "step": 699
+    },
+    {
+      "epoch": 0.6231916314266637,
+      "grad_norm": 2.2198028564453125,
+      "learning_rate": 4.2756787440540936e-05,
+      "loss": 1.4221,
+      "step": 700
+    },
+    {
+      "epoch": 0.6240819051858446,
+      "grad_norm": 1.8127583265304565,
+      "learning_rate": 4.249947479567218e-05,
+      "loss": 1.9228,
+      "step": 701
+    },
+    {
+      "epoch": 0.6249721789450255,
+      "grad_norm": 1.6736197471618652,
+      "learning_rate": 4.224272965777326e-05,
+      "loss": 1.7612,
+      "step": 702
+    },
+    {
+      "epoch": 0.6258624527042066,
+      "grad_norm": 1.928165316581726,
+      "learning_rate": 4.1986554560815096e-05,
+      "loss": 2.1468,
+      "step": 703
+    },
+    {
+      "epoch": 0.6267527264633875,
+      "grad_norm": 2.0662543773651123,
+      "learning_rate": 4.173095203314241e-05,
+      "loss": 2.3367,
+      "step": 704
+    },
+    {
+      "epoch": 0.6276430002225685,
+      "grad_norm": 2.434934377670288,
+      "learning_rate": 4.1475924597449024e-05,
+      "loss": 1.6921,
+      "step": 705
+    },
+    {
+      "epoch": 0.6285332739817494,
+      "grad_norm": 2.181518316268921,
+      "learning_rate": 4.12214747707527e-05,
+      "loss": 1.8562,
+      "step": 706
+    },
+    {
+      "epoch": 0.6294235477409303,
+      "grad_norm": 1.5964933633804321,
+      "learning_rate": 4.096760506437057e-05,
+      "loss": 1.5161,
+      "step": 707
+    },
+    {
+      "epoch": 0.6303138215001113,
+      "grad_norm": 1.5785959959030151,
+      "learning_rate": 4.071431798389408e-05,
+      "loss": 1.6356,
+      "step": 708
+    },
+    {
+      "epoch": 0.6312040952592922,
+      "grad_norm": 1.6601256132125854,
+      "learning_rate": 4.0461616029164526e-05,
+      "loss": 1.9933,
+      "step": 709
+    },
+    {
+      "epoch": 0.6320943690184732,
+      "grad_norm": 1.9473941326141357,
+      "learning_rate": 4.020950169424815e-05,
+      "loss": 1.6581,
+      "step": 710
+    },
+    {
+      "epoch": 0.6329846427776541,
+      "grad_norm": 1.2505872249603271,
+      "learning_rate": 3.9957977467411615e-05,
+      "loss": 1.5246,
+      "step": 711
+    },
+    {
+      "epoch": 0.633874916536835,
+      "grad_norm": 1.948551058769226,
+      "learning_rate": 3.9707045831097555e-05,
+      "loss": 1.7574,
+      "step": 712
+    },
+    {
+      "epoch": 0.6347651902960161,
+      "grad_norm": 1.8740789890289307,
+      "learning_rate": 3.945670926189987e-05,
+      "loss": 1.7806,
+      "step": 713
+    },
+    {
+      "epoch": 0.635655464055197,
+      "grad_norm": 2.870927572250366,
+      "learning_rate": 3.920697023053949e-05,
+      "loss": 1.8539,
+      "step": 714
+    },
+    {
+      "epoch": 0.636545737814378,
+      "grad_norm": 2.207949638366699,
+      "learning_rate": 3.895783120183976e-05,
+      "loss": 1.6277,
+      "step": 715
+    },
+    {
+      "epoch": 0.6374360115735589,
+      "grad_norm": 1.841289758682251,
+      "learning_rate": 3.8709294634702376e-05,
+      "loss": 1.9111,
+      "step": 716
+    },
+    {
+      "epoch": 0.6383262853327398,
+      "grad_norm": 1.5128315687179565,
+      "learning_rate": 3.846136298208285e-05,
+      "loss": 1.8701,
+      "step": 717
+    },
+    {
+      "epoch": 0.6392165590919208,
+      "grad_norm": 2.8079261779785156,
+      "learning_rate": 3.821403869096658e-05,
+      "loss": 2.3968,
+      "step": 718
+    },
+    {
+      "epoch": 0.6401068328511017,
+      "grad_norm": 2.5751731395721436,
+      "learning_rate": 3.796732420234443e-05,
+      "loss": 2.1135,
+      "step": 719
+    },
+    {
+      "epoch": 0.6409971066102826,
+      "grad_norm": 1.8290863037109375,
+      "learning_rate": 3.7721221951188765e-05,
+      "loss": 1.5428,
+      "step": 720
+    },
+    {
+      "epoch": 0.6418873803694636,
+      "grad_norm": 1.6619089841842651,
+      "learning_rate": 3.747573436642951e-05,
+      "loss": 2.0478,
+      "step": 721
+    },
+    {
+      "epoch": 0.6427776541286445,
+      "grad_norm": 5.214348316192627,
+      "learning_rate": 3.7230863870929964e-05,
+      "loss": 2.2313,
+      "step": 722
+    },
+    {
+      "epoch": 0.6436679278878255,
+      "grad_norm": 1.9198572635650635,
+      "learning_rate": 3.698661288146311e-05,
+      "loss": 2.1648,
+      "step": 723
+    },
+    {
+      "epoch": 0.6445582016470065,
+      "grad_norm": 2.1984057426452637,
+      "learning_rate": 3.674298380868756e-05,
+      "loss": 1.95,
+      "step": 724
+    },
+    {
+      "epoch": 0.6454484754061874,
+      "grad_norm": 1.4841092824935913,
+      "learning_rate": 3.649997905712396e-05,
+      "loss": 1.6187,
+      "step": 725
+    },
+    {
+      "epoch": 0.6463387491653684,
+      "grad_norm": 2.1685807704925537,
+      "learning_rate": 3.6257601025131026e-05,
+      "loss": 2.5758,
+      "step": 726
+    },
+    {
+      "epoch": 0.6472290229245493,
+      "grad_norm": 3.2427711486816406,
+      "learning_rate": 3.601585210488218e-05,
+      "loss": 2.4056,
+      "step": 727
+    },
+    {
+      "epoch": 0.6481192966837303,
+      "grad_norm": 1.696851134300232,
+      "learning_rate": 3.577473468234156e-05,
+      "loss": 1.7485,
+      "step": 728
+    },
+    {
+      "epoch": 0.6490095704429112,
+      "grad_norm": 2.0988123416900635,
+      "learning_rate": 3.553425113724088e-05,
+      "loss": 1.894,
+      "step": 729
+    },
+    {
+      "epoch": 0.6498998442020921,
+      "grad_norm": 1.6171239614486694,
+      "learning_rate": 3.52944038430556e-05,
+      "loss": 1.6306,
+      "step": 730
+    },
+    {
+      "epoch": 0.6507901179612731,
+      "grad_norm": 2.2144834995269775,
+      "learning_rate": 3.5055195166981645e-05,
+      "loss": 2.1848,
+      "step": 731
+    },
+    {
+      "epoch": 0.651680391720454,
+      "grad_norm": 1.90192711353302,
+      "learning_rate": 3.481662746991214e-05,
+      "loss": 2.0253,
+      "step": 732
+    },
+    {
+      "epoch": 0.6525706654796349,
+      "grad_norm": 1.956634283065796,
+      "learning_rate": 3.4578703106413904e-05,
+      "loss": 1.7608,
+      "step": 733
+    },
+    {
+      "epoch": 0.653460939238816,
+      "grad_norm": 1.6562048196792603,
+      "learning_rate": 3.4341424424704375e-05,
+      "loss": 1.828,
+      "step": 734
+    },
+    {
+      "epoch": 0.6543512129979969,
+      "grad_norm": 2.268554925918579,
+      "learning_rate": 3.4104793766628304e-05,
+      "loss": 1.9865,
+      "step": 735
+    },
+    {
+      "epoch": 0.6552414867571779,
+      "grad_norm": 2.734750986099243,
+      "learning_rate": 3.386881346763483e-05,
+      "loss": 1.8182,
+      "step": 736
+    },
+    {
+      "epoch": 0.6561317605163588,
+      "grad_norm": 1.684584379196167,
+      "learning_rate": 3.363348585675414e-05,
+      "loss": 1.4571,
+      "step": 737
+    },
+    {
+      "epoch": 0.6570220342755397,
+      "grad_norm": 1.4472819566726685,
+      "learning_rate": 3.339881325657484e-05,
+      "loss": 1.2802,
+      "step": 738
+    },
+    {
+      "epoch": 0.6579123080347207,
+      "grad_norm": 2.2351865768432617,
+      "learning_rate": 3.316479798322072e-05,
+      "loss": 1.5775,
+      "step": 739
+    },
+    {
+      "epoch": 0.6588025817939016,
+      "grad_norm": 2.1901495456695557,
+      "learning_rate": 3.2931442346328004e-05,
+      "loss": 1.9047,
+      "step": 740
+    },
+    {
+      "epoch": 0.6596928555530825,
+      "grad_norm": 2.646942377090454,
+      "learning_rate": 3.269874864902269e-05,
+      "loss": 2.0635,
+      "step": 741
+    },
+    {
+      "epoch": 0.6605831293122635,
+      "grad_norm": 2.0031111240386963,
+      "learning_rate": 3.246671918789755e-05,
+      "loss": 1.6451,
+      "step": 742
+    },
+    {
+      "epoch": 0.6614734030714444,
+      "grad_norm": 1.944433569908142,
+      "learning_rate": 3.223535625298979e-05,
+      "loss": 2.0588,
+      "step": 743
+    },
+    {
+      "epoch": 0.6623636768306255,
+      "grad_norm": 2.4715232849121094,
+      "learning_rate": 3.200466212775808e-05,
+      "loss": 2.2318,
+      "step": 744
+    },
+    {
+      "epoch": 0.6632539505898064,
+      "grad_norm": 1.5102177858352661,
+      "learning_rate": 3.1774639089060363e-05,
+      "loss": 1.6319,
+      "step": 745
+    },
+    {
+      "epoch": 0.6641442243489873,
+      "grad_norm": 1.6239656209945679,
+      "learning_rate": 3.154528940713113e-05,
+      "loss": 2.011,
+      "step": 746
+    },
+    {
+      "epoch": 0.6650344981081683,
+      "grad_norm": 1.6042510271072388,
+      "learning_rate": 3.1316615345559185e-05,
+      "loss": 1.6868,
+      "step": 747
+    },
+    {
+      "epoch": 0.6659247718673492,
+      "grad_norm": 1.2276384830474854,
+      "learning_rate": 3.108861916126518e-05,
+      "loss": 1.1603,
+      "step": 748
+    },
+    {
+      "epoch": 0.6668150456265302,
+      "grad_norm": 1.999118685722351,
+      "learning_rate": 3.086130310447937e-05,
+      "loss": 1.8479,
+      "step": 749
+    },
+    {
+      "epoch": 0.6677053193857111,
+      "grad_norm": 1.7513126134872437,
+      "learning_rate": 3.063466941871952e-05,
+      "loss": 2.1436,
+      "step": 750
+    },
+    {
+      "epoch": 0.668595593144892,
+      "grad_norm": 2.9473211765289307,
+      "learning_rate": 3.0408720340768572e-05,
+      "loss": 2.5041,
+      "step": 751
+    },
+    {
+      "epoch": 0.669485866904073,
+      "grad_norm": 1.2190455198287964,
+      "learning_rate": 3.018345810065275e-05,
+      "loss": 1.5207,
+      "step": 752
+    },
+    {
+      "epoch": 0.6703761406632539,
+      "grad_norm": 1.731210470199585,
+      "learning_rate": 2.9958884921619367e-05,
+      "loss": 1.4926,
+      "step": 753
+    },
+    {
+      "epoch": 0.671266414422435,
+      "grad_norm": 1.8084713220596313,
+      "learning_rate": 2.9735003020115092e-05,
+      "loss": 1.7094,
+      "step": 754
+    },
+    {
+      "epoch": 0.6721566881816159,
+      "grad_norm": 2.106186866760254,
+      "learning_rate": 2.9511814605763855e-05,
+      "loss": 2.2084,
+      "step": 755
+    },
+    {
+      "epoch": 0.6730469619407968,
+      "grad_norm": 1.5817265510559082,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 1.2095,
+      "step": 756
+    },
+    {
+      "epoch": 0.6739372356999778,
+      "grad_norm": 2.0624277591705322,
+      "learning_rate": 2.9067527042772636e-05,
+      "loss": 1.7659,
+      "step": 757
+    },
+    {
+      "epoch": 0.6748275094591587,
+      "grad_norm": 1.7278014421463013,
+      "learning_rate": 2.8846432279071467e-05,
+      "loss": 2.2494,
+      "step": 758
+    },
+    {
+      "epoch": 0.6757177832183396,
+      "grad_norm": 1.3449331521987915,
+      "learning_rate": 2.8626039772357882e-05,
+      "loss": 1.503,
+      "step": 759
+    },
+    {
+      "epoch": 0.6766080569775206,
+      "grad_norm": 2.799799680709839,
+      "learning_rate": 2.840635169781688e-05,
+      "loss": 2.1368,
+      "step": 760
+    },
+    {
+      "epoch": 0.6774983307367015,
+      "grad_norm": 2.179213523864746,
+      "learning_rate": 2.8187370223681132e-05,
+      "loss": 2.0393,
+      "step": 761
+    },
+    {
+      "epoch": 0.6783886044958825,
+      "grad_norm": 1.5615414381027222,
+      "learning_rate": 2.7969097511209308e-05,
+      "loss": 1.4168,
+      "step": 762
+    },
+    {
+      "epoch": 0.6792788782550634,
+      "grad_norm": 2.0087335109710693,
+      "learning_rate": 2.775153571466502e-05,
+      "loss": 1.7341,
+      "step": 763
+    },
+    {
+      "epoch": 0.6801691520142443,
+      "grad_norm": 1.9231607913970947,
+      "learning_rate": 2.753468698129533e-05,
+      "loss": 1.9537,
+      "step": 764
+    },
+    {
+      "epoch": 0.6810594257734254,
+      "grad_norm": 2.165048599243164,
+      "learning_rate": 2.7318553451309726e-05,
+      "loss": 1.6802,
+      "step": 765
+    },
+    {
+      "epoch": 0.6819496995326063,
+      "grad_norm": 1.7000216245651245,
+      "learning_rate": 2.7103137257858868e-05,
+      "loss": 1.5879,
+      "step": 766
+    },
+    {
+      "epoch": 0.6828399732917872,
+      "grad_norm": 2.1674106121063232,
+      "learning_rate": 2.688844052701359e-05,
+      "loss": 2.4558,
+      "step": 767
+    },
+    {
+      "epoch": 0.6837302470509682,
+      "grad_norm": 2.2125790119171143,
+      "learning_rate": 2.6674465377744017e-05,
+      "loss": 1.5141,
+      "step": 768
+    },
+    {
+      "epoch": 0.6846205208101491,
+      "grad_norm": 1.5611724853515625,
+      "learning_rate": 2.646121392189841e-05,
+      "loss": 1.3355,
+      "step": 769
+    },
+    {
+      "epoch": 0.6855107945693301,
+      "grad_norm": 1.870052695274353,
+      "learning_rate": 2.624868826418262e-05,
+      "loss": 1.287,
+      "step": 770
+    },
+    {
+      "epoch": 0.686401068328511,
+      "grad_norm": 1.8563505411148071,
+      "learning_rate": 2.603689050213902e-05,
+      "loss": 2.1767,
+      "step": 771
+    },
+    {
+      "epoch": 0.6872913420876919,
+      "grad_norm": 1.7890762090682983,
+      "learning_rate": 2.582582272612609e-05,
+      "loss": 1.7168,
+      "step": 772
+    },
+    {
+      "epoch": 0.6881816158468729,
+      "grad_norm": 1.420925498008728,
+      "learning_rate": 2.561548701929749e-05,
+      "loss": 1.602,
+      "step": 773
+    },
+    {
+      "epoch": 0.6890718896060538,
+      "grad_norm": 2.628692865371704,
+      "learning_rate": 2.540588545758179e-05,
+      "loss": 1.9804,
+      "step": 774
+    },
+    {
+      "epoch": 0.6899621633652349,
+      "grad_norm": 2.270979404449463,
+      "learning_rate": 2.5197020109661772e-05,
+      "loss": 2.0805,
+      "step": 775
+    },
+    {
+      "epoch": 0.6908524371244158,
+      "grad_norm": 1.3668320178985596,
+      "learning_rate": 2.4988893036954043e-05,
+      "loss": 1.3515,
+      "step": 776
+    },
+    {
+      "epoch": 0.6917427108835967,
+      "grad_norm": 2.867128849029541,
+      "learning_rate": 2.4781506293588873e-05,
+      "loss": 1.9102,
+      "step": 777
+    },
+    {
+      "epoch": 0.6926329846427777,
+      "grad_norm": 2.0665462017059326,
+      "learning_rate": 2.4574861926389615e-05,
+      "loss": 1.5223,
+      "step": 778
+    },
+    {
+      "epoch": 0.6935232584019586,
+      "grad_norm": 1.425742506980896,
+      "learning_rate": 2.436896197485282e-05,
+      "loss": 1.3977,
+      "step": 779
+    },
+    {
+      "epoch": 0.6944135321611395,
+      "grad_norm": 1.6747236251831055,
+      "learning_rate": 2.4163808471127812e-05,
+      "loss": 1.7501,
+      "step": 780
+    },
+    {
+      "epoch": 0.6953038059203205,
+      "grad_norm": 1.8426623344421387,
+      "learning_rate": 2.3959403439996907e-05,
+      "loss": 1.8335,
+      "step": 781
+    },
+    {
+      "epoch": 0.6961940796795014,
+      "grad_norm": 2.007152795791626,
+      "learning_rate": 2.37557488988552e-05,
+      "loss": 2.4554,
+      "step": 782
+    },
+    {
+      "epoch": 0.6970843534386824,
+      "grad_norm": 2.151026964187622,
+      "learning_rate": 2.3552846857690846e-05,
+      "loss": 2.0088,
+      "step": 783
+    },
+    {
+      "epoch": 0.6979746271978633,
+      "grad_norm": 1.486572027206421,
+      "learning_rate": 2.3350699319065026e-05,
+      "loss": 1.6484,
+      "step": 784
+    },
+    {
+      "epoch": 0.6988649009570443,
+      "grad_norm": 2.3491599559783936,
+      "learning_rate": 2.3149308278092342e-05,
+      "loss": 2.1194,
+      "step": 785
+    },
+    {
+      "epoch": 0.6997551747162253,
+      "grad_norm": 2.2394917011260986,
+      "learning_rate": 2.2948675722421086e-05,
+      "loss": 1.646,
+      "step": 786
+    },
+    {
+      "epoch": 0.7006454484754062,
+      "grad_norm": 1.754228949546814,
+      "learning_rate": 2.2748803632213557e-05,
+      "loss": 2.0966,
+      "step": 787
+    },
+    {
+      "epoch": 0.7015357222345872,
+      "grad_norm": 2.0761775970458984,
+      "learning_rate": 2.254969398012663e-05,
+      "loss": 1.7768,
+      "step": 788
+    },
+    {
+      "epoch": 0.7024259959937681,
+      "grad_norm": 1.6214390993118286,
+      "learning_rate": 2.235134873129213e-05,
+      "loss": 1.799,
+      "step": 789
+    },
+    {
+      "epoch": 0.703316269752949,
+      "grad_norm": 2.5304348468780518,
+      "learning_rate": 2.2153769843297667e-05,
+      "loss": 2.4849,
+      "step": 790
+    },
+    {
+      "epoch": 0.70420654351213,
+      "grad_norm": 1.9643628597259521,
+      "learning_rate": 2.195695926616702e-05,
+      "loss": 1.5525,
+      "step": 791
+    },
+    {
+      "epoch": 0.7050968172713109,
+      "grad_norm": 2.729534149169922,
+      "learning_rate": 2.1760918942341192e-05,
+      "loss": 2.0789,
+      "step": 792
+    },
+    {
+      "epoch": 0.7059870910304918,
+      "grad_norm": 1.8052536249160767,
+      "learning_rate": 2.1565650806658975e-05,
+      "loss": 1.6703,
+      "step": 793
+    },
+    {
+      "epoch": 0.7068773647896728,
+      "grad_norm": 2.2261719703674316,
+      "learning_rate": 2.137115678633811e-05,
+      "loss": 2.0943,
+      "step": 794
+    },
+    {
+      "epoch": 0.7077676385488538,
+      "grad_norm": 1.9751397371292114,
+      "learning_rate": 2.1177438800956007e-05,
+      "loss": 2.0637,
+      "step": 795
+    },
+    {
+      "epoch": 0.7086579123080348,
+      "grad_norm": 1.6201308965682983,
+      "learning_rate": 2.098449876243096e-05,
+      "loss": 1.8145,
+      "step": 796
+    },
+    {
+      "epoch": 0.7095481860672157,
+      "grad_norm": 3.4927635192871094,
+      "learning_rate": 2.07923385750033e-05,
+      "loss": 1.2709,
+      "step": 797
+    },
+    {
+      "epoch": 0.7104384598263966,
+      "grad_norm": 1.7658709287643433,
+      "learning_rate": 2.0600960135216462e-05,
+      "loss": 2.021,
+      "step": 798
+    },
+    {
+      "epoch": 0.7113287335855776,
+      "grad_norm": 1.4747573137283325,
+      "learning_rate": 2.0410365331898416e-05,
+      "loss": 1.425,
+      "step": 799
+    },
+    {
+      "epoch": 0.7122190073447585,
+      "grad_norm": 2.7042531967163086,
+      "learning_rate": 2.0220556046142893e-05,
+      "loss": 1.6427,
+      "step": 800
+    },
+    {
+      "epoch": 0.7131092811039395,
+      "grad_norm": 3.2484536170959473,
+      "learning_rate": 2.0031534151290943e-05,
+      "loss": 3.4625,
+      "step": 801
+    },
+    {
+      "epoch": 0.7139995548631204,
+      "grad_norm": 2.409848928451538,
+      "learning_rate": 1.9843301512912327e-05,
+      "loss": 1.8531,
+      "step": 802
+    },
+    {
+      "epoch": 0.7148898286223013,
+      "grad_norm": 1.7639962434768677,
+      "learning_rate": 1.965585998878724e-05,
+      "loss": 2.042,
+      "step": 803
+    },
+    {
+      "epoch": 0.7157801023814823,
+      "grad_norm": 1.6976604461669922,
+      "learning_rate": 1.946921142888781e-05,
+      "loss": 1.6346,
+      "step": 804
+    },
+    {
+      "epoch": 0.7166703761406632,
+      "grad_norm": 2.473877191543579,
+      "learning_rate": 1.928335767535997e-05,
+      "loss": 2.1879,
+      "step": 805
+    },
+    {
+      "epoch": 0.7175606498998442,
+      "grad_norm": 2.551295518875122,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 1.6309,
+      "step": 806
+    },
+    {
+      "epoch": 0.7184509236590252,
+      "grad_norm": 1.5652191638946533,
+      "learning_rate": 1.891404191676265e-05,
+      "loss": 1.7005,
+      "step": 807
+    },
+    {
+      "epoch": 0.7193411974182061,
+      "grad_norm": 2.0537121295928955,
+      "learning_rate": 1.8730583556690605e-05,
+      "loss": 1.694,
+      "step": 808
+    },
+    {
+      "epoch": 0.7202314711773871,
+      "grad_norm": 1.8218454122543335,
+      "learning_rate": 1.854792729294905e-05,
+      "loss": 1.7962,
+      "step": 809
+    },
+    {
+      "epoch": 0.721121744936568,
+      "grad_norm": 2.0948023796081543,
+      "learning_rate": 1.8366074928281607e-05,
+      "loss": 1.8436,
+      "step": 810
+    },
+    {
+      "epoch": 0.7220120186957489,
+      "grad_norm": 1.4422553777694702,
+      "learning_rate": 1.818502825749764e-05,
+      "loss": 1.4781,
+      "step": 811
+    },
+    {
+      "epoch": 0.7229022924549299,
+      "grad_norm": 2.543961763381958,
+      "learning_rate": 1.8004789067454764e-05,
+      "loss": 1.8586,
+      "step": 812
+    },
+    {
+      "epoch": 0.7237925662141108,
+      "grad_norm": 1.2190775871276855,
+      "learning_rate": 1.7825359137040988e-05,
+      "loss": 1.0705,
+      "step": 813
+    },
+    {
+      "epoch": 0.7246828399732917,
+      "grad_norm": 2.0991899967193604,
+      "learning_rate": 1.7646740237157256e-05,
+      "loss": 2.5025,
+      "step": 814
+    },
+    {
+      "epoch": 0.7255731137324727,
+      "grad_norm": 1.7105249166488647,
+      "learning_rate": 1.7468934130700044e-05,
+      "loss": 1.6559,
+      "step": 815
+    },
+    {
+      "epoch": 0.7264633874916537,
+      "grad_norm": 1.533126711845398,
+      "learning_rate": 1.7291942572543807e-05,
+      "loss": 1.6451,
+      "step": 816
+    },
+    {
+      "epoch": 0.7273536612508347,
+      "grad_norm": 2.375067949295044,
+      "learning_rate": 1.7115767309523812e-05,
+      "loss": 2.1343,
+      "step": 817
+    },
+    {
+      "epoch": 0.7282439350100156,
+      "grad_norm": 1.9037030935287476,
+      "learning_rate": 1.6940410080418723e-05,
+      "loss": 1.8725,
+      "step": 818
+    },
+    {
+      "epoch": 0.7291342087691965,
+      "grad_norm": 2.0840301513671875,
+      "learning_rate": 1.6765872615933677e-05,
+      "loss": 2.1541,
+      "step": 819
+    },
+    {
+      "epoch": 0.7300244825283775,
+      "grad_norm": 2.6141037940979004,
+      "learning_rate": 1.6592156638682886e-05,
+      "loss": 2.3102,
+      "step": 820
+    },
+    {
+      "epoch": 0.7309147562875584,
+      "grad_norm": 1.8828359842300415,
+      "learning_rate": 1.6419263863172997e-05,
+      "loss": 1.9035,
+      "step": 821
+    },
+    {
+      "epoch": 0.7318050300467394,
+      "grad_norm": 1.8121448755264282,
+      "learning_rate": 1.6247195995785837e-05,
+      "loss": 1.6091,
+      "step": 822
+    },
+    {
+      "epoch": 0.7326953038059203,
+      "grad_norm": 2.16495418548584,
+      "learning_rate": 1.6075954734761845e-05,
+      "loss": 1.7576,
+      "step": 823
+    },
+    {
+      "epoch": 0.7335855775651012,
+      "grad_norm": 1.4242593050003052,
+      "learning_rate": 1.5905541770183096e-05,
+      "loss": 1.5635,
+      "step": 824
+    },
+    {
+      "epoch": 0.7344758513242822,
+      "grad_norm": 2.510634183883667,
+      "learning_rate": 1.5735958783956794e-05,
+      "loss": 1.7408,
+      "step": 825
+    },
+    {
+      "epoch": 0.7353661250834632,
+      "grad_norm": 1.7546478509902954,
+      "learning_rate": 1.5567207449798515e-05,
+      "loss": 1.5751,
+      "step": 826
+    },
+    {
+      "epoch": 0.7362563988426442,
+      "grad_norm": 1.723543643951416,
+      "learning_rate": 1.539928943321579e-05,
+      "loss": 1.4336,
+      "step": 827
+    },
+    {
+      "epoch": 0.7371466726018251,
+      "grad_norm": 1.7334325313568115,
+      "learning_rate": 1.5232206391491699e-05,
+      "loss": 1.7615,
+      "step": 828
+    },
+    {
+      "epoch": 0.738036946361006,
+      "grad_norm": 1.8361815214157104,
+      "learning_rate": 1.5065959973668353e-05,
+      "loss": 1.9695,
+      "step": 829
+    },
+    {
+      "epoch": 0.738927220120187,
+      "grad_norm": 1.947911262512207,
+      "learning_rate": 1.4900551820530828e-05,
+      "loss": 1.851,
+      "step": 830
+    },
+    {
+      "epoch": 0.7398174938793679,
+      "grad_norm": 1.4021551609039307,
+      "learning_rate": 1.4735983564590783e-05,
+      "loss": 1.845,
+      "step": 831
+    },
+    {
+      "epoch": 0.7407077676385488,
+      "grad_norm": 1.4573607444763184,
+      "learning_rate": 1.4572256830070497e-05,
+      "loss": 1.4406,
+      "step": 832
+    },
+    {
+      "epoch": 0.7415980413977298,
+      "grad_norm": 1.9009228944778442,
+      "learning_rate": 1.4409373232886702e-05,
+      "loss": 2.1688,
+      "step": 833
+    },
+    {
+      "epoch": 0.7424883151569107,
+      "grad_norm": 2.001500129699707,
+      "learning_rate": 1.4247334380634792e-05,
+      "loss": 2.0726,
+      "step": 834
+    },
+    {
+      "epoch": 0.7433785889160917,
+      "grad_norm": 2.0300889015197754,
+      "learning_rate": 1.4086141872572789e-05,
+      "loss": 1.8231,
+      "step": 835
+    },
+    {
+      "epoch": 0.7442688626752726,
+      "grad_norm": 12.630093574523926,
+      "learning_rate": 1.3925797299605647e-05,
+      "loss": 1.6159,
+      "step": 836
+    },
+    {
+      "epoch": 0.7451591364344536,
+      "grad_norm": 2.9155776500701904,
+      "learning_rate": 1.3766302244269624e-05,
+      "loss": 1.6208,
+      "step": 837
+    },
+    {
+      "epoch": 0.7460494101936346,
+      "grad_norm": 11.810160636901855,
+      "learning_rate": 1.3607658280716473e-05,
+      "loss": 1.6429,
+      "step": 838
+    },
+    {
+      "epoch": 0.7469396839528155,
+      "grad_norm": 2.091851234436035,
+      "learning_rate": 1.3449866974698122e-05,
+      "loss": 1.751,
+      "step": 839
+    },
+    {
+      "epoch": 0.7478299577119965,
+      "grad_norm": 1.2974621057510376,
+      "learning_rate": 1.3292929883550998e-05,
+      "loss": 1.5679,
+      "step": 840
+    },
+    {
+      "epoch": 0.7487202314711774,
+      "grad_norm": 1.8211385011672974,
+      "learning_rate": 1.3136848556180892e-05,
+      "loss": 1.3824,
+      "step": 841
+    },
+    {
+      "epoch": 0.7496105052303583,
+      "grad_norm": 1.7687885761260986,
+      "learning_rate": 1.2981624533047432e-05,
+      "loss": 1.7996,
+      "step": 842
+    },
+    {
+      "epoch": 0.7505007789895393,
+      "grad_norm": 2.209022283554077,
+      "learning_rate": 1.2827259346149122e-05,
+      "loss": 1.6272,
+      "step": 843
+    },
+    {
+      "epoch": 0.7513910527487202,
+      "grad_norm": 2.4203338623046875,
+      "learning_rate": 1.2673754519008008e-05,
+      "loss": 1.564,
+      "step": 844
+    },
+    {
+      "epoch": 0.7522813265079011,
+      "grad_norm": 2.287649393081665,
+      "learning_rate": 1.2521111566654731e-05,
+      "loss": 1.815,
+      "step": 845
+    },
+    {
+      "epoch": 0.7531716002670821,
+      "grad_norm": 2.2657222747802734,
+      "learning_rate": 1.2369331995613665e-05,
+      "loss": 2.0586,
+      "step": 846
+    },
+    {
+      "epoch": 0.7540618740262631,
+      "grad_norm": 1.5107150077819824,
+      "learning_rate": 1.2218417303887842e-05,
+      "loss": 1.543,
+      "step": 847
+    },
+    {
+      "epoch": 0.7549521477854441,
+      "grad_norm": 1.3881908655166626,
+      "learning_rate": 1.206836898094439e-05,
+      "loss": 1.3387,
+      "step": 848
+    },
+    {
+      "epoch": 0.755842421544625,
+      "grad_norm": 2.1925411224365234,
+      "learning_rate": 1.191918850769964e-05,
+      "loss": 1.5979,
+      "step": 849
+    },
+    {
+      "epoch": 0.7567326953038059,
+      "grad_norm": 1.8776897192001343,
+      "learning_rate": 1.1770877356504683e-05,
+      "loss": 1.7517,
+      "step": 850
+    },
+    {
+      "epoch": 0.7576229690629869,
+      "grad_norm": 2.166015625,
+      "learning_rate": 1.1623436991130654e-05,
+      "loss": 1.8777,
+      "step": 851
+    },
+    {
+      "epoch": 0.7585132428221678,
+      "grad_norm": 1.5148497819900513,
+      "learning_rate": 1.1476868866754486e-05,
+      "loss": 1.9234,
+      "step": 852
+    },
+    {
+      "epoch": 0.7594035165813487,
+      "grad_norm": 2.4111993312835693,
+      "learning_rate": 1.1331174429944347e-05,
+      "loss": 1.4803,
+      "step": 853
+    },
+    {
+      "epoch": 0.7602937903405297,
+      "grad_norm": 1.7421929836273193,
+      "learning_rate": 1.1186355118645554e-05,
+      "loss": 1.3535,
+      "step": 854
+    },
+    {
+      "epoch": 0.7611840640997106,
+      "grad_norm": 1.5131878852844238,
+      "learning_rate": 1.1042412362166222e-05,
+      "loss": 1.7501,
+      "step": 855
+    },
+    {
+      "epoch": 0.7620743378588916,
+      "grad_norm": 6.059530735015869,
+      "learning_rate": 1.0899347581163221e-05,
+      "loss": 1.743,
+      "step": 856
+    },
+    {
+      "epoch": 0.7629646116180726,
+      "grad_norm": 2.317833423614502,
+      "learning_rate": 1.0757162187628222e-05,
+      "loss": 2.0051,
+      "step": 857
+    },
+    {
+      "epoch": 0.7638548853772535,
+      "grad_norm": 1.7132494449615479,
+      "learning_rate": 1.0615857584873623e-05,
+      "loss": 2.1177,
+      "step": 858
+    },
+    {
+      "epoch": 0.7647451591364345,
+      "grad_norm": 2.202890634536743,
+      "learning_rate": 1.0475435167518843e-05,
+      "loss": 1.9601,
+      "step": 859
+    },
+    {
+      "epoch": 0.7656354328956154,
+      "grad_norm": 3.0330166816711426,
+      "learning_rate": 1.0335896321476413e-05,
+      "loss": 1.5331,
+      "step": 860
+    },
+    {
+      "epoch": 0.7665257066547964,
+      "grad_norm": 1.7549813985824585,
+      "learning_rate": 1.0197242423938446e-05,
+      "loss": 1.7902,
+      "step": 861
+    },
+    {
+      "epoch": 0.7674159804139773,
+      "grad_norm": 2.6962575912475586,
+      "learning_rate": 1.0059474843362892e-05,
+      "loss": 2.384,
+      "step": 862
+    },
+    {
+      "epoch": 0.7683062541731582,
+      "grad_norm": 1.2838563919067383,
+      "learning_rate": 9.922594939460194e-06,
+      "loss": 1.1263,
+      "step": 863
+    },
+    {
+      "epoch": 0.7691965279323392,
+      "grad_norm": 1.1030938625335693,
+      "learning_rate": 9.786604063179728e-06,
+      "loss": 1.3106,
+      "step": 864
+    },
+    {
+      "epoch": 0.7700868016915201,
+      "grad_norm": 2.963934898376465,
+      "learning_rate": 9.651503556696516e-06,
+      "loss": 1.7046,
+      "step": 865
+    },
+    {
+      "epoch": 0.770977075450701,
+      "grad_norm": 1.6622592210769653,
+      "learning_rate": 9.517294753398064e-06,
+      "loss": 1.9185,
+      "step": 866
+    },
+    {
+      "epoch": 0.7718673492098821,
+      "grad_norm": 1.9364992380142212,
+      "learning_rate": 9.383978977871021e-06,
+      "loss": 1.5313,
+      "step": 867
+    },
+    {
+      "epoch": 0.772757622969063,
+      "grad_norm": 1.8039541244506836,
+      "learning_rate": 9.251557545888312e-06,
+      "loss": 1.9076,
+      "step": 868
+    },
+    {
+      "epoch": 0.773647896728244,
+      "grad_norm": 1.801265001296997,
+      "learning_rate": 9.120031764395987e-06,
+      "loss": 1.9091,
+      "step": 869
+    },
+    {
+      "epoch": 0.7745381704874249,
+      "grad_norm": 1.7942633628845215,
+      "learning_rate": 8.989402931500434e-06,
+      "loss": 1.9505,
+      "step": 870
+    },
+    {
+      "epoch": 0.7754284442466058,
+      "grad_norm": 1.9900010824203491,
+      "learning_rate": 8.85967233645547e-06,
+      "loss": 1.7563,
+      "step": 871
+    },
+    {
+      "epoch": 0.7763187180057868,
+      "grad_norm": 1.809123158454895,
+      "learning_rate": 8.730841259649725e-06,
+      "loss": 1.6804,
+      "step": 872
+    },
+    {
+      "epoch": 0.7772089917649677,
+      "grad_norm": 1.3528798818588257,
+      "learning_rate": 8.602910972593892e-06,
+      "loss": 1.6276,
+      "step": 873
+    },
+    {
+      "epoch": 0.7780992655241487,
+      "grad_norm": 3.139510154724121,
+      "learning_rate": 8.475882737908248e-06,
+      "loss": 2.2761,
+      "step": 874
+    },
+    {
+      "epoch": 0.7789895392833296,
+      "grad_norm": 2.2909891605377197,
+      "learning_rate": 8.34975780931021e-06,
+      "loss": 2.1253,
+      "step": 875
+    },
+    {
+      "epoch": 0.7798798130425105,
+      "grad_norm": 1.4801619052886963,
+      "learning_rate": 8.224537431601886e-06,
+      "loss": 1.5527,
+      "step": 876
+    },
+    {
+      "epoch": 0.7807700868016915,
+      "grad_norm": 1.4977777004241943,
+      "learning_rate": 8.100222840657878e-06,
+      "loss": 1.4283,
+      "step": 877
+    },
+    {
+      "epoch": 0.7816603605608725,
+      "grad_norm": 1.879576325416565,
+      "learning_rate": 7.976815263412963e-06,
+      "loss": 1.6929,
+      "step": 878
+    },
+    {
+      "epoch": 0.7825506343200535,
+      "grad_norm": 1.6205519437789917,
+      "learning_rate": 7.854315917850163e-06,
+      "loss": 1.8613,
+      "step": 879
+    },
+    {
+      "epoch": 0.7834409080792344,
+      "grad_norm": 1.9095968008041382,
+      "learning_rate": 7.73272601298851e-06,
+      "loss": 1.9905,
+      "step": 880
+    },
+    {
+      "epoch": 0.7843311818384153,
+      "grad_norm": 1.1806037425994873,
+      "learning_rate": 7.612046748871327e-06,
+      "loss": 1.4975,
+      "step": 881
+    },
+    {
+      "epoch": 0.7852214555975963,
+      "grad_norm": 1.2940500974655151,
+      "learning_rate": 7.492279316554207e-06,
+      "loss": 1.6006,
+      "step": 882
+    },
+    {
+      "epoch": 0.7861117293567772,
+      "grad_norm": 1.8334342241287231,
+      "learning_rate": 7.3734248980933395e-06,
+      "loss": 1.3934,
+      "step": 883
+    },
+    {
+      "epoch": 0.7870020031159581,
+      "grad_norm": 2.4100828170776367,
+      "learning_rate": 7.255484666533874e-06,
+      "loss": 1.692,
+      "step": 884
+    },
+    {
+      "epoch": 0.7878922768751391,
+      "grad_norm": 3.7759780883789062,
+      "learning_rate": 7.138459785898266e-06,
+      "loss": 1.9625,
+      "step": 885
+    },
+    {
+      "epoch": 0.78878255063432,
+      "grad_norm": 2.20845365524292,
+      "learning_rate": 7.022351411174866e-06,
+      "loss": 2.6353,
+      "step": 886
+    },
+    {
+      "epoch": 0.789672824393501,
+      "grad_norm": 1.8291293382644653,
+      "learning_rate": 6.907160688306425e-06,
+      "loss": 1.5929,
+      "step": 887
+    },
+    {
+      "epoch": 0.790563098152682,
+      "grad_norm": 2.1727707386016846,
+      "learning_rate": 6.7928887541789055e-06,
+      "loss": 2.1755,
+      "step": 888
+    },
+    {
+      "epoch": 0.7914533719118629,
+      "grad_norm": 1.7652902603149414,
+      "learning_rate": 6.679536736610137e-06,
+      "loss": 1.5469,
+      "step": 889
+    },
+    {
+      "epoch": 0.7923436456710439,
+      "grad_norm": 1.6923081874847412,
+      "learning_rate": 6.5671057543387985e-06,
+      "loss": 1.4707,
+      "step": 890
+    },
+    {
+      "epoch": 0.7932339194302248,
+      "grad_norm": 1.6656982898712158,
+      "learning_rate": 6.455596917013273e-06,
+      "loss": 1.498,
+      "step": 891
+    },
+    {
+      "epoch": 0.7941241931894057,
+      "grad_norm": 1.6052128076553345,
+      "learning_rate": 6.345011325180772e-06,
+      "loss": 1.6739,
+      "step": 892
+    },
+    {
+      "epoch": 0.7950144669485867,
+      "grad_norm": 2.754615306854248,
+      "learning_rate": 6.235350070276447e-06,
+      "loss": 2.0599,
+      "step": 893
+    },
+    {
+      "epoch": 0.7959047407077676,
+      "grad_norm": 1.7544282674789429,
+      "learning_rate": 6.126614234612593e-06,
+      "loss": 1.7979,
+      "step": 894
+    },
+    {
+      "epoch": 0.7967950144669486,
+      "grad_norm": 1.9163486957550049,
+      "learning_rate": 6.018804891368035e-06,
+      "loss": 1.7769,
+      "step": 895
+    },
+    {
+      "epoch": 0.7976852882261295,
+      "grad_norm": 2.594054937362671,
+      "learning_rate": 5.911923104577455e-06,
+      "loss": 2.1903,
+      "step": 896
+    },
+    {
+      "epoch": 0.7985755619853104,
+      "grad_norm": 2.1390740871429443,
+      "learning_rate": 5.805969929120947e-06,
+      "loss": 2.3515,
+      "step": 897
+    },
+    {
+      "epoch": 0.7994658357444915,
+      "grad_norm": 2.4581687450408936,
+      "learning_rate": 5.700946410713548e-06,
+      "loss": 1.7991,
+      "step": 898
+    },
+    {
+      "epoch": 0.8003561095036724,
+      "grad_norm": 1.9051916599273682,
+      "learning_rate": 5.5968535858950345e-06,
+      "loss": 1.8116,
+      "step": 899
+    },
+    {
+      "epoch": 0.8012463832628534,
+      "grad_norm": 1.8434404134750366,
+      "learning_rate": 5.49369248201953e-06,
+      "loss": 1.879,
+      "step": 900
+    },
+    {
+      "epoch": 0.8021366570220343,
+      "grad_norm": 2.205735206604004,
+      "learning_rate": 5.39146411724547e-06,
+      "loss": 2.0231,
+      "step": 901
+    },
+    {
+      "epoch": 0.8030269307812152,
+      "grad_norm": 2.084808588027954,
+      "learning_rate": 5.290169500525577e-06,
+      "loss": 2.0352,
+      "step": 902
+    },
+    {
+      "epoch": 0.8039172045403962,
+      "grad_norm": 2.2908926010131836,
+      "learning_rate": 5.189809631596798e-06,
+      "loss": 1.8234,
+      "step": 903
+    },
+    {
+      "epoch": 0.8048074782995771,
+      "grad_norm": 2.077094078063965,
+      "learning_rate": 5.0903855009705514e-06,
+      "loss": 2.0776,
+      "step": 904
+    },
+    {
+      "epoch": 0.805697752058758,
+      "grad_norm": 1.568692922592163,
+      "learning_rate": 4.991898089922819e-06,
+      "loss": 1.4186,
+      "step": 905
+    },
+    {
+      "epoch": 0.806588025817939,
+      "grad_norm": 1.4289157390594482,
+      "learning_rate": 4.8943483704846475e-06,
+      "loss": 1.1468,
+      "step": 906
+    },
+    {
+      "epoch": 0.8074782995771199,
+      "grad_norm": 3.5975162982940674,
+      "learning_rate": 4.797737305432337e-06,
+      "loss": 2.2185,
+      "step": 907
+    },
+    {
+      "epoch": 0.808368573336301,
+      "grad_norm": 1.8757998943328857,
+      "learning_rate": 4.702065848278126e-06,
+      "loss": 1.788,
+      "step": 908
+    },
+    {
+      "epoch": 0.8092588470954819,
+      "grad_norm": 2.071596145629883,
+      "learning_rate": 4.607334943260655e-06,
+      "loss": 1.4828,
+      "step": 909
+    },
+    {
+      "epoch": 0.8101491208546628,
+      "grad_norm": 1.8785046339035034,
+      "learning_rate": 4.513545525335705e-06,
+      "loss": 1.8259,
+      "step": 910
+    },
+    {
+      "epoch": 0.8110393946138438,
+      "grad_norm": 2.3740673065185547,
+      "learning_rate": 4.420698520166988e-06,
+      "loss": 1.9712,
+      "step": 911
+    },
+    {
+      "epoch": 0.8119296683730247,
+      "grad_norm": 1.753470540046692,
+      "learning_rate": 4.328794844116946e-06,
+      "loss": 1.1425,
+      "step": 912
+    },
+    {
+      "epoch": 0.8128199421322057,
+      "grad_norm": 2.1722190380096436,
+      "learning_rate": 4.237835404237778e-06,
+      "loss": 2.1072,
+      "step": 913
+    },
+    {
+      "epoch": 0.8137102158913866,
+      "grad_norm": 1.5812511444091797,
+      "learning_rate": 4.147821098262405e-06,
+      "loss": 1.8418,
+      "step": 914
+    },
+    {
+      "epoch": 0.8146004896505675,
+      "grad_norm": 2.335020065307617,
+      "learning_rate": 4.0587528145957235e-06,
+      "loss": 2.2968,
+      "step": 915
+    },
+    {
+      "epoch": 0.8154907634097485,
+      "grad_norm": 4.78397798538208,
+      "learning_rate": 3.970631432305694e-06,
+      "loss": 2.5117,
+      "step": 916
+    },
+    {
+      "epoch": 0.8163810371689294,
+      "grad_norm": 2.6600706577301025,
+      "learning_rate": 3.883457821114811e-06,
+      "loss": 1.7201,
+      "step": 917
+    },
+    {
+      "epoch": 0.8172713109281103,
+      "grad_norm": 1.75503671169281,
+      "learning_rate": 3.797232841391407e-06,
+      "loss": 2.0005,
+      "step": 918
+    },
+    {
+      "epoch": 0.8181615846872914,
+      "grad_norm": 1.8299975395202637,
+      "learning_rate": 3.711957344141237e-06,
+      "loss": 1.883,
+      "step": 919
+    },
+    {
+      "epoch": 0.8190518584464723,
+      "grad_norm": 2.177534818649292,
+      "learning_rate": 3.627632170999029e-06,
+      "loss": 1.5191,
+      "step": 920
+    },
+    {
+      "epoch": 0.8199421322056533,
+      "grad_norm": 1.4951258897781372,
+      "learning_rate": 3.5442581542201923e-06,
+      "loss": 1.4525,
+      "step": 921
+    },
+    {
+      "epoch": 0.8208324059648342,
+      "grad_norm": 2.103001117706299,
+      "learning_rate": 3.461836116672612e-06,
+      "loss": 2.0439,
+      "step": 922
+    },
+    {
+      "epoch": 0.8217226797240151,
+      "grad_norm": 1.925445318222046,
+      "learning_rate": 3.380366871828522e-06,
+      "loss": 2.3512,
+      "step": 923
+    },
+    {
+      "epoch": 0.8226129534831961,
+      "grad_norm": 1.6522331237792969,
+      "learning_rate": 3.2998512237565005e-06,
+      "loss": 1.7079,
+      "step": 924
+    },
+    {
+      "epoch": 0.823503227242377,
+      "grad_norm": 1.6218456029891968,
+      "learning_rate": 3.2202899671134546e-06,
+      "loss": 1.6155,
+      "step": 925
+    },
+    {
+      "epoch": 0.824393501001558,
+      "grad_norm": 2.5410165786743164,
+      "learning_rate": 3.1416838871368924e-06,
+      "loss": 1.8986,
+      "step": 926
+    },
+    {
+      "epoch": 0.8252837747607389,
+      "grad_norm": 1.5301979780197144,
+      "learning_rate": 3.064033759637064e-06,
+      "loss": 1.6242,
+      "step": 927
+    },
+    {
+      "epoch": 0.8261740485199198,
+      "grad_norm": 2.0955357551574707,
+      "learning_rate": 2.9873403509894203e-06,
+      "loss": 2.2897,
+      "step": 928
+    },
+    {
+      "epoch": 0.8270643222791009,
+      "grad_norm": 2.5796597003936768,
+      "learning_rate": 2.9116044181269007e-06,
+      "loss": 2.3979,
+      "step": 929
+    },
+    {
+      "epoch": 0.8279545960382818,
+      "grad_norm": 1.8714462518692017,
+      "learning_rate": 2.836826708532603e-06,
+      "loss": 2.3742,
+      "step": 930
+    },
+    {
+      "epoch": 0.8288448697974627,
+      "grad_norm": 2.4107353687286377,
+      "learning_rate": 2.7630079602323442e-06,
+      "loss": 1.8276,
+      "step": 931
+    },
+    {
+      "epoch": 0.8297351435566437,
+      "grad_norm": 1.8410996198654175,
+      "learning_rate": 2.690148901787337e-06,
+      "loss": 1.5216,
+      "step": 932
+    },
+    {
+      "epoch": 0.8306254173158246,
+      "grad_norm": 2.153083324432373,
+      "learning_rate": 2.618250252287113e-06,
+      "loss": 2.5236,
+      "step": 933
+    },
+    {
+      "epoch": 0.8315156910750056,
+      "grad_norm": 2.0374317169189453,
+      "learning_rate": 2.5473127213422763e-06,
+      "loss": 2.2,
+      "step": 934
+    },
+    {
+      "epoch": 0.8324059648341865,
+      "grad_norm": 1.670549988746643,
+      "learning_rate": 2.4773370090776626e-06,
+      "loss": 1.7859,
+      "step": 935
+    },
+    {
+      "epoch": 0.8332962385933674,
+      "grad_norm": 2.1211376190185547,
+      "learning_rate": 2.4083238061252567e-06,
+      "loss": 2.1081,
+      "step": 936
+    },
+    {
+      "epoch": 0.8341865123525484,
+      "grad_norm": 1.8752716779708862,
+      "learning_rate": 2.3402737936175425e-06,
+      "loss": 1.8866,
+      "step": 937
+    },
+    {
+      "epoch": 0.8350767861117293,
+      "grad_norm": 2.2133710384368896,
+      "learning_rate": 2.273187643180652e-06,
+      "loss": 1.6484,
+      "step": 938
+    },
+    {
+      "epoch": 0.8359670598709104,
+      "grad_norm": 2.0927562713623047,
+      "learning_rate": 2.2070660169278166e-06,
+      "loss": 1.9145,
+      "step": 939
+    },
+    {
+      "epoch": 0.8368573336300913,
+      "grad_norm": 2.2053141593933105,
+      "learning_rate": 2.141909567452793e-06,
+      "loss": 1.6497,
+      "step": 940
+    },
+    {
+      "epoch": 0.8377476073892722,
+      "grad_norm": 1.718266248703003,
+      "learning_rate": 2.0777189378234143e-06,
+      "loss": 1.8699,
+      "step": 941
+    },
+    {
+      "epoch": 0.8386378811484532,
+      "grad_norm": 1.7345627546310425,
+      "learning_rate": 2.014494761575314e-06,
+      "loss": 1.9545,
+      "step": 942
+    },
+    {
+      "epoch": 0.8395281549076341,
+      "grad_norm": 2.2770743370056152,
+      "learning_rate": 1.9522376627055583e-06,
+      "loss": 2.4747,
+      "step": 943
+    },
+    {
+      "epoch": 0.840418428666815,
+      "grad_norm": 3.990893840789795,
+      "learning_rate": 1.8909482556666024e-06,
+      "loss": 2.4903,
+      "step": 944
+    },
+    {
+      "epoch": 0.841308702425996,
+      "grad_norm": 1.6306087970733643,
+      "learning_rate": 1.8306271453601199e-06,
+      "loss": 1.4113,
+      "step": 945
+    },
+    {
+      "epoch": 0.8421989761851769,
+      "grad_norm": 1.6917046308517456,
+      "learning_rate": 1.771274927131139e-06,
+      "loss": 1.7981,
+      "step": 946
+    },
+    {
+      "epoch": 0.8430892499443579,
+      "grad_norm": 2.0122792720794678,
+      "learning_rate": 1.712892186762083e-06,
+      "loss": 1.671,
+      "step": 947
+    },
+    {
+      "epoch": 0.8439795237035388,
+      "grad_norm": 1.5916146039962769,
+      "learning_rate": 1.6554795004670388e-06,
+      "loss": 1.5368,
+      "step": 948
+    },
+    {
+      "epoch": 0.8448697974627198,
+      "grad_norm": 1.4738613367080688,
+      "learning_rate": 1.5990374348860305e-06,
+      "loss": 1.7981,
+      "step": 949
+    },
+    {
+      "epoch": 0.8457600712219008,
+      "grad_norm": 1.550717830657959,
+      "learning_rate": 1.543566547079467e-06,
+      "loss": 1.4818,
+      "step": 950
+    },
+    {
+      "epoch": 0.8466503449810817,
+      "grad_norm": 1.380691409111023,
+      "learning_rate": 1.4890673845226133e-06,
+      "loss": 1.5712,
+      "step": 951
+    },
+    {
+      "epoch": 0.8475406187402627,
+      "grad_norm": 1.9173582792282104,
+      "learning_rate": 1.4355404851001952e-06,
+      "loss": 1.7729,
+      "step": 952
+    },
+    {
+      "epoch": 0.8484308924994436,
+      "grad_norm": 1.9216915369033813,
+      "learning_rate": 1.3829863771011253e-06,
+      "loss": 1.767,
+      "step": 953
+    },
+    {
+      "epoch": 0.8493211662586245,
+      "grad_norm": 1.4184794425964355,
+      "learning_rate": 1.3314055792131964e-06,
+      "loss": 1.2916,
+      "step": 954
+    },
+    {
+      "epoch": 0.8502114400178055,
+      "grad_norm": 2.1560451984405518,
+      "learning_rate": 1.280798600518085e-06,
+      "loss": 1.5511,
+      "step": 955
+    },
+    {
+      "epoch": 0.8511017137769864,
+      "grad_norm": 1.8579264879226685,
+      "learning_rate": 1.231165940486234e-06,
+      "loss": 2.1381,
+      "step": 956
+    },
+    {
+      "epoch": 0.8519919875361673,
+      "grad_norm": 1.4371646642684937,
+      "learning_rate": 1.1825080889719563e-06,
+      "loss": 1.6049,
+      "step": 957
+    },
+    {
+      "epoch": 0.8528822612953483,
+      "grad_norm": 2.4612481594085693,
+      "learning_rate": 1.134825526208605e-06,
+      "loss": 1.804,
+      "step": 958
+    },
+    {
+      "epoch": 0.8537725350545292,
+      "grad_norm": 1.62546706199646,
+      "learning_rate": 1.0881187228038215e-06,
+      "loss": 1.8068,
+      "step": 959
+    },
+    {
+      "epoch": 0.8546628088137103,
+      "grad_norm": 2.140608787536621,
+      "learning_rate": 1.0423881397349068e-06,
+      "loss": 1.8131,
+      "step": 960
+    },
+    {
+      "epoch": 0.8555530825728912,
+      "grad_norm": 1.6791179180145264,
+      "learning_rate": 9.976342283442463e-07,
+      "loss": 2.104,
+      "step": 961
+    },
+    {
+      "epoch": 0.8564433563320721,
+      "grad_norm": 1.730907917022705,
+      "learning_rate": 9.538574303348813e-07,
+      "loss": 1.8383,
+      "step": 962
+    },
+    {
+      "epoch": 0.8573336300912531,
+      "grad_norm": 1.5879576206207275,
+      "learning_rate": 9.110581777661331e-07,
+      "loss": 1.705,
+      "step": 963
+    },
+    {
+      "epoch": 0.858223903850434,
+      "grad_norm": 2.378976821899414,
+      "learning_rate": 8.692368930493521e-07,
+      "loss": 2.2417,
+      "step": 964
+    },
+    {
+      "epoch": 0.859114177609615,
+      "grad_norm": 1.8429821729660034,
+      "learning_rate": 8.283939889437209e-07,
+      "loss": 2.1216,
+      "step": 965
+    },
+    {
+      "epoch": 0.8600044513687959,
+      "grad_norm": 3.7237606048583984,
+      "learning_rate": 7.885298685522235e-07,
+      "loss": 2.1695,
+      "step": 966
+    },
+    {
+      "epoch": 0.8608947251279768,
+      "grad_norm": 2.168633222579956,
+      "learning_rate": 7.496449253176274e-07,
+      "loss": 2.2198,
+      "step": 967
+    },
+    {
+      "epoch": 0.8617849988871578,
+      "grad_norm": 1.5449988842010498,
+      "learning_rate": 7.117395430186414e-07,
+      "loss": 1.6742,
+      "step": 968
+    },
+    {
+      "epoch": 0.8626752726463387,
+      "grad_norm": 1.8865556716918945,
+      "learning_rate": 6.748140957660631e-07,
+      "loss": 1.8883,
+      "step": 969
+    },
+    {
+      "epoch": 0.8635655464055197,
+      "grad_norm": 1.589005708694458,
+      "learning_rate": 6.388689479991605e-07,
+      "loss": 1.2163,
+      "step": 970
+    },
+    {
+      "epoch": 0.8644558201647007,
+      "grad_norm": 1.6373053789138794,
+      "learning_rate": 6.039044544820404e-07,
+      "loss": 1.4072,
+      "step": 971
+    },
+    {
+      "epoch": 0.8653460939238816,
+      "grad_norm": 2.5038976669311523,
+      "learning_rate": 5.699209603001076e-07,
+      "loss": 2.0531,
+      "step": 972
+    },
+    {
+      "epoch": 0.8662363676830626,
+      "grad_norm": 1.6263477802276611,
+      "learning_rate": 5.369188008567672e-07,
+      "loss": 1.5389,
+      "step": 973
+    },
+    {
+      "epoch": 0.8671266414422435,
+      "grad_norm": 1.6909034252166748,
+      "learning_rate": 5.048983018699827e-07,
+      "loss": 1.3181,
+      "step": 974
+    },
+    {
+      "epoch": 0.8680169152014244,
+      "grad_norm": 1.7488054037094116,
+      "learning_rate": 4.738597793691679e-07,
+      "loss": 1.5009,
+      "step": 975
+    },
+    {
+      "epoch": 0.8689071889606054,
+      "grad_norm": 1.4903292655944824,
+      "learning_rate": 4.438035396920004e-07,
+      "loss": 1.283,
+      "step": 976
+    },
+    {
+      "epoch": 0.8697974627197863,
+      "grad_norm": 2.250135898590088,
+      "learning_rate": 4.1472987948143473e-07,
+      "loss": 1.8839,
+      "step": 977
+    },
+    {
+      "epoch": 0.8706877364789672,
+      "grad_norm": 2.169273614883423,
+      "learning_rate": 3.866390856827495e-07,
+      "loss": 1.9408,
+      "step": 978
+    },
+    {
+      "epoch": 0.8715780102381482,
+      "grad_norm": 1.683918833732605,
+      "learning_rate": 3.595314355407609e-07,
+      "loss": 1.6478,
+      "step": 979
+    },
+    {
+      "epoch": 0.8724682839973292,
+      "grad_norm": 2.058692455291748,
+      "learning_rate": 3.3340719659701313e-07,
+      "loss": 2.2342,
+      "step": 980
+    },
+    {
+      "epoch": 0.8733585577565102,
+      "grad_norm": 1.406325101852417,
+      "learning_rate": 3.0826662668720364e-07,
+      "loss": 1.4541,
+      "step": 981
+    },
+    {
+      "epoch": 0.8742488315156911,
+      "grad_norm": 3.1064553260803223,
+      "learning_rate": 2.841099739386066e-07,
+      "loss": 3.0294,
+      "step": 982
+    },
+    {
+      "epoch": 0.875139105274872,
+      "grad_norm": 1.261873722076416,
+      "learning_rate": 2.609374767676309e-07,
+      "loss": 1.4086,
+      "step": 983
+    },
+    {
+      "epoch": 0.876029379034053,
+      "grad_norm": 1.0917588472366333,
+      "learning_rate": 2.387493638774774e-07,
+      "loss": 1.159,
+      "step": 984
+    },
+    {
+      "epoch": 0.8769196527932339,
+      "grad_norm": 1.7324159145355225,
+      "learning_rate": 2.175458542558517e-07,
+      "loss": 1.469,
+      "step": 985
+    },
+    {
+      "epoch": 0.8778099265524149,
+      "grad_norm": 1.3332997560501099,
+      "learning_rate": 1.973271571728441e-07,
+      "loss": 1.0762,
+      "step": 986
+    },
+    {
+      "epoch": 0.8787002003115958,
+      "grad_norm": 1.856528878211975,
+      "learning_rate": 1.7809347217881966e-07,
+      "loss": 2.0178,
+      "step": 987
+    },
+    {
+      "epoch": 0.8795904740707767,
+      "grad_norm": 2.5236916542053223,
+      "learning_rate": 1.598449891024978e-07,
+      "loss": 1.7862,
+      "step": 988
+    },
+    {
+      "epoch": 0.8804807478299577,
+      "grad_norm": 2.023587703704834,
+      "learning_rate": 1.425818880490315e-07,
+      "loss": 1.8994,
+      "step": 989
+    },
+    {
+      "epoch": 0.8813710215891387,
+      "grad_norm": 1.4901554584503174,
+      "learning_rate": 1.2630433939825327e-07,
+      "loss": 1.5095,
+      "step": 990
+    },
+    {
+      "epoch": 0.8822612953483197,
+      "grad_norm": 1.6693906784057617,
+      "learning_rate": 1.1101250380300965e-07,
+      "loss": 1.4956,
+      "step": 991
+    },
+    {
+      "epoch": 0.8831515691075006,
+      "grad_norm": 2.0669729709625244,
+      "learning_rate": 9.670653218752934e-08,
+      "loss": 2.1793,
+      "step": 992
+    },
+    {
+      "epoch": 0.8840418428666815,
+      "grad_norm": 2.204921007156372,
+      "learning_rate": 8.33865657459909e-08,
+      "loss": 1.751,
+      "step": 993
+    },
+    {
+      "epoch": 0.8849321166258625,
+      "grad_norm": 1.2985179424285889,
+      "learning_rate": 7.105273594107953e-08,
+      "loss": 1.4216,
+      "step": 994
+    },
+    {
+      "epoch": 0.8858223903850434,
+      "grad_norm": 1.683183193206787,
+      "learning_rate": 5.970516450271025e-08,
+      "loss": 2.0145,
+      "step": 995
+    },
+    {
+      "epoch": 0.8867126641442243,
+      "grad_norm": 1.415525197982788,
+      "learning_rate": 4.934396342684e-08,
+      "loss": 1.3802,
+      "step": 996
+    },
+    {
+      "epoch": 0.8876029379034053,
+      "grad_norm": 1.679310917854309,
+      "learning_rate": 3.996923497434635e-08,
+      "loss": 1.5799,
+      "step": 997
+    },
+    {
+      "epoch": 0.8884932116625862,
+      "grad_norm": 3.5218520164489746,
+      "learning_rate": 3.1581071670006015e-08,
+      "loss": 1.5458,
+      "step": 998
+    },
+    {
+      "epoch": 0.8893834854217672,
+      "grad_norm": 1.5060161352157593,
+      "learning_rate": 2.417955630159563e-08,
+      "loss": 1.0504,
+      "step": 999
+    },
+    {
+      "epoch": 0.8902737591809481,
+      "grad_norm": 2.1258907318115234,
+      "learning_rate": 1.7764761919103477e-08,
+      "loss": 2.3434,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "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": 6.34658190524416e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}