diff --git "a/checkpoint-1500/trainer_state.json" "b/checkpoint-1500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1500/trainer_state.json"
@@ -0,0 +1,10533 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9759271307742355,
+  "eval_steps": 500,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0006506180871828237,
+      "grad_norm": 3.778571605682373,
+      "learning_rate": 0.0001,
+      "loss": 4.706,
+      "step": 1
+    },
+    {
+      "epoch": 0.0013012361743656475,
+      "grad_norm": 0.7331739068031311,
+      "learning_rate": 0.0001,
+      "loss": 2.6402,
+      "step": 2
+    },
+    {
+      "epoch": 0.001951854261548471,
+      "grad_norm": 0.5679969191551208,
+      "learning_rate": 0.0001,
+      "loss": 2.5315,
+      "step": 3
+    },
+    {
+      "epoch": 0.002602472348731295,
+      "grad_norm": 0.6543067693710327,
+      "learning_rate": 0.0001,
+      "loss": 2.5226,
+      "step": 4
+    },
+    {
+      "epoch": 0.0032530904359141183,
+      "grad_norm": 0.42487671971321106,
+      "learning_rate": 0.0001,
+      "loss": 2.1375,
+      "step": 5
+    },
+    {
+      "epoch": 0.003903708523096942,
+      "grad_norm": 0.48795655369758606,
+      "learning_rate": 0.0001,
+      "loss": 2.253,
+      "step": 6
+    },
+    {
+      "epoch": 0.004554326610279766,
+      "grad_norm": 0.6054234504699707,
+      "learning_rate": 0.0001,
+      "loss": 2.3411,
+      "step": 7
+    },
+    {
+      "epoch": 0.00520494469746259,
+      "grad_norm": 0.3039970397949219,
+      "learning_rate": 0.0001,
+      "loss": 2.1293,
+      "step": 8
+    },
+    {
+      "epoch": 0.005855562784645413,
+      "grad_norm": 0.6592361330986023,
+      "learning_rate": 0.0001,
+      "loss": 3.1615,
+      "step": 9
+    },
+    {
+      "epoch": 0.006506180871828237,
+      "grad_norm": 0.4017999470233917,
+      "learning_rate": 0.0001,
+      "loss": 2.5068,
+      "step": 10
+    },
+    {
+      "epoch": 0.0071567989590110605,
+      "grad_norm": 0.31507641077041626,
+      "learning_rate": 0.0001,
+      "loss": 2.1894,
+      "step": 11
+    },
+    {
+      "epoch": 0.007807417046193884,
+      "grad_norm": 0.33226895332336426,
+      "learning_rate": 0.0001,
+      "loss": 2.2006,
+      "step": 12
+    },
+    {
+      "epoch": 0.008458035133376708,
+      "grad_norm": 0.2632739841938019,
+      "learning_rate": 0.0001,
+      "loss": 2.0998,
+      "step": 13
+    },
+    {
+      "epoch": 0.009108653220559532,
+      "grad_norm": 0.2794795036315918,
+      "learning_rate": 0.0001,
+      "loss": 2.113,
+      "step": 14
+    },
+    {
+      "epoch": 0.009759271307742356,
+      "grad_norm": 0.29168492555618286,
+      "learning_rate": 0.0001,
+      "loss": 2.354,
+      "step": 15
+    },
+    {
+      "epoch": 0.01040988939492518,
+      "grad_norm": 0.2537970244884491,
+      "learning_rate": 0.0001,
+      "loss": 2.2939,
+      "step": 16
+    },
+    {
+      "epoch": 0.011060507482108002,
+      "grad_norm": 0.5140053033828735,
+      "learning_rate": 0.0001,
+      "loss": 2.6237,
+      "step": 17
+    },
+    {
+      "epoch": 0.011711125569290826,
+      "grad_norm": 0.3093675971031189,
+      "learning_rate": 0.0001,
+      "loss": 2.3502,
+      "step": 18
+    },
+    {
+      "epoch": 0.01236174365647365,
+      "grad_norm": 0.29241421818733215,
+      "learning_rate": 0.0001,
+      "loss": 2.5365,
+      "step": 19
+    },
+    {
+      "epoch": 0.013012361743656473,
+      "grad_norm": 0.3164322078227997,
+      "learning_rate": 0.0001,
+      "loss": 2.396,
+      "step": 20
+    },
+    {
+      "epoch": 0.013662979830839297,
+      "grad_norm": 0.24512743949890137,
+      "learning_rate": 0.0001,
+      "loss": 2.2759,
+      "step": 21
+    },
+    {
+      "epoch": 0.014313597918022121,
+      "grad_norm": 0.24328342080116272,
+      "learning_rate": 0.0001,
+      "loss": 2.2103,
+      "step": 22
+    },
+    {
+      "epoch": 0.014964216005204945,
+      "grad_norm": 0.2563220262527466,
+      "learning_rate": 0.0001,
+      "loss": 2.4836,
+      "step": 23
+    },
+    {
+      "epoch": 0.015614834092387769,
+      "grad_norm": 0.33601588010787964,
+      "learning_rate": 0.0001,
+      "loss": 2.4446,
+      "step": 24
+    },
+    {
+      "epoch": 0.01626545217957059,
+      "grad_norm": 0.28699007630348206,
+      "learning_rate": 0.0001,
+      "loss": 2.8504,
+      "step": 25
+    },
+    {
+      "epoch": 0.016916070266753416,
+      "grad_norm": 0.3181653618812561,
+      "learning_rate": 0.0001,
+      "loss": 2.3042,
+      "step": 26
+    },
+    {
+      "epoch": 0.01756668835393624,
+      "grad_norm": 0.2349390834569931,
+      "learning_rate": 0.0001,
+      "loss": 2.1024,
+      "step": 27
+    },
+    {
+      "epoch": 0.018217306441119064,
+      "grad_norm": 0.2751820981502533,
+      "learning_rate": 0.0001,
+      "loss": 2.2646,
+      "step": 28
+    },
+    {
+      "epoch": 0.018867924528301886,
+      "grad_norm": 0.25547271966934204,
+      "learning_rate": 0.0001,
+      "loss": 2.1928,
+      "step": 29
+    },
+    {
+      "epoch": 0.01951854261548471,
+      "grad_norm": 0.283507764339447,
+      "learning_rate": 0.0001,
+      "loss": 2.3073,
+      "step": 30
+    },
+    {
+      "epoch": 0.020169160702667534,
+      "grad_norm": 0.3354213237762451,
+      "learning_rate": 0.0001,
+      "loss": 2.6273,
+      "step": 31
+    },
+    {
+      "epoch": 0.02081977878985036,
+      "grad_norm": 0.40484553575515747,
+      "learning_rate": 0.0001,
+      "loss": 2.4919,
+      "step": 32
+    },
+    {
+      "epoch": 0.02147039687703318,
+      "grad_norm": 0.34319421648979187,
+      "learning_rate": 0.0001,
+      "loss": 2.8381,
+      "step": 33
+    },
+    {
+      "epoch": 0.022121014964216004,
+      "grad_norm": 0.32958984375,
+      "learning_rate": 0.0001,
+      "loss": 2.3062,
+      "step": 34
+    },
+    {
+      "epoch": 0.02277163305139883,
+      "grad_norm": 0.4503105878829956,
+      "learning_rate": 0.0001,
+      "loss": 2.4647,
+      "step": 35
+    },
+    {
+      "epoch": 0.02342225113858165,
+      "grad_norm": 0.5084238052368164,
+      "learning_rate": 0.0001,
+      "loss": 3.0047,
+      "step": 36
+    },
+    {
+      "epoch": 0.024072869225764477,
+      "grad_norm": 0.5192400813102722,
+      "learning_rate": 0.0001,
+      "loss": 2.2899,
+      "step": 37
+    },
+    {
+      "epoch": 0.0247234873129473,
+      "grad_norm": 0.4197874665260315,
+      "learning_rate": 0.0001,
+      "loss": 2.4057,
+      "step": 38
+    },
+    {
+      "epoch": 0.025374105400130124,
+      "grad_norm": 0.5170285105705261,
+      "learning_rate": 0.0001,
+      "loss": 3.2918,
+      "step": 39
+    },
+    {
+      "epoch": 0.026024723487312947,
+      "grad_norm": 0.2491147667169571,
+      "learning_rate": 0.0001,
+      "loss": 2.1957,
+      "step": 40
+    },
+    {
+      "epoch": 0.026675341574495772,
+      "grad_norm": 0.6597635746002197,
+      "learning_rate": 0.0001,
+      "loss": 2.7474,
+      "step": 41
+    },
+    {
+      "epoch": 0.027325959661678594,
+      "grad_norm": 0.40205034613609314,
+      "learning_rate": 0.0001,
+      "loss": 2.4561,
+      "step": 42
+    },
+    {
+      "epoch": 0.02797657774886142,
+      "grad_norm": 0.27388331294059753,
+      "learning_rate": 0.0001,
+      "loss": 2.0477,
+      "step": 43
+    },
+    {
+      "epoch": 0.028627195836044242,
+      "grad_norm": 0.9163908958435059,
+      "learning_rate": 0.0001,
+      "loss": 3.334,
+      "step": 44
+    },
+    {
+      "epoch": 0.029277813923227064,
+      "grad_norm": 0.2747696042060852,
+      "learning_rate": 0.0001,
+      "loss": 2.1604,
+      "step": 45
+    },
+    {
+      "epoch": 0.02992843201040989,
+      "grad_norm": 0.36308085918426514,
+      "learning_rate": 0.0001,
+      "loss": 2.693,
+      "step": 46
+    },
+    {
+      "epoch": 0.03057905009759271,
+      "grad_norm": 0.6159886121749878,
+      "learning_rate": 0.0001,
+      "loss": 2.5515,
+      "step": 47
+    },
+    {
+      "epoch": 0.031229668184775537,
+      "grad_norm": 0.4801373779773712,
+      "learning_rate": 0.0001,
+      "loss": 2.809,
+      "step": 48
+    },
+    {
+      "epoch": 0.03188028627195836,
+      "grad_norm": 0.32580915093421936,
+      "learning_rate": 0.0001,
+      "loss": 2.5236,
+      "step": 49
+    },
+    {
+      "epoch": 0.03253090435914118,
+      "grad_norm": 0.3028671443462372,
+      "learning_rate": 0.0001,
+      "loss": 2.2685,
+      "step": 50
+    },
+    {
+      "epoch": 0.03318152244632401,
+      "grad_norm": 0.5660931468009949,
+      "learning_rate": 0.0001,
+      "loss": 2.2564,
+      "step": 51
+    },
+    {
+      "epoch": 0.03383214053350683,
+      "grad_norm": 0.24634602665901184,
+      "learning_rate": 0.0001,
+      "loss": 2.1355,
+      "step": 52
+    },
+    {
+      "epoch": 0.034482758620689655,
+      "grad_norm": 0.24830913543701172,
+      "learning_rate": 0.0001,
+      "loss": 2.0425,
+      "step": 53
+    },
+    {
+      "epoch": 0.03513337670787248,
+      "grad_norm": 0.23614570498466492,
+      "learning_rate": 0.0001,
+      "loss": 2.1975,
+      "step": 54
+    },
+    {
+      "epoch": 0.035783994795055306,
+      "grad_norm": 0.2624325156211853,
+      "learning_rate": 0.0001,
+      "loss": 2.3071,
+      "step": 55
+    },
+    {
+      "epoch": 0.03643461288223813,
+      "grad_norm": 0.3967755436897278,
+      "learning_rate": 0.0001,
+      "loss": 2.6088,
+      "step": 56
+    },
+    {
+      "epoch": 0.03708523096942095,
+      "grad_norm": 0.22147373855113983,
+      "learning_rate": 0.0001,
+      "loss": 2.003,
+      "step": 57
+    },
+    {
+      "epoch": 0.03773584905660377,
+      "grad_norm": 0.47795867919921875,
+      "learning_rate": 0.0001,
+      "loss": 2.1473,
+      "step": 58
+    },
+    {
+      "epoch": 0.038386467143786594,
+      "grad_norm": 0.43953707814216614,
+      "learning_rate": 0.0001,
+      "loss": 2.6595,
+      "step": 59
+    },
+    {
+      "epoch": 0.03903708523096942,
+      "grad_norm": 0.29031845927238464,
+      "learning_rate": 0.0001,
+      "loss": 2.3173,
+      "step": 60
+    },
+    {
+      "epoch": 0.039687703318152245,
+      "grad_norm": 0.2491024285554886,
+      "learning_rate": 0.0001,
+      "loss": 2.0575,
+      "step": 61
+    },
+    {
+      "epoch": 0.04033832140533507,
+      "grad_norm": 0.3025687634944916,
+      "learning_rate": 0.0001,
+      "loss": 2.0965,
+      "step": 62
+    },
+    {
+      "epoch": 0.04098893949251789,
+      "grad_norm": 0.26097819209098816,
+      "learning_rate": 0.0001,
+      "loss": 2.2583,
+      "step": 63
+    },
+    {
+      "epoch": 0.04163955757970072,
+      "grad_norm": 0.2413238286972046,
+      "learning_rate": 0.0001,
+      "loss": 2.2441,
+      "step": 64
+    },
+    {
+      "epoch": 0.04229017566688354,
+      "grad_norm": 0.2332315295934677,
+      "learning_rate": 0.0001,
+      "loss": 2.185,
+      "step": 65
+    },
+    {
+      "epoch": 0.04294079375406636,
+      "grad_norm": 0.4037252366542816,
+      "learning_rate": 0.0001,
+      "loss": 2.3875,
+      "step": 66
+    },
+    {
+      "epoch": 0.043591411841249185,
+      "grad_norm": 0.34149354696273804,
+      "learning_rate": 0.0001,
+      "loss": 2.3835,
+      "step": 67
+    },
+    {
+      "epoch": 0.04424202992843201,
+      "grad_norm": 0.23793481290340424,
+      "learning_rate": 0.0001,
+      "loss": 2.3521,
+      "step": 68
+    },
+    {
+      "epoch": 0.044892648015614836,
+      "grad_norm": 0.24252744019031525,
+      "learning_rate": 0.0001,
+      "loss": 2.0984,
+      "step": 69
+    },
+    {
+      "epoch": 0.04554326610279766,
+      "grad_norm": 0.2870447635650635,
+      "learning_rate": 0.0001,
+      "loss": 2.5408,
+      "step": 70
+    },
+    {
+      "epoch": 0.04619388418998048,
+      "grad_norm": 0.5050077438354492,
+      "learning_rate": 0.0001,
+      "loss": 2.7091,
+      "step": 71
+    },
+    {
+      "epoch": 0.0468445022771633,
+      "grad_norm": 0.2391565591096878,
+      "learning_rate": 0.0001,
+      "loss": 2.1601,
+      "step": 72
+    },
+    {
+      "epoch": 0.04749512036434613,
+      "grad_norm": 0.20647507905960083,
+      "learning_rate": 0.0001,
+      "loss": 1.9582,
+      "step": 73
+    },
+    {
+      "epoch": 0.048145738451528954,
+      "grad_norm": 0.26072338223457336,
+      "learning_rate": 0.0001,
+      "loss": 2.3577,
+      "step": 74
+    },
+    {
+      "epoch": 0.048796356538711776,
+      "grad_norm": 0.28378504514694214,
+      "learning_rate": 0.0001,
+      "loss": 2.349,
+      "step": 75
+    },
+    {
+      "epoch": 0.0494469746258946,
+      "grad_norm": 0.2536943256855011,
+      "learning_rate": 0.0001,
+      "loss": 2.375,
+      "step": 76
+    },
+    {
+      "epoch": 0.05009759271307743,
+      "grad_norm": 0.29276445508003235,
+      "learning_rate": 0.0001,
+      "loss": 2.5003,
+      "step": 77
+    },
+    {
+      "epoch": 0.05074821080026025,
+      "grad_norm": 0.2649310231208801,
+      "learning_rate": 0.0001,
+      "loss": 2.3247,
+      "step": 78
+    },
+    {
+      "epoch": 0.05139882888744307,
+      "grad_norm": 0.38125383853912354,
+      "learning_rate": 0.0001,
+      "loss": 2.5405,
+      "step": 79
+    },
+    {
+      "epoch": 0.05204944697462589,
+      "grad_norm": 0.40980008244514465,
+      "learning_rate": 0.0001,
+      "loss": 2.212,
+      "step": 80
+    },
+    {
+      "epoch": 0.052700065061808715,
+      "grad_norm": 0.5363492965698242,
+      "learning_rate": 0.0001,
+      "loss": 2.6499,
+      "step": 81
+    },
+    {
+      "epoch": 0.053350683148991544,
+      "grad_norm": 0.34647300839424133,
+      "learning_rate": 0.0001,
+      "loss": 2.6302,
+      "step": 82
+    },
+    {
+      "epoch": 0.054001301236174366,
+      "grad_norm": 0.27607980370521545,
+      "learning_rate": 0.0001,
+      "loss": 2.1819,
+      "step": 83
+    },
+    {
+      "epoch": 0.05465191932335719,
+      "grad_norm": 0.27654680609703064,
+      "learning_rate": 0.0001,
+      "loss": 2.1763,
+      "step": 84
+    },
+    {
+      "epoch": 0.05530253741054001,
+      "grad_norm": 0.24596217274665833,
+      "learning_rate": 0.0001,
+      "loss": 2.2585,
+      "step": 85
+    },
+    {
+      "epoch": 0.05595315549772284,
+      "grad_norm": 0.24279890954494476,
+      "learning_rate": 0.0001,
+      "loss": 2.4247,
+      "step": 86
+    },
+    {
+      "epoch": 0.05660377358490566,
+      "grad_norm": 0.2918747365474701,
+      "learning_rate": 0.0001,
+      "loss": 2.3986,
+      "step": 87
+    },
+    {
+      "epoch": 0.057254391672088484,
+      "grad_norm": 0.26778745651245117,
+      "learning_rate": 0.0001,
+      "loss": 2.3592,
+      "step": 88
+    },
+    {
+      "epoch": 0.057905009759271306,
+      "grad_norm": 0.39637815952301025,
+      "learning_rate": 0.0001,
+      "loss": 2.8006,
+      "step": 89
+    },
+    {
+      "epoch": 0.05855562784645413,
+      "grad_norm": 0.2676962614059448,
+      "learning_rate": 0.0001,
+      "loss": 2.2384,
+      "step": 90
+    },
+    {
+      "epoch": 0.05920624593363696,
+      "grad_norm": 0.3044937252998352,
+      "learning_rate": 0.0001,
+      "loss": 2.7762,
+      "step": 91
+    },
+    {
+      "epoch": 0.05985686402081978,
+      "grad_norm": 0.23922136425971985,
+      "learning_rate": 0.0001,
+      "loss": 2.0873,
+      "step": 92
+    },
+    {
+      "epoch": 0.0605074821080026,
+      "grad_norm": 0.25385046005249023,
+      "learning_rate": 0.0001,
+      "loss": 2.2708,
+      "step": 93
+    },
+    {
+      "epoch": 0.06115810019518542,
+      "grad_norm": 0.378401517868042,
+      "learning_rate": 0.0001,
+      "loss": 3.0583,
+      "step": 94
+    },
+    {
+      "epoch": 0.06180871828236825,
+      "grad_norm": 0.37193092703819275,
+      "learning_rate": 0.0001,
+      "loss": 2.3632,
+      "step": 95
+    },
+    {
+      "epoch": 0.062459336369551074,
+      "grad_norm": 0.3757643699645996,
+      "learning_rate": 0.0001,
+      "loss": 2.4071,
+      "step": 96
+    },
+    {
+      "epoch": 0.0631099544567339,
+      "grad_norm": 0.272833913564682,
+      "learning_rate": 0.0001,
+      "loss": 2.3989,
+      "step": 97
+    },
+    {
+      "epoch": 0.06376057254391672,
+      "grad_norm": 0.26533326506614685,
+      "learning_rate": 0.0001,
+      "loss": 2.1716,
+      "step": 98
+    },
+    {
+      "epoch": 0.06441119063109954,
+      "grad_norm": 0.5787199139595032,
+      "learning_rate": 0.0001,
+      "loss": 2.9445,
+      "step": 99
+    },
+    {
+      "epoch": 0.06506180871828236,
+      "grad_norm": 0.29046157002449036,
+      "learning_rate": 0.0001,
+      "loss": 2.3325,
+      "step": 100
+    },
+    {
+      "epoch": 0.06571242680546518,
+      "grad_norm": 0.531452476978302,
+      "learning_rate": 0.0001,
+      "loss": 2.7445,
+      "step": 101
+    },
+    {
+      "epoch": 0.06636304489264802,
+      "grad_norm": 0.3969165086746216,
+      "learning_rate": 0.0001,
+      "loss": 2.7126,
+      "step": 102
+    },
+    {
+      "epoch": 0.06701366297983084,
+      "grad_norm": 0.24183356761932373,
+      "learning_rate": 0.0001,
+      "loss": 1.9971,
+      "step": 103
+    },
+    {
+      "epoch": 0.06766428106701367,
+      "grad_norm": 0.3268399238586426,
+      "learning_rate": 0.0001,
+      "loss": 2.1055,
+      "step": 104
+    },
+    {
+      "epoch": 0.06831489915419649,
+      "grad_norm": 0.2625877559185028,
+      "learning_rate": 0.0001,
+      "loss": 1.9946,
+      "step": 105
+    },
+    {
+      "epoch": 0.06896551724137931,
+      "grad_norm": 0.2720443308353424,
+      "learning_rate": 0.0001,
+      "loss": 2.0764,
+      "step": 106
+    },
+    {
+      "epoch": 0.06961613532856213,
+      "grad_norm": 0.20969334244728088,
+      "learning_rate": 0.0001,
+      "loss": 1.8687,
+      "step": 107
+    },
+    {
+      "epoch": 0.07026675341574495,
+      "grad_norm": 0.26211223006248474,
+      "learning_rate": 0.0001,
+      "loss": 2.2042,
+      "step": 108
+    },
+    {
+      "epoch": 0.07091737150292778,
+      "grad_norm": 0.27889683842658997,
+      "learning_rate": 0.0001,
+      "loss": 2.3146,
+      "step": 109
+    },
+    {
+      "epoch": 0.07156798959011061,
+      "grad_norm": 0.2657179832458496,
+      "learning_rate": 0.0001,
+      "loss": 2.1021,
+      "step": 110
+    },
+    {
+      "epoch": 0.07221860767729343,
+      "grad_norm": 0.26620885729789734,
+      "learning_rate": 0.0001,
+      "loss": 2.3488,
+      "step": 111
+    },
+    {
+      "epoch": 0.07286922576447626,
+      "grad_norm": 0.4223373830318451,
+      "learning_rate": 0.0001,
+      "loss": 2.5289,
+      "step": 112
+    },
+    {
+      "epoch": 0.07351984385165908,
+      "grad_norm": 0.35398781299591064,
+      "learning_rate": 0.0001,
+      "loss": 2.5702,
+      "step": 113
+    },
+    {
+      "epoch": 0.0741704619388419,
+      "grad_norm": 0.23328129947185516,
+      "learning_rate": 0.0001,
+      "loss": 2.1292,
+      "step": 114
+    },
+    {
+      "epoch": 0.07482108002602472,
+      "grad_norm": 0.33508536219596863,
+      "learning_rate": 0.0001,
+      "loss": 2.2049,
+      "step": 115
+    },
+    {
+      "epoch": 0.07547169811320754,
+      "grad_norm": 0.2646953761577606,
+      "learning_rate": 0.0001,
+      "loss": 2.3445,
+      "step": 116
+    },
+    {
+      "epoch": 0.07612231620039037,
+      "grad_norm": 0.27866706252098083,
+      "learning_rate": 0.0001,
+      "loss": 2.2472,
+      "step": 117
+    },
+    {
+      "epoch": 0.07677293428757319,
+      "grad_norm": 0.35688602924346924,
+      "learning_rate": 0.0001,
+      "loss": 2.5045,
+      "step": 118
+    },
+    {
+      "epoch": 0.07742355237475602,
+      "grad_norm": 0.24262933433055878,
+      "learning_rate": 0.0001,
+      "loss": 2.4565,
+      "step": 119
+    },
+    {
+      "epoch": 0.07807417046193885,
+      "grad_norm": 0.44757333397865295,
+      "learning_rate": 0.0001,
+      "loss": 2.1619,
+      "step": 120
+    },
+    {
+      "epoch": 0.07872478854912167,
+      "grad_norm": 0.3279111385345459,
+      "learning_rate": 0.0001,
+      "loss": 2.3996,
+      "step": 121
+    },
+    {
+      "epoch": 0.07937540663630449,
+      "grad_norm": 0.25862693786621094,
+      "learning_rate": 0.0001,
+      "loss": 2.3214,
+      "step": 122
+    },
+    {
+      "epoch": 0.08002602472348731,
+      "grad_norm": 0.30093592405319214,
+      "learning_rate": 0.0001,
+      "loss": 2.6446,
+      "step": 123
+    },
+    {
+      "epoch": 0.08067664281067013,
+      "grad_norm": 0.25440871715545654,
+      "learning_rate": 0.0001,
+      "loss": 2.1181,
+      "step": 124
+    },
+    {
+      "epoch": 0.08132726089785296,
+      "grad_norm": 0.19935627281665802,
+      "learning_rate": 0.0001,
+      "loss": 2.0904,
+      "step": 125
+    },
+    {
+      "epoch": 0.08197787898503578,
+      "grad_norm": 0.27385473251342773,
+      "learning_rate": 0.0001,
+      "loss": 2.0829,
+      "step": 126
+    },
+    {
+      "epoch": 0.0826284970722186,
+      "grad_norm": 0.24417711794376373,
+      "learning_rate": 0.0001,
+      "loss": 2.0019,
+      "step": 127
+    },
+    {
+      "epoch": 0.08327911515940144,
+      "grad_norm": 0.27386653423309326,
+      "learning_rate": 0.0001,
+      "loss": 2.2743,
+      "step": 128
+    },
+    {
+      "epoch": 0.08392973324658426,
+      "grad_norm": 0.22413575649261475,
+      "learning_rate": 0.0001,
+      "loss": 2.1584,
+      "step": 129
+    },
+    {
+      "epoch": 0.08458035133376708,
+      "grad_norm": 0.27748343348503113,
+      "learning_rate": 0.0001,
+      "loss": 2.1428,
+      "step": 130
+    },
+    {
+      "epoch": 0.0852309694209499,
+      "grad_norm": 0.18890976905822754,
+      "learning_rate": 0.0001,
+      "loss": 1.9474,
+      "step": 131
+    },
+    {
+      "epoch": 0.08588158750813273,
+      "grad_norm": 0.3067719340324402,
+      "learning_rate": 0.0001,
+      "loss": 2.287,
+      "step": 132
+    },
+    {
+      "epoch": 0.08653220559531555,
+      "grad_norm": 0.35126858949661255,
+      "learning_rate": 0.0001,
+      "loss": 2.5086,
+      "step": 133
+    },
+    {
+      "epoch": 0.08718282368249837,
+      "grad_norm": 0.19619591534137726,
+      "learning_rate": 0.0001,
+      "loss": 2.0132,
+      "step": 134
+    },
+    {
+      "epoch": 0.08783344176968119,
+      "grad_norm": 0.360569566488266,
+      "learning_rate": 0.0001,
+      "loss": 2.607,
+      "step": 135
+    },
+    {
+      "epoch": 0.08848405985686401,
+      "grad_norm": 0.22566738724708557,
+      "learning_rate": 0.0001,
+      "loss": 2.0942,
+      "step": 136
+    },
+    {
+      "epoch": 0.08913467794404685,
+      "grad_norm": 0.27346086502075195,
+      "learning_rate": 0.0001,
+      "loss": 2.3139,
+      "step": 137
+    },
+    {
+      "epoch": 0.08978529603122967,
+      "grad_norm": 0.2500152289867401,
+      "learning_rate": 0.0001,
+      "loss": 2.0815,
+      "step": 138
+    },
+    {
+      "epoch": 0.0904359141184125,
+      "grad_norm": 0.22101153433322906,
+      "learning_rate": 0.0001,
+      "loss": 2.374,
+      "step": 139
+    },
+    {
+      "epoch": 0.09108653220559532,
+      "grad_norm": 0.2173723727464676,
+      "learning_rate": 0.0001,
+      "loss": 2.0084,
+      "step": 140
+    },
+    {
+      "epoch": 0.09173715029277814,
+      "grad_norm": 0.28956499695777893,
+      "learning_rate": 0.0001,
+      "loss": 2.6283,
+      "step": 141
+    },
+    {
+      "epoch": 0.09238776837996096,
+      "grad_norm": 0.27032795548439026,
+      "learning_rate": 0.0001,
+      "loss": 2.142,
+      "step": 142
+    },
+    {
+      "epoch": 0.09303838646714378,
+      "grad_norm": 0.24320480227470398,
+      "learning_rate": 0.0001,
+      "loss": 2.1402,
+      "step": 143
+    },
+    {
+      "epoch": 0.0936890045543266,
+      "grad_norm": 0.3127799332141876,
+      "learning_rate": 0.0001,
+      "loss": 2.6671,
+      "step": 144
+    },
+    {
+      "epoch": 0.09433962264150944,
+      "grad_norm": 0.30706024169921875,
+      "learning_rate": 0.0001,
+      "loss": 2.3026,
+      "step": 145
+    },
+    {
+      "epoch": 0.09499024072869226,
+      "grad_norm": 0.2378646731376648,
+      "learning_rate": 0.0001,
+      "loss": 2.0422,
+      "step": 146
+    },
+    {
+      "epoch": 0.09564085881587508,
+      "grad_norm": 0.24755406379699707,
+      "learning_rate": 0.0001,
+      "loss": 2.2574,
+      "step": 147
+    },
+    {
+      "epoch": 0.09629147690305791,
+      "grad_norm": 0.34464696049690247,
+      "learning_rate": 0.0001,
+      "loss": 2.2817,
+      "step": 148
+    },
+    {
+      "epoch": 0.09694209499024073,
+      "grad_norm": 0.30485469102859497,
+      "learning_rate": 0.0001,
+      "loss": 2.7303,
+      "step": 149
+    },
+    {
+      "epoch": 0.09759271307742355,
+      "grad_norm": 0.1860698163509369,
+      "learning_rate": 0.0001,
+      "loss": 1.8582,
+      "step": 150
+    },
+    {
+      "epoch": 0.09824333116460637,
+      "grad_norm": 0.23853841423988342,
+      "learning_rate": 0.0001,
+      "loss": 2.1378,
+      "step": 151
+    },
+    {
+      "epoch": 0.0988939492517892,
+      "grad_norm": 0.20248261094093323,
+      "learning_rate": 0.0001,
+      "loss": 2.1888,
+      "step": 152
+    },
+    {
+      "epoch": 0.09954456733897202,
+      "grad_norm": 0.3582792282104492,
+      "learning_rate": 0.0001,
+      "loss": 2.6726,
+      "step": 153
+    },
+    {
+      "epoch": 0.10019518542615485,
+      "grad_norm": 0.2576686441898346,
+      "learning_rate": 0.0001,
+      "loss": 2.4494,
+      "step": 154
+    },
+    {
+      "epoch": 0.10084580351333768,
+      "grad_norm": 0.306029349565506,
+      "learning_rate": 0.0001,
+      "loss": 2.2273,
+      "step": 155
+    },
+    {
+      "epoch": 0.1014964216005205,
+      "grad_norm": 0.31375500559806824,
+      "learning_rate": 0.0001,
+      "loss": 2.2474,
+      "step": 156
+    },
+    {
+      "epoch": 0.10214703968770332,
+      "grad_norm": 0.253250390291214,
+      "learning_rate": 0.0001,
+      "loss": 2.0142,
+      "step": 157
+    },
+    {
+      "epoch": 0.10279765777488614,
+      "grad_norm": 0.3098273277282715,
+      "learning_rate": 0.0001,
+      "loss": 2.2516,
+      "step": 158
+    },
+    {
+      "epoch": 0.10344827586206896,
+      "grad_norm": 0.3239591717720032,
+      "learning_rate": 0.0001,
+      "loss": 2.2432,
+      "step": 159
+    },
+    {
+      "epoch": 0.10409889394925179,
+      "grad_norm": 0.24929773807525635,
+      "learning_rate": 0.0001,
+      "loss": 2.2495,
+      "step": 160
+    },
+    {
+      "epoch": 0.10474951203643461,
+      "grad_norm": 0.3203783929347992,
+      "learning_rate": 0.0001,
+      "loss": 2.68,
+      "step": 161
+    },
+    {
+      "epoch": 0.10540013012361743,
+      "grad_norm": 0.38844674825668335,
+      "learning_rate": 0.0001,
+      "loss": 2.7457,
+      "step": 162
+    },
+    {
+      "epoch": 0.10605074821080027,
+      "grad_norm": 0.21753644943237305,
+      "learning_rate": 0.0001,
+      "loss": 2.1284,
+      "step": 163
+    },
+    {
+      "epoch": 0.10670136629798309,
+      "grad_norm": 0.20610418915748596,
+      "learning_rate": 0.0001,
+      "loss": 1.8377,
+      "step": 164
+    },
+    {
+      "epoch": 0.10735198438516591,
+      "grad_norm": 0.3555772304534912,
+      "learning_rate": 0.0001,
+      "loss": 2.3599,
+      "step": 165
+    },
+    {
+      "epoch": 0.10800260247234873,
+      "grad_norm": 0.3971005380153656,
+      "learning_rate": 0.0001,
+      "loss": 2.2771,
+      "step": 166
+    },
+    {
+      "epoch": 0.10865322055953155,
+      "grad_norm": 0.28628769516944885,
+      "learning_rate": 0.0001,
+      "loss": 2.2438,
+      "step": 167
+    },
+    {
+      "epoch": 0.10930383864671438,
+      "grad_norm": 0.38728833198547363,
+      "learning_rate": 0.0001,
+      "loss": 2.4103,
+      "step": 168
+    },
+    {
+      "epoch": 0.1099544567338972,
+      "grad_norm": 0.26340189576148987,
+      "learning_rate": 0.0001,
+      "loss": 2.6832,
+      "step": 169
+    },
+    {
+      "epoch": 0.11060507482108002,
+      "grad_norm": 0.20119386911392212,
+      "learning_rate": 0.0001,
+      "loss": 1.9622,
+      "step": 170
+    },
+    {
+      "epoch": 0.11125569290826284,
+      "grad_norm": 0.2929171621799469,
+      "learning_rate": 0.0001,
+      "loss": 2.2762,
+      "step": 171
+    },
+    {
+      "epoch": 0.11190631099544568,
+      "grad_norm": 0.422146201133728,
+      "learning_rate": 0.0001,
+      "loss": 2.4015,
+      "step": 172
+    },
+    {
+      "epoch": 0.1125569290826285,
+      "grad_norm": 0.29050537943840027,
+      "learning_rate": 0.0001,
+      "loss": 2.4399,
+      "step": 173
+    },
+    {
+      "epoch": 0.11320754716981132,
+      "grad_norm": 0.2646816074848175,
+      "learning_rate": 0.0001,
+      "loss": 2.3058,
+      "step": 174
+    },
+    {
+      "epoch": 0.11385816525699415,
+      "grad_norm": 0.2643061578273773,
+      "learning_rate": 0.0001,
+      "loss": 2.1892,
+      "step": 175
+    },
+    {
+      "epoch": 0.11450878334417697,
+      "grad_norm": 0.5878323316574097,
+      "learning_rate": 0.0001,
+      "loss": 3.2198,
+      "step": 176
+    },
+    {
+      "epoch": 0.11515940143135979,
+      "grad_norm": 0.36881884932518005,
+      "learning_rate": 0.0001,
+      "loss": 2.4112,
+      "step": 177
+    },
+    {
+      "epoch": 0.11581001951854261,
+      "grad_norm": 0.25198304653167725,
+      "learning_rate": 0.0001,
+      "loss": 2.1667,
+      "step": 178
+    },
+    {
+      "epoch": 0.11646063760572543,
+      "grad_norm": 0.34164664149284363,
+      "learning_rate": 0.0001,
+      "loss": 2.6248,
+      "step": 179
+    },
+    {
+      "epoch": 0.11711125569290826,
+      "grad_norm": 0.41471973061561584,
+      "learning_rate": 0.0001,
+      "loss": 2.5616,
+      "step": 180
+    },
+    {
+      "epoch": 0.11776187378009109,
+      "grad_norm": 0.26372480392456055,
+      "learning_rate": 0.0001,
+      "loss": 2.2904,
+      "step": 181
+    },
+    {
+      "epoch": 0.11841249186727391,
+      "grad_norm": 0.2271176278591156,
+      "learning_rate": 0.0001,
+      "loss": 2.0312,
+      "step": 182
+    },
+    {
+      "epoch": 0.11906310995445674,
+      "grad_norm": 0.2106996774673462,
+      "learning_rate": 0.0001,
+      "loss": 1.9661,
+      "step": 183
+    },
+    {
+      "epoch": 0.11971372804163956,
+      "grad_norm": 0.22870291769504547,
+      "learning_rate": 0.0001,
+      "loss": 1.9052,
+      "step": 184
+    },
+    {
+      "epoch": 0.12036434612882238,
+      "grad_norm": 0.41253864765167236,
+      "learning_rate": 0.0001,
+      "loss": 2.3747,
+      "step": 185
+    },
+    {
+      "epoch": 0.1210149642160052,
+      "grad_norm": 0.3258817791938782,
+      "learning_rate": 0.0001,
+      "loss": 2.5401,
+      "step": 186
+    },
+    {
+      "epoch": 0.12166558230318802,
+      "grad_norm": 0.3461870551109314,
+      "learning_rate": 0.0001,
+      "loss": 2.8027,
+      "step": 187
+    },
+    {
+      "epoch": 0.12231620039037085,
+      "grad_norm": 0.3704046607017517,
+      "learning_rate": 0.0001,
+      "loss": 2.799,
+      "step": 188
+    },
+    {
+      "epoch": 0.12296681847755368,
+      "grad_norm": 0.30265969038009644,
+      "learning_rate": 0.0001,
+      "loss": 2.4287,
+      "step": 189
+    },
+    {
+      "epoch": 0.1236174365647365,
+      "grad_norm": 0.4215582013130188,
+      "learning_rate": 0.0001,
+      "loss": 2.6857,
+      "step": 190
+    },
+    {
+      "epoch": 0.12426805465191933,
+      "grad_norm": 0.3003520965576172,
+      "learning_rate": 0.0001,
+      "loss": 2.4155,
+      "step": 191
+    },
+    {
+      "epoch": 0.12491867273910215,
+      "grad_norm": 0.412749320268631,
+      "learning_rate": 0.0001,
+      "loss": 2.6352,
+      "step": 192
+    },
+    {
+      "epoch": 0.12556929082628496,
+      "grad_norm": 0.2772350013256073,
+      "learning_rate": 0.0001,
+      "loss": 2.2452,
+      "step": 193
+    },
+    {
+      "epoch": 0.1262199089134678,
+      "grad_norm": 0.21457143127918243,
+      "learning_rate": 0.0001,
+      "loss": 2.0172,
+      "step": 194
+    },
+    {
+      "epoch": 0.12687052700065063,
+      "grad_norm": 0.40995845198631287,
+      "learning_rate": 0.0001,
+      "loss": 2.6218,
+      "step": 195
+    },
+    {
+      "epoch": 0.12752114508783344,
+      "grad_norm": 0.2253209501504898,
+      "learning_rate": 0.0001,
+      "loss": 2.2319,
+      "step": 196
+    },
+    {
+      "epoch": 0.12817176317501627,
+      "grad_norm": 0.36564287543296814,
+      "learning_rate": 0.0001,
+      "loss": 2.4585,
+      "step": 197
+    },
+    {
+      "epoch": 0.12882238126219908,
+      "grad_norm": 0.41084784269332886,
+      "learning_rate": 0.0001,
+      "loss": 2.6326,
+      "step": 198
+    },
+    {
+      "epoch": 0.12947299934938192,
+      "grad_norm": 0.36012157797813416,
+      "learning_rate": 0.0001,
+      "loss": 2.0168,
+      "step": 199
+    },
+    {
+      "epoch": 0.13012361743656473,
+      "grad_norm": 0.5138425230979919,
+      "learning_rate": 0.0001,
+      "loss": 2.3377,
+      "step": 200
+    },
+    {
+      "epoch": 0.13077423552374756,
+      "grad_norm": 0.2799031436443329,
+      "learning_rate": 0.0001,
+      "loss": 2.532,
+      "step": 201
+    },
+    {
+      "epoch": 0.13142485361093037,
+      "grad_norm": 0.3078779876232147,
+      "learning_rate": 0.0001,
+      "loss": 2.044,
+      "step": 202
+    },
+    {
+      "epoch": 0.1320754716981132,
+      "grad_norm": 0.31270912289619446,
+      "learning_rate": 0.0001,
+      "loss": 1.8576,
+      "step": 203
+    },
+    {
+      "epoch": 0.13272608978529604,
+      "grad_norm": 0.23117204010486603,
+      "learning_rate": 0.0001,
+      "loss": 2.1908,
+      "step": 204
+    },
+    {
+      "epoch": 0.13337670787247885,
+      "grad_norm": 0.2531285285949707,
+      "learning_rate": 0.0001,
+      "loss": 2.143,
+      "step": 205
+    },
+    {
+      "epoch": 0.1340273259596617,
+      "grad_norm": 0.28053218126296997,
+      "learning_rate": 0.0001,
+      "loss": 2.6902,
+      "step": 206
+    },
+    {
+      "epoch": 0.1346779440468445,
+      "grad_norm": 0.2600589692592621,
+      "learning_rate": 0.0001,
+      "loss": 2.0355,
+      "step": 207
+    },
+    {
+      "epoch": 0.13532856213402733,
+      "grad_norm": 0.2725912630558014,
+      "learning_rate": 0.0001,
+      "loss": 2.3949,
+      "step": 208
+    },
+    {
+      "epoch": 0.13597918022121014,
+      "grad_norm": 0.6166338324546814,
+      "learning_rate": 0.0001,
+      "loss": 2.8146,
+      "step": 209
+    },
+    {
+      "epoch": 0.13662979830839297,
+      "grad_norm": 0.4028575122356415,
+      "learning_rate": 0.0001,
+      "loss": 2.888,
+      "step": 210
+    },
+    {
+      "epoch": 0.1372804163955758,
+      "grad_norm": 0.23181548714637756,
+      "learning_rate": 0.0001,
+      "loss": 2.1406,
+      "step": 211
+    },
+    {
+      "epoch": 0.13793103448275862,
+      "grad_norm": 0.24338063597679138,
+      "learning_rate": 0.0001,
+      "loss": 2.1564,
+      "step": 212
+    },
+    {
+      "epoch": 0.13858165256994145,
+      "grad_norm": 0.233146533370018,
+      "learning_rate": 0.0001,
+      "loss": 2.1695,
+      "step": 213
+    },
+    {
+      "epoch": 0.13923227065712426,
+      "grad_norm": 0.21236726641654968,
+      "learning_rate": 0.0001,
+      "loss": 1.9272,
+      "step": 214
+    },
+    {
+      "epoch": 0.1398828887443071,
+      "grad_norm": 0.25471317768096924,
+      "learning_rate": 0.0001,
+      "loss": 2.3447,
+      "step": 215
+    },
+    {
+      "epoch": 0.1405335068314899,
+      "grad_norm": 0.35532835125923157,
+      "learning_rate": 0.0001,
+      "loss": 2.4328,
+      "step": 216
+    },
+    {
+      "epoch": 0.14118412491867274,
+      "grad_norm": 0.32900944352149963,
+      "learning_rate": 0.0001,
+      "loss": 2.385,
+      "step": 217
+    },
+    {
+      "epoch": 0.14183474300585555,
+      "grad_norm": 0.45404863357543945,
+      "learning_rate": 0.0001,
+      "loss": 2.8053,
+      "step": 218
+    },
+    {
+      "epoch": 0.1424853610930384,
+      "grad_norm": 0.33968400955200195,
+      "learning_rate": 0.0001,
+      "loss": 2.4524,
+      "step": 219
+    },
+    {
+      "epoch": 0.14313597918022122,
+      "grad_norm": 0.3250170946121216,
+      "learning_rate": 0.0001,
+      "loss": 2.6173,
+      "step": 220
+    },
+    {
+      "epoch": 0.14378659726740403,
+      "grad_norm": 0.34765559434890747,
+      "learning_rate": 0.0001,
+      "loss": 2.8468,
+      "step": 221
+    },
+    {
+      "epoch": 0.14443721535458687,
+      "grad_norm": 0.2274564653635025,
+      "learning_rate": 0.0001,
+      "loss": 2.1305,
+      "step": 222
+    },
+    {
+      "epoch": 0.14508783344176968,
+      "grad_norm": 0.42719507217407227,
+      "learning_rate": 0.0001,
+      "loss": 2.3682,
+      "step": 223
+    },
+    {
+      "epoch": 0.1457384515289525,
+      "grad_norm": 0.2848481833934784,
+      "learning_rate": 0.0001,
+      "loss": 2.0923,
+      "step": 224
+    },
+    {
+      "epoch": 0.14638906961613532,
+      "grad_norm": 0.266548752784729,
+      "learning_rate": 0.0001,
+      "loss": 2.0393,
+      "step": 225
+    },
+    {
+      "epoch": 0.14703968770331816,
+      "grad_norm": 0.24076099693775177,
+      "learning_rate": 0.0001,
+      "loss": 2.2674,
+      "step": 226
+    },
+    {
+      "epoch": 0.14769030579050096,
+      "grad_norm": 0.23347622156143188,
+      "learning_rate": 0.0001,
+      "loss": 1.9455,
+      "step": 227
+    },
+    {
+      "epoch": 0.1483409238776838,
+      "grad_norm": 0.3925648033618927,
+      "learning_rate": 0.0001,
+      "loss": 2.7117,
+      "step": 228
+    },
+    {
+      "epoch": 0.14899154196486664,
+      "grad_norm": 0.27654924988746643,
+      "learning_rate": 0.0001,
+      "loss": 2.1306,
+      "step": 229
+    },
+    {
+      "epoch": 0.14964216005204944,
+      "grad_norm": 0.2853853702545166,
+      "learning_rate": 0.0001,
+      "loss": 2.4369,
+      "step": 230
+    },
+    {
+      "epoch": 0.15029277813923228,
+      "grad_norm": 0.4509859085083008,
+      "learning_rate": 0.0001,
+      "loss": 2.6047,
+      "step": 231
+    },
+    {
+      "epoch": 0.1509433962264151,
+      "grad_norm": 0.2515909671783447,
+      "learning_rate": 0.0001,
+      "loss": 2.2065,
+      "step": 232
+    },
+    {
+      "epoch": 0.15159401431359792,
+      "grad_norm": 0.5977367162704468,
+      "learning_rate": 0.0001,
+      "loss": 2.7133,
+      "step": 233
+    },
+    {
+      "epoch": 0.15224463240078073,
+      "grad_norm": 0.30381399393081665,
+      "learning_rate": 0.0001,
+      "loss": 2.343,
+      "step": 234
+    },
+    {
+      "epoch": 0.15289525048796357,
+      "grad_norm": 0.27204832434654236,
+      "learning_rate": 0.0001,
+      "loss": 2.2908,
+      "step": 235
+    },
+    {
+      "epoch": 0.15354586857514638,
+      "grad_norm": 0.6246710419654846,
+      "learning_rate": 0.0001,
+      "loss": 2.7862,
+      "step": 236
+    },
+    {
+      "epoch": 0.1541964866623292,
+      "grad_norm": 0.4803178012371063,
+      "learning_rate": 0.0001,
+      "loss": 3.4388,
+      "step": 237
+    },
+    {
+      "epoch": 0.15484710474951205,
+      "grad_norm": 0.3038940727710724,
+      "learning_rate": 0.0001,
+      "loss": 2.7409,
+      "step": 238
+    },
+    {
+      "epoch": 0.15549772283669486,
+      "grad_norm": 0.2494591474533081,
+      "learning_rate": 0.0001,
+      "loss": 2.2601,
+      "step": 239
+    },
+    {
+      "epoch": 0.1561483409238777,
+      "grad_norm": 0.23808616399765015,
+      "learning_rate": 0.0001,
+      "loss": 2.1319,
+      "step": 240
+    },
+    {
+      "epoch": 0.1567989590110605,
+      "grad_norm": 0.3111306130886078,
+      "learning_rate": 0.0001,
+      "loss": 2.7414,
+      "step": 241
+    },
+    {
+      "epoch": 0.15744957709824334,
+      "grad_norm": 0.22197599709033966,
+      "learning_rate": 0.0001,
+      "loss": 2.1346,
+      "step": 242
+    },
+    {
+      "epoch": 0.15810019518542615,
+      "grad_norm": 0.2681500315666199,
+      "learning_rate": 0.0001,
+      "loss": 2.3779,
+      "step": 243
+    },
+    {
+      "epoch": 0.15875081327260898,
+      "grad_norm": 0.2612643241882324,
+      "learning_rate": 0.0001,
+      "loss": 2.5743,
+      "step": 244
+    },
+    {
+      "epoch": 0.1594014313597918,
+      "grad_norm": 0.201397106051445,
+      "learning_rate": 0.0001,
+      "loss": 2.0312,
+      "step": 245
+    },
+    {
+      "epoch": 0.16005204944697463,
+      "grad_norm": 0.25662410259246826,
+      "learning_rate": 0.0001,
+      "loss": 2.5085,
+      "step": 246
+    },
+    {
+      "epoch": 0.16070266753415746,
+      "grad_norm": 0.21460294723510742,
+      "learning_rate": 0.0001,
+      "loss": 2.1099,
+      "step": 247
+    },
+    {
+      "epoch": 0.16135328562134027,
+      "grad_norm": 0.19971312582492828,
+      "learning_rate": 0.0001,
+      "loss": 2.1024,
+      "step": 248
+    },
+    {
+      "epoch": 0.1620039037085231,
+      "grad_norm": 0.1986059844493866,
+      "learning_rate": 0.0001,
+      "loss": 1.9306,
+      "step": 249
+    },
+    {
+      "epoch": 0.16265452179570591,
+      "grad_norm": 0.21961884200572968,
+      "learning_rate": 0.0001,
+      "loss": 2.1218,
+      "step": 250
+    },
+    {
+      "epoch": 0.16330513988288875,
+      "grad_norm": 0.20071017742156982,
+      "learning_rate": 0.0001,
+      "loss": 2.0581,
+      "step": 251
+    },
+    {
+      "epoch": 0.16395575797007156,
+      "grad_norm": 0.32734909653663635,
+      "learning_rate": 0.0001,
+      "loss": 2.6229,
+      "step": 252
+    },
+    {
+      "epoch": 0.1646063760572544,
+      "grad_norm": 0.21822451055049896,
+      "learning_rate": 0.0001,
+      "loss": 1.9954,
+      "step": 253
+    },
+    {
+      "epoch": 0.1652569941444372,
+      "grad_norm": 0.3013177216053009,
+      "learning_rate": 0.0001,
+      "loss": 2.454,
+      "step": 254
+    },
+    {
+      "epoch": 0.16590761223162004,
+      "grad_norm": 0.31199347972869873,
+      "learning_rate": 0.0001,
+      "loss": 2.815,
+      "step": 255
+    },
+    {
+      "epoch": 0.16655823031880287,
+      "grad_norm": 0.2255464345216751,
+      "learning_rate": 0.0001,
+      "loss": 2.0232,
+      "step": 256
+    },
+    {
+      "epoch": 0.16720884840598568,
+      "grad_norm": 0.21208804845809937,
+      "learning_rate": 0.0001,
+      "loss": 1.9663,
+      "step": 257
+    },
+    {
+      "epoch": 0.16785946649316852,
+      "grad_norm": 0.2432132512331009,
+      "learning_rate": 0.0001,
+      "loss": 2.4189,
+      "step": 258
+    },
+    {
+      "epoch": 0.16851008458035133,
+      "grad_norm": 0.21116623282432556,
+      "learning_rate": 0.0001,
+      "loss": 2.0761,
+      "step": 259
+    },
+    {
+      "epoch": 0.16916070266753416,
+      "grad_norm": 0.18722975254058838,
+      "learning_rate": 0.0001,
+      "loss": 1.9537,
+      "step": 260
+    },
+    {
+      "epoch": 0.16981132075471697,
+      "grad_norm": 0.2683362662792206,
+      "learning_rate": 0.0001,
+      "loss": 2.4483,
+      "step": 261
+    },
+    {
+      "epoch": 0.1704619388418998,
+      "grad_norm": 0.2739648222923279,
+      "learning_rate": 0.0001,
+      "loss": 2.3754,
+      "step": 262
+    },
+    {
+      "epoch": 0.17111255692908262,
+      "grad_norm": 0.1836375594139099,
+      "learning_rate": 0.0001,
+      "loss": 2.0103,
+      "step": 263
+    },
+    {
+      "epoch": 0.17176317501626545,
+      "grad_norm": 0.34002602100372314,
+      "learning_rate": 0.0001,
+      "loss": 2.2626,
+      "step": 264
+    },
+    {
+      "epoch": 0.1724137931034483,
+      "grad_norm": 0.19341516494750977,
+      "learning_rate": 0.0001,
+      "loss": 1.9751,
+      "step": 265
+    },
+    {
+      "epoch": 0.1730644111906311,
+      "grad_norm": 0.25080743432044983,
+      "learning_rate": 0.0001,
+      "loss": 2.2162,
+      "step": 266
+    },
+    {
+      "epoch": 0.17371502927781393,
+      "grad_norm": 0.2362661212682724,
+      "learning_rate": 0.0001,
+      "loss": 2.0226,
+      "step": 267
+    },
+    {
+      "epoch": 0.17436564736499674,
+      "grad_norm": 0.25844064354896545,
+      "learning_rate": 0.0001,
+      "loss": 2.3176,
+      "step": 268
+    },
+    {
+      "epoch": 0.17501626545217958,
+      "grad_norm": 0.3904498517513275,
+      "learning_rate": 0.0001,
+      "loss": 2.4871,
+      "step": 269
+    },
+    {
+      "epoch": 0.17566688353936238,
+      "grad_norm": 0.22143317759037018,
+      "learning_rate": 0.0001,
+      "loss": 2.2073,
+      "step": 270
+    },
+    {
+      "epoch": 0.17631750162654522,
+      "grad_norm": 0.20974211394786835,
+      "learning_rate": 0.0001,
+      "loss": 2.1393,
+      "step": 271
+    },
+    {
+      "epoch": 0.17696811971372803,
+      "grad_norm": 0.24463056027889252,
+      "learning_rate": 0.0001,
+      "loss": 2.0203,
+      "step": 272
+    },
+    {
+      "epoch": 0.17761873780091086,
+      "grad_norm": 0.23296399414539337,
+      "learning_rate": 0.0001,
+      "loss": 2.1096,
+      "step": 273
+    },
+    {
+      "epoch": 0.1782693558880937,
+      "grad_norm": 0.4122619926929474,
+      "learning_rate": 0.0001,
+      "loss": 3.1512,
+      "step": 274
+    },
+    {
+      "epoch": 0.1789199739752765,
+      "grad_norm": 0.2744470536708832,
+      "learning_rate": 0.0001,
+      "loss": 2.2211,
+      "step": 275
+    },
+    {
+      "epoch": 0.17957059206245934,
+      "grad_norm": 0.21010619401931763,
+      "learning_rate": 0.0001,
+      "loss": 2.2203,
+      "step": 276
+    },
+    {
+      "epoch": 0.18022121014964215,
+      "grad_norm": 0.27855056524276733,
+      "learning_rate": 0.0001,
+      "loss": 2.2903,
+      "step": 277
+    },
+    {
+      "epoch": 0.180871828236825,
+      "grad_norm": 0.2909989058971405,
+      "learning_rate": 0.0001,
+      "loss": 2.237,
+      "step": 278
+    },
+    {
+      "epoch": 0.1815224463240078,
+      "grad_norm": 0.21754448115825653,
+      "learning_rate": 0.0001,
+      "loss": 2.0138,
+      "step": 279
+    },
+    {
+      "epoch": 0.18217306441119063,
+      "grad_norm": 0.35209745168685913,
+      "learning_rate": 0.0001,
+      "loss": 2.652,
+      "step": 280
+    },
+    {
+      "epoch": 0.18282368249837344,
+      "grad_norm": 0.29994750022888184,
+      "learning_rate": 0.0001,
+      "loss": 2.1868,
+      "step": 281
+    },
+    {
+      "epoch": 0.18347430058555628,
+      "grad_norm": 0.2645902633666992,
+      "learning_rate": 0.0001,
+      "loss": 2.2925,
+      "step": 282
+    },
+    {
+      "epoch": 0.1841249186727391,
+      "grad_norm": 0.3492202162742615,
+      "learning_rate": 0.0001,
+      "loss": 2.4176,
+      "step": 283
+    },
+    {
+      "epoch": 0.18477553675992192,
+      "grad_norm": 0.256651371717453,
+      "learning_rate": 0.0001,
+      "loss": 2.3414,
+      "step": 284
+    },
+    {
+      "epoch": 0.18542615484710476,
+      "grad_norm": 0.23287786543369293,
+      "learning_rate": 0.0001,
+      "loss": 2.5488,
+      "step": 285
+    },
+    {
+      "epoch": 0.18607677293428757,
+      "grad_norm": 0.26059290766716003,
+      "learning_rate": 0.0001,
+      "loss": 2.4551,
+      "step": 286
+    },
+    {
+      "epoch": 0.1867273910214704,
+      "grad_norm": 0.2482365071773529,
+      "learning_rate": 0.0001,
+      "loss": 2.0818,
+      "step": 287
+    },
+    {
+      "epoch": 0.1873780091086532,
+      "grad_norm": 0.23024773597717285,
+      "learning_rate": 0.0001,
+      "loss": 2.2592,
+      "step": 288
+    },
+    {
+      "epoch": 0.18802862719583605,
+      "grad_norm": 0.2590011656284332,
+      "learning_rate": 0.0001,
+      "loss": 2.4177,
+      "step": 289
+    },
+    {
+      "epoch": 0.18867924528301888,
+      "grad_norm": 0.19760870933532715,
+      "learning_rate": 0.0001,
+      "loss": 2.0731,
+      "step": 290
+    },
+    {
+      "epoch": 0.1893298633702017,
+      "grad_norm": 0.20266428589820862,
+      "learning_rate": 0.0001,
+      "loss": 2.1221,
+      "step": 291
+    },
+    {
+      "epoch": 0.18998048145738453,
+      "grad_norm": 0.20199884474277496,
+      "learning_rate": 0.0001,
+      "loss": 2.0489,
+      "step": 292
+    },
+    {
+      "epoch": 0.19063109954456733,
+      "grad_norm": 0.23876360058784485,
+      "learning_rate": 0.0001,
+      "loss": 2.1392,
+      "step": 293
+    },
+    {
+      "epoch": 0.19128171763175017,
+      "grad_norm": 0.23555997014045715,
+      "learning_rate": 0.0001,
+      "loss": 2.4116,
+      "step": 294
+    },
+    {
+      "epoch": 0.19193233571893298,
+      "grad_norm": 0.5010725259780884,
+      "learning_rate": 0.0001,
+      "loss": 2.7444,
+      "step": 295
+    },
+    {
+      "epoch": 0.19258295380611581,
+      "grad_norm": 0.37809622287750244,
+      "learning_rate": 0.0001,
+      "loss": 2.2635,
+      "step": 296
+    },
+    {
+      "epoch": 0.19323357189329862,
+      "grad_norm": 0.499888151884079,
+      "learning_rate": 0.0001,
+      "loss": 2.1984,
+      "step": 297
+    },
+    {
+      "epoch": 0.19388418998048146,
+      "grad_norm": 0.43810585141181946,
+      "learning_rate": 0.0001,
+      "loss": 3.084,
+      "step": 298
+    },
+    {
+      "epoch": 0.1945348080676643,
+      "grad_norm": 0.35633769631385803,
+      "learning_rate": 0.0001,
+      "loss": 2.0351,
+      "step": 299
+    },
+    {
+      "epoch": 0.1951854261548471,
+      "grad_norm": 0.3693079650402069,
+      "learning_rate": 0.0001,
+      "loss": 1.9525,
+      "step": 300
+    },
+    {
+      "epoch": 0.19583604424202994,
+      "grad_norm": 0.36550503969192505,
+      "learning_rate": 0.0001,
+      "loss": 2.2469,
+      "step": 301
+    },
+    {
+      "epoch": 0.19648666232921275,
+      "grad_norm": 0.2579827308654785,
+      "learning_rate": 0.0001,
+      "loss": 2.3585,
+      "step": 302
+    },
+    {
+      "epoch": 0.19713728041639558,
+      "grad_norm": 0.2603841722011566,
+      "learning_rate": 0.0001,
+      "loss": 2.3959,
+      "step": 303
+    },
+    {
+      "epoch": 0.1977878985035784,
+      "grad_norm": 0.33103683590888977,
+      "learning_rate": 0.0001,
+      "loss": 2.2197,
+      "step": 304
+    },
+    {
+      "epoch": 0.19843851659076123,
+      "grad_norm": 0.2977697551250458,
+      "learning_rate": 0.0001,
+      "loss": 2.2569,
+      "step": 305
+    },
+    {
+      "epoch": 0.19908913467794404,
+      "grad_norm": 0.2085130512714386,
+      "learning_rate": 0.0001,
+      "loss": 2.2284,
+      "step": 306
+    },
+    {
+      "epoch": 0.19973975276512687,
+      "grad_norm": 0.409212201833725,
+      "learning_rate": 0.0001,
+      "loss": 2.7014,
+      "step": 307
+    },
+    {
+      "epoch": 0.2003903708523097,
+      "grad_norm": 0.2447553277015686,
+      "learning_rate": 0.0001,
+      "loss": 2.2826,
+      "step": 308
+    },
+    {
+      "epoch": 0.20104098893949252,
+      "grad_norm": 0.21881726384162903,
+      "learning_rate": 0.0001,
+      "loss": 1.8573,
+      "step": 309
+    },
+    {
+      "epoch": 0.20169160702667535,
+      "grad_norm": 0.24484936892986298,
+      "learning_rate": 0.0001,
+      "loss": 2.318,
+      "step": 310
+    },
+    {
+      "epoch": 0.20234222511385816,
+      "grad_norm": 0.3251173198223114,
+      "learning_rate": 0.0001,
+      "loss": 2.3346,
+      "step": 311
+    },
+    {
+      "epoch": 0.202992843201041,
+      "grad_norm": 0.22313712537288666,
+      "learning_rate": 0.0001,
+      "loss": 1.9119,
+      "step": 312
+    },
+    {
+      "epoch": 0.2036434612882238,
+      "grad_norm": 0.3086949288845062,
+      "learning_rate": 0.0001,
+      "loss": 2.1809,
+      "step": 313
+    },
+    {
+      "epoch": 0.20429407937540664,
+      "grad_norm": 0.28272122144699097,
+      "learning_rate": 0.0001,
+      "loss": 2.3335,
+      "step": 314
+    },
+    {
+      "epoch": 0.20494469746258945,
+      "grad_norm": 0.208637535572052,
+      "learning_rate": 0.0001,
+      "loss": 2.1947,
+      "step": 315
+    },
+    {
+      "epoch": 0.20559531554977228,
+      "grad_norm": 0.2913041114807129,
+      "learning_rate": 0.0001,
+      "loss": 2.3009,
+      "step": 316
+    },
+    {
+      "epoch": 0.20624593363695512,
+      "grad_norm": 0.2813785970211029,
+      "learning_rate": 0.0001,
+      "loss": 2.0133,
+      "step": 317
+    },
+    {
+      "epoch": 0.20689655172413793,
+      "grad_norm": 0.2324337363243103,
+      "learning_rate": 0.0001,
+      "loss": 2.0827,
+      "step": 318
+    },
+    {
+      "epoch": 0.20754716981132076,
+      "grad_norm": 0.25195491313934326,
+      "learning_rate": 0.0001,
+      "loss": 2.5201,
+      "step": 319
+    },
+    {
+      "epoch": 0.20819778789850357,
+      "grad_norm": 0.3435034453868866,
+      "learning_rate": 0.0001,
+      "loss": 2.321,
+      "step": 320
+    },
+    {
+      "epoch": 0.2088484059856864,
+      "grad_norm": 0.2735581696033478,
+      "learning_rate": 0.0001,
+      "loss": 2.2218,
+      "step": 321
+    },
+    {
+      "epoch": 0.20949902407286922,
+      "grad_norm": 0.2250661551952362,
+      "learning_rate": 0.0001,
+      "loss": 1.9416,
+      "step": 322
+    },
+    {
+      "epoch": 0.21014964216005205,
+      "grad_norm": 0.3160262107849121,
+      "learning_rate": 0.0001,
+      "loss": 2.5494,
+      "step": 323
+    },
+    {
+      "epoch": 0.21080026024723486,
+      "grad_norm": 0.3669279217720032,
+      "learning_rate": 0.0001,
+      "loss": 2.7751,
+      "step": 324
+    },
+    {
+      "epoch": 0.2114508783344177,
+      "grad_norm": 0.2052752673625946,
+      "learning_rate": 0.0001,
+      "loss": 2.0139,
+      "step": 325
+    },
+    {
+      "epoch": 0.21210149642160053,
+      "grad_norm": 0.2906612455844879,
+      "learning_rate": 0.0001,
+      "loss": 2.227,
+      "step": 326
+    },
+    {
+      "epoch": 0.21275211450878334,
+      "grad_norm": 0.30327048897743225,
+      "learning_rate": 0.0001,
+      "loss": 2.2905,
+      "step": 327
+    },
+    {
+      "epoch": 0.21340273259596618,
+      "grad_norm": 0.33950623869895935,
+      "learning_rate": 0.0001,
+      "loss": 3.0731,
+      "step": 328
+    },
+    {
+      "epoch": 0.21405335068314899,
+      "grad_norm": 0.31319788098335266,
+      "learning_rate": 0.0001,
+      "loss": 2.1374,
+      "step": 329
+    },
+    {
+      "epoch": 0.21470396877033182,
+      "grad_norm": 0.21442054212093353,
+      "learning_rate": 0.0001,
+      "loss": 1.7588,
+      "step": 330
+    },
+    {
+      "epoch": 0.21535458685751463,
+      "grad_norm": 0.23125174641609192,
+      "learning_rate": 0.0001,
+      "loss": 1.9295,
+      "step": 331
+    },
+    {
+      "epoch": 0.21600520494469747,
+      "grad_norm": 0.23220308125019073,
+      "learning_rate": 0.0001,
+      "loss": 2.2606,
+      "step": 332
+    },
+    {
+      "epoch": 0.21665582303188027,
+      "grad_norm": 0.24599219858646393,
+      "learning_rate": 0.0001,
+      "loss": 2.2687,
+      "step": 333
+    },
+    {
+      "epoch": 0.2173064411190631,
+      "grad_norm": 0.22226236760616302,
+      "learning_rate": 0.0001,
+      "loss": 2.1428,
+      "step": 334
+    },
+    {
+      "epoch": 0.21795705920624595,
+      "grad_norm": 0.2653510570526123,
+      "learning_rate": 0.0001,
+      "loss": 2.4381,
+      "step": 335
+    },
+    {
+      "epoch": 0.21860767729342875,
+      "grad_norm": 0.23770929872989655,
+      "learning_rate": 0.0001,
+      "loss": 1.9655,
+      "step": 336
+    },
+    {
+      "epoch": 0.2192582953806116,
+      "grad_norm": 0.1932332068681717,
+      "learning_rate": 0.0001,
+      "loss": 1.9465,
+      "step": 337
+    },
+    {
+      "epoch": 0.2199089134677944,
+      "grad_norm": 0.181661456823349,
+      "learning_rate": 0.0001,
+      "loss": 1.9912,
+      "step": 338
+    },
+    {
+      "epoch": 0.22055953155497723,
+      "grad_norm": 0.22275297343730927,
+      "learning_rate": 0.0001,
+      "loss": 2.1964,
+      "step": 339
+    },
+    {
+      "epoch": 0.22121014964216004,
+      "grad_norm": 0.22086840867996216,
+      "learning_rate": 0.0001,
+      "loss": 2.2216,
+      "step": 340
+    },
+    {
+      "epoch": 0.22186076772934288,
+      "grad_norm": 0.22807130217552185,
+      "learning_rate": 0.0001,
+      "loss": 2.2434,
+      "step": 341
+    },
+    {
+      "epoch": 0.2225113858165257,
+      "grad_norm": 0.26616647839546204,
+      "learning_rate": 0.0001,
+      "loss": 2.442,
+      "step": 342
+    },
+    {
+      "epoch": 0.22316200390370852,
+      "grad_norm": 0.2841719388961792,
+      "learning_rate": 0.0001,
+      "loss": 2.2358,
+      "step": 343
+    },
+    {
+      "epoch": 0.22381262199089136,
+      "grad_norm": 0.23251943290233612,
+      "learning_rate": 0.0001,
+      "loss": 2.3436,
+      "step": 344
+    },
+    {
+      "epoch": 0.22446324007807417,
+      "grad_norm": 0.20406994223594666,
+      "learning_rate": 0.0001,
+      "loss": 2.101,
+      "step": 345
+    },
+    {
+      "epoch": 0.225113858165257,
+      "grad_norm": 0.18677304685115814,
+      "learning_rate": 0.0001,
+      "loss": 2.0596,
+      "step": 346
+    },
+    {
+      "epoch": 0.2257644762524398,
+      "grad_norm": 0.22367873787879944,
+      "learning_rate": 0.0001,
+      "loss": 2.2051,
+      "step": 347
+    },
+    {
+      "epoch": 0.22641509433962265,
+      "grad_norm": 0.2521246671676636,
+      "learning_rate": 0.0001,
+      "loss": 2.1718,
+      "step": 348
+    },
+    {
+      "epoch": 0.22706571242680545,
+      "grad_norm": 0.23043319582939148,
+      "learning_rate": 0.0001,
+      "loss": 2.2818,
+      "step": 349
+    },
+    {
+      "epoch": 0.2277163305139883,
+      "grad_norm": 0.22021251916885376,
+      "learning_rate": 0.0001,
+      "loss": 2.0337,
+      "step": 350
+    },
+    {
+      "epoch": 0.2283669486011711,
+      "grad_norm": 0.18043603003025055,
+      "learning_rate": 0.0001,
+      "loss": 1.9434,
+      "step": 351
+    },
+    {
+      "epoch": 0.22901756668835394,
+      "grad_norm": 0.4757142961025238,
+      "learning_rate": 0.0001,
+      "loss": 2.2467,
+      "step": 352
+    },
+    {
+      "epoch": 0.22966818477553677,
+      "grad_norm": 0.30740290880203247,
+      "learning_rate": 0.0001,
+      "loss": 2.5296,
+      "step": 353
+    },
+    {
+      "epoch": 0.23031880286271958,
+      "grad_norm": 0.23037666082382202,
+      "learning_rate": 0.0001,
+      "loss": 2.311,
+      "step": 354
+    },
+    {
+      "epoch": 0.23096942094990242,
+      "grad_norm": 0.22314564883708954,
+      "learning_rate": 0.0001,
+      "loss": 2.0494,
+      "step": 355
+    },
+    {
+      "epoch": 0.23162003903708522,
+      "grad_norm": 0.21417242288589478,
+      "learning_rate": 0.0001,
+      "loss": 2.2459,
+      "step": 356
+    },
+    {
+      "epoch": 0.23227065712426806,
+      "grad_norm": 0.2895831763744354,
+      "learning_rate": 0.0001,
+      "loss": 2.2705,
+      "step": 357
+    },
+    {
+      "epoch": 0.23292127521145087,
+      "grad_norm": 0.2110838145017624,
+      "learning_rate": 0.0001,
+      "loss": 2.1175,
+      "step": 358
+    },
+    {
+      "epoch": 0.2335718932986337,
+      "grad_norm": 0.3999682664871216,
+      "learning_rate": 0.0001,
+      "loss": 2.6891,
+      "step": 359
+    },
+    {
+      "epoch": 0.2342225113858165,
+      "grad_norm": 0.5169201493263245,
+      "learning_rate": 0.0001,
+      "loss": 2.5764,
+      "step": 360
+    },
+    {
+      "epoch": 0.23487312947299935,
+      "grad_norm": 0.24382548034191132,
+      "learning_rate": 0.0001,
+      "loss": 2.1065,
+      "step": 361
+    },
+    {
+      "epoch": 0.23552374756018218,
+      "grad_norm": 0.2830081582069397,
+      "learning_rate": 0.0001,
+      "loss": 2.1186,
+      "step": 362
+    },
+    {
+      "epoch": 0.236174365647365,
+      "grad_norm": 0.23680554330348969,
+      "learning_rate": 0.0001,
+      "loss": 2.118,
+      "step": 363
+    },
+    {
+      "epoch": 0.23682498373454783,
+      "grad_norm": 0.3790690302848816,
+      "learning_rate": 0.0001,
+      "loss": 2.3566,
+      "step": 364
+    },
+    {
+      "epoch": 0.23747560182173064,
+      "grad_norm": 0.2664685845375061,
+      "learning_rate": 0.0001,
+      "loss": 2.2118,
+      "step": 365
+    },
+    {
+      "epoch": 0.23812621990891347,
+      "grad_norm": 0.22439126670360565,
+      "learning_rate": 0.0001,
+      "loss": 2.0897,
+      "step": 366
+    },
+    {
+      "epoch": 0.23877683799609628,
+      "grad_norm": 0.2559892237186432,
+      "learning_rate": 0.0001,
+      "loss": 2.2559,
+      "step": 367
+    },
+    {
+      "epoch": 0.23942745608327912,
+      "grad_norm": 0.43989577889442444,
+      "learning_rate": 0.0001,
+      "loss": 2.5208,
+      "step": 368
+    },
+    {
+      "epoch": 0.24007807417046195,
+      "grad_norm": 0.24543894827365875,
+      "learning_rate": 0.0001,
+      "loss": 2.1692,
+      "step": 369
+    },
+    {
+      "epoch": 0.24072869225764476,
+      "grad_norm": 0.37020954489707947,
+      "learning_rate": 0.0001,
+      "loss": 2.1287,
+      "step": 370
+    },
+    {
+      "epoch": 0.2413793103448276,
+      "grad_norm": 0.41815564036369324,
+      "learning_rate": 0.0001,
+      "loss": 2.5952,
+      "step": 371
+    },
+    {
+      "epoch": 0.2420299284320104,
+      "grad_norm": 0.22579136490821838,
+      "learning_rate": 0.0001,
+      "loss": 2.2427,
+      "step": 372
+    },
+    {
+      "epoch": 0.24268054651919324,
+      "grad_norm": 0.3004798889160156,
+      "learning_rate": 0.0001,
+      "loss": 2.2767,
+      "step": 373
+    },
+    {
+      "epoch": 0.24333116460637605,
+      "grad_norm": 0.27470141649246216,
+      "learning_rate": 0.0001,
+      "loss": 2.092,
+      "step": 374
+    },
+    {
+      "epoch": 0.24398178269355889,
+      "grad_norm": 0.25301867723464966,
+      "learning_rate": 0.0001,
+      "loss": 2.1816,
+      "step": 375
+    },
+    {
+      "epoch": 0.2446324007807417,
+      "grad_norm": 0.21194620430469513,
+      "learning_rate": 0.0001,
+      "loss": 2.1322,
+      "step": 376
+    },
+    {
+      "epoch": 0.24528301886792453,
+      "grad_norm": 0.28737103939056396,
+      "learning_rate": 0.0001,
+      "loss": 2.6685,
+      "step": 377
+    },
+    {
+      "epoch": 0.24593363695510737,
+      "grad_norm": 0.28857922554016113,
+      "learning_rate": 0.0001,
+      "loss": 2.2219,
+      "step": 378
+    },
+    {
+      "epoch": 0.24658425504229017,
+      "grad_norm": 0.29493409395217896,
+      "learning_rate": 0.0001,
+      "loss": 2.717,
+      "step": 379
+    },
+    {
+      "epoch": 0.247234873129473,
+      "grad_norm": 0.33975929021835327,
+      "learning_rate": 0.0001,
+      "loss": 2.3499,
+      "step": 380
+    },
+    {
+      "epoch": 0.24788549121665582,
+      "grad_norm": 0.21486152708530426,
+      "learning_rate": 0.0001,
+      "loss": 2.306,
+      "step": 381
+    },
+    {
+      "epoch": 0.24853610930383865,
+      "grad_norm": 0.2686431109905243,
+      "learning_rate": 0.0001,
+      "loss": 2.0942,
+      "step": 382
+    },
+    {
+      "epoch": 0.24918672739102146,
+      "grad_norm": 0.2812007963657379,
+      "learning_rate": 0.0001,
+      "loss": 2.3729,
+      "step": 383
+    },
+    {
+      "epoch": 0.2498373454782043,
+      "grad_norm": 0.31875330209732056,
+      "learning_rate": 0.0001,
+      "loss": 2.5766,
+      "step": 384
+    },
+    {
+      "epoch": 0.2504879635653871,
+      "grad_norm": 0.2624376714229584,
+      "learning_rate": 0.0001,
+      "loss": 2.2057,
+      "step": 385
+    },
+    {
+      "epoch": 0.2511385816525699,
+      "grad_norm": 0.265286386013031,
+      "learning_rate": 0.0001,
+      "loss": 2.2405,
+      "step": 386
+    },
+    {
+      "epoch": 0.2517891997397528,
+      "grad_norm": 0.3202246129512787,
+      "learning_rate": 0.0001,
+      "loss": 2.2817,
+      "step": 387
+    },
+    {
+      "epoch": 0.2524398178269356,
+      "grad_norm": 0.22770161926746368,
+      "learning_rate": 0.0001,
+      "loss": 1.9564,
+      "step": 388
+    },
+    {
+      "epoch": 0.2530904359141184,
+      "grad_norm": 0.3313138484954834,
+      "learning_rate": 0.0001,
+      "loss": 2.4424,
+      "step": 389
+    },
+    {
+      "epoch": 0.25374105400130126,
+      "grad_norm": 0.2961839437484741,
+      "learning_rate": 0.0001,
+      "loss": 2.4122,
+      "step": 390
+    },
+    {
+      "epoch": 0.25439167208848407,
+      "grad_norm": 0.24270308017730713,
+      "learning_rate": 0.0001,
+      "loss": 1.99,
+      "step": 391
+    },
+    {
+      "epoch": 0.2550422901756669,
+      "grad_norm": 0.2306670844554901,
+      "learning_rate": 0.0001,
+      "loss": 2.3529,
+      "step": 392
+    },
+    {
+      "epoch": 0.2556929082628497,
+      "grad_norm": 0.28387176990509033,
+      "learning_rate": 0.0001,
+      "loss": 2.0824,
+      "step": 393
+    },
+    {
+      "epoch": 0.25634352635003255,
+      "grad_norm": 0.3105824291706085,
+      "learning_rate": 0.0001,
+      "loss": 2.437,
+      "step": 394
+    },
+    {
+      "epoch": 0.25699414443721535,
+      "grad_norm": 0.1932361125946045,
+      "learning_rate": 0.0001,
+      "loss": 1.9747,
+      "step": 395
+    },
+    {
+      "epoch": 0.25764476252439816,
+      "grad_norm": 0.31146278977394104,
+      "learning_rate": 0.0001,
+      "loss": 2.263,
+      "step": 396
+    },
+    {
+      "epoch": 0.258295380611581,
+      "grad_norm": 0.24420365691184998,
+      "learning_rate": 0.0001,
+      "loss": 2.015,
+      "step": 397
+    },
+    {
+      "epoch": 0.25894599869876384,
+      "grad_norm": 0.24144989252090454,
+      "learning_rate": 0.0001,
+      "loss": 2.2536,
+      "step": 398
+    },
+    {
+      "epoch": 0.25959661678594664,
+      "grad_norm": 0.3478517532348633,
+      "learning_rate": 0.0001,
+      "loss": 2.5835,
+      "step": 399
+    },
+    {
+      "epoch": 0.26024723487312945,
+      "grad_norm": 0.24381348490715027,
+      "learning_rate": 0.0001,
+      "loss": 2.2439,
+      "step": 400
+    },
+    {
+      "epoch": 0.2608978529603123,
+      "grad_norm": 0.2834983468055725,
+      "learning_rate": 0.0001,
+      "loss": 2.3991,
+      "step": 401
+    },
+    {
+      "epoch": 0.2615484710474951,
+      "grad_norm": 0.28689858317375183,
+      "learning_rate": 0.0001,
+      "loss": 1.9156,
+      "step": 402
+    },
+    {
+      "epoch": 0.26219908913467793,
+      "grad_norm": 0.23692357540130615,
+      "learning_rate": 0.0001,
+      "loss": 2.0189,
+      "step": 403
+    },
+    {
+      "epoch": 0.26284970722186074,
+      "grad_norm": 0.30104926228523254,
+      "learning_rate": 0.0001,
+      "loss": 2.4945,
+      "step": 404
+    },
+    {
+      "epoch": 0.2635003253090436,
+      "grad_norm": 0.23472270369529724,
+      "learning_rate": 0.0001,
+      "loss": 1.8892,
+      "step": 405
+    },
+    {
+      "epoch": 0.2641509433962264,
+      "grad_norm": 0.31508034467697144,
+      "learning_rate": 0.0001,
+      "loss": 2.4935,
+      "step": 406
+    },
+    {
+      "epoch": 0.2648015614834092,
+      "grad_norm": 0.25103551149368286,
+      "learning_rate": 0.0001,
+      "loss": 2.4428,
+      "step": 407
+    },
+    {
+      "epoch": 0.2654521795705921,
+      "grad_norm": 0.2387259602546692,
+      "learning_rate": 0.0001,
+      "loss": 2.0989,
+      "step": 408
+    },
+    {
+      "epoch": 0.2661027976577749,
+      "grad_norm": 0.2606028616428375,
+      "learning_rate": 0.0001,
+      "loss": 1.9494,
+      "step": 409
+    },
+    {
+      "epoch": 0.2667534157449577,
+      "grad_norm": 0.25114724040031433,
+      "learning_rate": 0.0001,
+      "loss": 2.2432,
+      "step": 410
+    },
+    {
+      "epoch": 0.2674040338321405,
+      "grad_norm": 0.3072582483291626,
+      "learning_rate": 0.0001,
+      "loss": 2.3506,
+      "step": 411
+    },
+    {
+      "epoch": 0.2680546519193234,
+      "grad_norm": 0.23917561769485474,
+      "learning_rate": 0.0001,
+      "loss": 2.2665,
+      "step": 412
+    },
+    {
+      "epoch": 0.2687052700065062,
+      "grad_norm": 0.2120814174413681,
+      "learning_rate": 0.0001,
+      "loss": 1.9625,
+      "step": 413
+    },
+    {
+      "epoch": 0.269355888093689,
+      "grad_norm": 0.22003813087940216,
+      "learning_rate": 0.0001,
+      "loss": 2.1179,
+      "step": 414
+    },
+    {
+      "epoch": 0.27000650618087185,
+      "grad_norm": 0.33217060565948486,
+      "learning_rate": 0.0001,
+      "loss": 2.6353,
+      "step": 415
+    },
+    {
+      "epoch": 0.27065712426805466,
+      "grad_norm": 0.2260630577802658,
+      "learning_rate": 0.0001,
+      "loss": 2.0355,
+      "step": 416
+    },
+    {
+      "epoch": 0.27130774235523747,
+      "grad_norm": 0.30081093311309814,
+      "learning_rate": 0.0001,
+      "loss": 2.1825,
+      "step": 417
+    },
+    {
+      "epoch": 0.2719583604424203,
+      "grad_norm": 0.27275893092155457,
+      "learning_rate": 0.0001,
+      "loss": 2.6183,
+      "step": 418
+    },
+    {
+      "epoch": 0.27260897852960314,
+      "grad_norm": 0.4902358651161194,
+      "learning_rate": 0.0001,
+      "loss": 3.0888,
+      "step": 419
+    },
+    {
+      "epoch": 0.27325959661678595,
+      "grad_norm": 0.21213112771511078,
+      "learning_rate": 0.0001,
+      "loss": 2.1172,
+      "step": 420
+    },
+    {
+      "epoch": 0.27391021470396876,
+      "grad_norm": 0.35953450202941895,
+      "learning_rate": 0.0001,
+      "loss": 2.5109,
+      "step": 421
+    },
+    {
+      "epoch": 0.2745608327911516,
+      "grad_norm": 0.2081584334373474,
+      "learning_rate": 0.0001,
+      "loss": 2.0894,
+      "step": 422
+    },
+    {
+      "epoch": 0.27521145087833443,
+      "grad_norm": 0.20892906188964844,
+      "learning_rate": 0.0001,
+      "loss": 1.9643,
+      "step": 423
+    },
+    {
+      "epoch": 0.27586206896551724,
+      "grad_norm": 0.30058735609054565,
+      "learning_rate": 0.0001,
+      "loss": 2.6503,
+      "step": 424
+    },
+    {
+      "epoch": 0.27651268705270005,
+      "grad_norm": 0.32902124524116516,
+      "learning_rate": 0.0001,
+      "loss": 2.3271,
+      "step": 425
+    },
+    {
+      "epoch": 0.2771633051398829,
+      "grad_norm": 0.2003614902496338,
+      "learning_rate": 0.0001,
+      "loss": 1.9881,
+      "step": 426
+    },
+    {
+      "epoch": 0.2778139232270657,
+      "grad_norm": 0.33349111676216125,
+      "learning_rate": 0.0001,
+      "loss": 2.7625,
+      "step": 427
+    },
+    {
+      "epoch": 0.2784645413142485,
+      "grad_norm": 0.25051257014274597,
+      "learning_rate": 0.0001,
+      "loss": 2.0825,
+      "step": 428
+    },
+    {
+      "epoch": 0.27911515940143133,
+      "grad_norm": 0.3301559388637543,
+      "learning_rate": 0.0001,
+      "loss": 2.85,
+      "step": 429
+    },
+    {
+      "epoch": 0.2797657774886142,
+      "grad_norm": 0.18224254250526428,
+      "learning_rate": 0.0001,
+      "loss": 1.9687,
+      "step": 430
+    },
+    {
+      "epoch": 0.280416395575797,
+      "grad_norm": 0.21809989213943481,
+      "learning_rate": 0.0001,
+      "loss": 2.2596,
+      "step": 431
+    },
+    {
+      "epoch": 0.2810670136629798,
+      "grad_norm": 0.2473779171705246,
+      "learning_rate": 0.0001,
+      "loss": 2.2042,
+      "step": 432
+    },
+    {
+      "epoch": 0.2817176317501627,
+      "grad_norm": 0.20744885504245758,
+      "learning_rate": 0.0001,
+      "loss": 2.1546,
+      "step": 433
+    },
+    {
+      "epoch": 0.2823682498373455,
+      "grad_norm": 0.2620698809623718,
+      "learning_rate": 0.0001,
+      "loss": 2.5195,
+      "step": 434
+    },
+    {
+      "epoch": 0.2830188679245283,
+      "grad_norm": 0.291421115398407,
+      "learning_rate": 0.0001,
+      "loss": 2.4983,
+      "step": 435
+    },
+    {
+      "epoch": 0.2836694860117111,
+      "grad_norm": 0.3294708728790283,
+      "learning_rate": 0.0001,
+      "loss": 2.3146,
+      "step": 436
+    },
+    {
+      "epoch": 0.28432010409889397,
+      "grad_norm": 0.26191362738609314,
+      "learning_rate": 0.0001,
+      "loss": 2.2818,
+      "step": 437
+    },
+    {
+      "epoch": 0.2849707221860768,
+      "grad_norm": 0.29155483841896057,
+      "learning_rate": 0.0001,
+      "loss": 2.4888,
+      "step": 438
+    },
+    {
+      "epoch": 0.2856213402732596,
+      "grad_norm": 0.19482360780239105,
+      "learning_rate": 0.0001,
+      "loss": 2.0061,
+      "step": 439
+    },
+    {
+      "epoch": 0.28627195836044245,
+      "grad_norm": 0.2594612240791321,
+      "learning_rate": 0.0001,
+      "loss": 2.1891,
+      "step": 440
+    },
+    {
+      "epoch": 0.28692257644762525,
+      "grad_norm": 0.21656309068202972,
+      "learning_rate": 0.0001,
+      "loss": 1.7911,
+      "step": 441
+    },
+    {
+      "epoch": 0.28757319453480806,
+      "grad_norm": 0.18664829432964325,
+      "learning_rate": 0.0001,
+      "loss": 1.9634,
+      "step": 442
+    },
+    {
+      "epoch": 0.28822381262199087,
+      "grad_norm": 0.2178332507610321,
+      "learning_rate": 0.0001,
+      "loss": 2.32,
+      "step": 443
+    },
+    {
+      "epoch": 0.28887443070917374,
+      "grad_norm": 0.351418673992157,
+      "learning_rate": 0.0001,
+      "loss": 3.0873,
+      "step": 444
+    },
+    {
+      "epoch": 0.28952504879635654,
+      "grad_norm": 0.23604457080364227,
+      "learning_rate": 0.0001,
+      "loss": 2.46,
+      "step": 445
+    },
+    {
+      "epoch": 0.29017566688353935,
+      "grad_norm": 0.2599848806858063,
+      "learning_rate": 0.0001,
+      "loss": 2.0207,
+      "step": 446
+    },
+    {
+      "epoch": 0.29082628497072216,
+      "grad_norm": 0.340314120054245,
+      "learning_rate": 0.0001,
+      "loss": 2.279,
+      "step": 447
+    },
+    {
+      "epoch": 0.291476903057905,
+      "grad_norm": 0.23228399455547333,
+      "learning_rate": 0.0001,
+      "loss": 2.3561,
+      "step": 448
+    },
+    {
+      "epoch": 0.29212752114508783,
+      "grad_norm": 0.25504687428474426,
+      "learning_rate": 0.0001,
+      "loss": 2.2251,
+      "step": 449
+    },
+    {
+      "epoch": 0.29277813923227064,
+      "grad_norm": 0.2465014010667801,
+      "learning_rate": 0.0001,
+      "loss": 2.1031,
+      "step": 450
+    },
+    {
+      "epoch": 0.2934287573194535,
+      "grad_norm": 0.2188328504562378,
+      "learning_rate": 0.0001,
+      "loss": 2.1483,
+      "step": 451
+    },
+    {
+      "epoch": 0.2940793754066363,
+      "grad_norm": 0.24546551704406738,
+      "learning_rate": 0.0001,
+      "loss": 2.2334,
+      "step": 452
+    },
+    {
+      "epoch": 0.2947299934938191,
+      "grad_norm": 0.23416215181350708,
+      "learning_rate": 0.0001,
+      "loss": 2.1846,
+      "step": 453
+    },
+    {
+      "epoch": 0.29538061158100193,
+      "grad_norm": 0.25267231464385986,
+      "learning_rate": 0.0001,
+      "loss": 2.2134,
+      "step": 454
+    },
+    {
+      "epoch": 0.2960312296681848,
+      "grad_norm": 0.26632416248321533,
+      "learning_rate": 0.0001,
+      "loss": 2.5012,
+      "step": 455
+    },
+    {
+      "epoch": 0.2966818477553676,
+      "grad_norm": 0.18289139866828918,
+      "learning_rate": 0.0001,
+      "loss": 2.0524,
+      "step": 456
+    },
+    {
+      "epoch": 0.2973324658425504,
+      "grad_norm": 0.19033563137054443,
+      "learning_rate": 0.0001,
+      "loss": 2.0165,
+      "step": 457
+    },
+    {
+      "epoch": 0.2979830839297333,
+      "grad_norm": 0.200730562210083,
+      "learning_rate": 0.0001,
+      "loss": 1.8021,
+      "step": 458
+    },
+    {
+      "epoch": 0.2986337020169161,
+      "grad_norm": 0.2109062522649765,
+      "learning_rate": 0.0001,
+      "loss": 2.0655,
+      "step": 459
+    },
+    {
+      "epoch": 0.2992843201040989,
+      "grad_norm": 0.23461318016052246,
+      "learning_rate": 0.0001,
+      "loss": 2.3335,
+      "step": 460
+    },
+    {
+      "epoch": 0.2999349381912817,
+      "grad_norm": 0.2085726112127304,
+      "learning_rate": 0.0001,
+      "loss": 2.0061,
+      "step": 461
+    },
+    {
+      "epoch": 0.30058555627846456,
+      "grad_norm": 0.2938329875469208,
+      "learning_rate": 0.0001,
+      "loss": 2.5245,
+      "step": 462
+    },
+    {
+      "epoch": 0.30123617436564737,
+      "grad_norm": 0.22131232917308807,
+      "learning_rate": 0.0001,
+      "loss": 2.4115,
+      "step": 463
+    },
+    {
+      "epoch": 0.3018867924528302,
+      "grad_norm": 0.3459152579307556,
+      "learning_rate": 0.0001,
+      "loss": 2.3896,
+      "step": 464
+    },
+    {
+      "epoch": 0.302537410540013,
+      "grad_norm": 0.27464184165000916,
+      "learning_rate": 0.0001,
+      "loss": 2.6592,
+      "step": 465
+    },
+    {
+      "epoch": 0.30318802862719585,
+      "grad_norm": 0.28379327058792114,
+      "learning_rate": 0.0001,
+      "loss": 2.1453,
+      "step": 466
+    },
+    {
+      "epoch": 0.30383864671437866,
+      "grad_norm": 0.28283926844596863,
+      "learning_rate": 0.0001,
+      "loss": 2.1704,
+      "step": 467
+    },
+    {
+      "epoch": 0.30448926480156147,
+      "grad_norm": 0.22243599593639374,
+      "learning_rate": 0.0001,
+      "loss": 2.1175,
+      "step": 468
+    },
+    {
+      "epoch": 0.30513988288874433,
+      "grad_norm": 0.22331124544143677,
+      "learning_rate": 0.0001,
+      "loss": 1.8857,
+      "step": 469
+    },
+    {
+      "epoch": 0.30579050097592714,
+      "grad_norm": 0.21995989978313446,
+      "learning_rate": 0.0001,
+      "loss": 2.1316,
+      "step": 470
+    },
+    {
+      "epoch": 0.30644111906310995,
+      "grad_norm": 0.21140341460704803,
+      "learning_rate": 0.0001,
+      "loss": 2.0742,
+      "step": 471
+    },
+    {
+      "epoch": 0.30709173715029275,
+      "grad_norm": 0.31053757667541504,
+      "learning_rate": 0.0001,
+      "loss": 2.615,
+      "step": 472
+    },
+    {
+      "epoch": 0.3077423552374756,
+      "grad_norm": 0.2768484354019165,
+      "learning_rate": 0.0001,
+      "loss": 2.713,
+      "step": 473
+    },
+    {
+      "epoch": 0.3083929733246584,
+      "grad_norm": 0.2538318336009979,
+      "learning_rate": 0.0001,
+      "loss": 2.1917,
+      "step": 474
+    },
+    {
+      "epoch": 0.30904359141184123,
+      "grad_norm": 0.2105240672826767,
+      "learning_rate": 0.0001,
+      "loss": 2.2741,
+      "step": 475
+    },
+    {
+      "epoch": 0.3096942094990241,
+      "grad_norm": 0.2915903925895691,
+      "learning_rate": 0.0001,
+      "loss": 2.115,
+      "step": 476
+    },
+    {
+      "epoch": 0.3103448275862069,
+      "grad_norm": 0.30282047390937805,
+      "learning_rate": 0.0001,
+      "loss": 2.7806,
+      "step": 477
+    },
+    {
+      "epoch": 0.3109954456733897,
+      "grad_norm": 0.2707601487636566,
+      "learning_rate": 0.0001,
+      "loss": 2.6137,
+      "step": 478
+    },
+    {
+      "epoch": 0.3116460637605725,
+      "grad_norm": 0.34574300050735474,
+      "learning_rate": 0.0001,
+      "loss": 2.5957,
+      "step": 479
+    },
+    {
+      "epoch": 0.3122966818477554,
+      "grad_norm": 0.22767509520053864,
+      "learning_rate": 0.0001,
+      "loss": 2.3543,
+      "step": 480
+    },
+    {
+      "epoch": 0.3129472999349382,
+      "grad_norm": 0.25194215774536133,
+      "learning_rate": 0.0001,
+      "loss": 2.6586,
+      "step": 481
+    },
+    {
+      "epoch": 0.313597918022121,
+      "grad_norm": 0.20427219569683075,
+      "learning_rate": 0.0001,
+      "loss": 1.9091,
+      "step": 482
+    },
+    {
+      "epoch": 0.3142485361093038,
+      "grad_norm": 0.2993704378604889,
+      "learning_rate": 0.0001,
+      "loss": 2.4704,
+      "step": 483
+    },
+    {
+      "epoch": 0.3148991541964867,
+      "grad_norm": 0.18951758742332458,
+      "learning_rate": 0.0001,
+      "loss": 2.1108,
+      "step": 484
+    },
+    {
+      "epoch": 0.3155497722836695,
+      "grad_norm": 0.2622709572315216,
+      "learning_rate": 0.0001,
+      "loss": 2.4144,
+      "step": 485
+    },
+    {
+      "epoch": 0.3162003903708523,
+      "grad_norm": 0.20735126733779907,
+      "learning_rate": 0.0001,
+      "loss": 2.3065,
+      "step": 486
+    },
+    {
+      "epoch": 0.31685100845803515,
+      "grad_norm": 0.22782085835933685,
+      "learning_rate": 0.0001,
+      "loss": 2.4377,
+      "step": 487
+    },
+    {
+      "epoch": 0.31750162654521796,
+      "grad_norm": 0.2568935453891754,
+      "learning_rate": 0.0001,
+      "loss": 2.1199,
+      "step": 488
+    },
+    {
+      "epoch": 0.31815224463240077,
+      "grad_norm": 0.23917409777641296,
+      "learning_rate": 0.0001,
+      "loss": 2.2457,
+      "step": 489
+    },
+    {
+      "epoch": 0.3188028627195836,
+      "grad_norm": 0.21531902253627777,
+      "learning_rate": 0.0001,
+      "loss": 2.0489,
+      "step": 490
+    },
+    {
+      "epoch": 0.31945348080676644,
+      "grad_norm": 0.21461109817028046,
+      "learning_rate": 0.0001,
+      "loss": 2.1915,
+      "step": 491
+    },
+    {
+      "epoch": 0.32010409889394925,
+      "grad_norm": 0.2458680123090744,
+      "learning_rate": 0.0001,
+      "loss": 2.3939,
+      "step": 492
+    },
+    {
+      "epoch": 0.32075471698113206,
+      "grad_norm": 0.2617323696613312,
+      "learning_rate": 0.0001,
+      "loss": 2.5611,
+      "step": 493
+    },
+    {
+      "epoch": 0.3214053350683149,
+      "grad_norm": 0.22562618553638458,
+      "learning_rate": 0.0001,
+      "loss": 2.2703,
+      "step": 494
+    },
+    {
+      "epoch": 0.32205595315549773,
+      "grad_norm": 0.2290688008069992,
+      "learning_rate": 0.0001,
+      "loss": 2.3049,
+      "step": 495
+    },
+    {
+      "epoch": 0.32270657124268054,
+      "grad_norm": 0.4118833541870117,
+      "learning_rate": 0.0001,
+      "loss": 2.9194,
+      "step": 496
+    },
+    {
+      "epoch": 0.32335718932986335,
+      "grad_norm": 0.22502999007701874,
+      "learning_rate": 0.0001,
+      "loss": 2.2362,
+      "step": 497
+    },
+    {
+      "epoch": 0.3240078074170462,
+      "grad_norm": 0.23599191009998322,
+      "learning_rate": 0.0001,
+      "loss": 2.35,
+      "step": 498
+    },
+    {
+      "epoch": 0.324658425504229,
+      "grad_norm": 0.3065047860145569,
+      "learning_rate": 0.0001,
+      "loss": 2.3984,
+      "step": 499
+    },
+    {
+      "epoch": 0.32530904359141183,
+      "grad_norm": 0.19241982698440552,
+      "learning_rate": 0.0001,
+      "loss": 1.8787,
+      "step": 500
+    },
+    {
+      "epoch": 0.3259596616785947,
+      "grad_norm": 0.20695632696151733,
+      "learning_rate": 0.0001,
+      "loss": 1.9397,
+      "step": 501
+    },
+    {
+      "epoch": 0.3266102797657775,
+      "grad_norm": 0.1998564749956131,
+      "learning_rate": 0.0001,
+      "loss": 2.1463,
+      "step": 502
+    },
+    {
+      "epoch": 0.3272608978529603,
+      "grad_norm": 0.27775317430496216,
+      "learning_rate": 0.0001,
+      "loss": 2.7956,
+      "step": 503
+    },
+    {
+      "epoch": 0.3279115159401431,
+      "grad_norm": 0.2393936961889267,
+      "learning_rate": 0.0001,
+      "loss": 2.3785,
+      "step": 504
+    },
+    {
+      "epoch": 0.328562134027326,
+      "grad_norm": 0.20921163260936737,
+      "learning_rate": 0.0001,
+      "loss": 2.1909,
+      "step": 505
+    },
+    {
+      "epoch": 0.3292127521145088,
+      "grad_norm": 0.25875911116600037,
+      "learning_rate": 0.0001,
+      "loss": 2.129,
+      "step": 506
+    },
+    {
+      "epoch": 0.3298633702016916,
+      "grad_norm": 0.2382909208536148,
+      "learning_rate": 0.0001,
+      "loss": 2.3786,
+      "step": 507
+    },
+    {
+      "epoch": 0.3305139882888744,
+      "grad_norm": 0.19657136499881744,
+      "learning_rate": 0.0001,
+      "loss": 1.951,
+      "step": 508
+    },
+    {
+      "epoch": 0.33116460637605727,
+      "grad_norm": 0.23688004910945892,
+      "learning_rate": 0.0001,
+      "loss": 2.4348,
+      "step": 509
+    },
+    {
+      "epoch": 0.3318152244632401,
+      "grad_norm": 0.1988734006881714,
+      "learning_rate": 0.0001,
+      "loss": 2.2352,
+      "step": 510
+    },
+    {
+      "epoch": 0.3324658425504229,
+      "grad_norm": 0.2078763097524643,
+      "learning_rate": 0.0001,
+      "loss": 2.1376,
+      "step": 511
+    },
+    {
+      "epoch": 0.33311646063760575,
+      "grad_norm": 0.18860888481140137,
+      "learning_rate": 0.0001,
+      "loss": 1.9367,
+      "step": 512
+    },
+    {
+      "epoch": 0.33376707872478856,
+      "grad_norm": 0.30205249786376953,
+      "learning_rate": 0.0001,
+      "loss": 2.6822,
+      "step": 513
+    },
+    {
+      "epoch": 0.33441769681197137,
+      "grad_norm": 0.2146618664264679,
+      "learning_rate": 0.0001,
+      "loss": 2.1927,
+      "step": 514
+    },
+    {
+      "epoch": 0.3350683148991542,
+      "grad_norm": 0.19332504272460938,
+      "learning_rate": 0.0001,
+      "loss": 2.0442,
+      "step": 515
+    },
+    {
+      "epoch": 0.33571893298633704,
+      "grad_norm": 0.2289431244134903,
+      "learning_rate": 0.0001,
+      "loss": 2.0152,
+      "step": 516
+    },
+    {
+      "epoch": 0.33636955107351985,
+      "grad_norm": 0.21815945208072662,
+      "learning_rate": 0.0001,
+      "loss": 2.0015,
+      "step": 517
+    },
+    {
+      "epoch": 0.33702016916070265,
+      "grad_norm": 0.2226189821958542,
+      "learning_rate": 0.0001,
+      "loss": 2.2989,
+      "step": 518
+    },
+    {
+      "epoch": 0.3376707872478855,
+      "grad_norm": 0.22195078432559967,
+      "learning_rate": 0.0001,
+      "loss": 2.2237,
+      "step": 519
+    },
+    {
+      "epoch": 0.3383214053350683,
+      "grad_norm": 0.1946515589952469,
+      "learning_rate": 0.0001,
+      "loss": 1.9459,
+      "step": 520
+    },
+    {
+      "epoch": 0.33897202342225113,
+      "grad_norm": 0.21510568261146545,
+      "learning_rate": 0.0001,
+      "loss": 2.1305,
+      "step": 521
+    },
+    {
+      "epoch": 0.33962264150943394,
+      "grad_norm": 0.23448903858661652,
+      "learning_rate": 0.0001,
+      "loss": 2.1838,
+      "step": 522
+    },
+    {
+      "epoch": 0.3402732595966168,
+      "grad_norm": 0.19046911597251892,
+      "learning_rate": 0.0001,
+      "loss": 1.9739,
+      "step": 523
+    },
+    {
+      "epoch": 0.3409238776837996,
+      "grad_norm": 0.2314033806324005,
+      "learning_rate": 0.0001,
+      "loss": 2.2053,
+      "step": 524
+    },
+    {
+      "epoch": 0.3415744957709824,
+      "grad_norm": 0.2206612378358841,
+      "learning_rate": 0.0001,
+      "loss": 2.2566,
+      "step": 525
+    },
+    {
+      "epoch": 0.34222511385816523,
+      "grad_norm": 0.19578076899051666,
+      "learning_rate": 0.0001,
+      "loss": 2.045,
+      "step": 526
+    },
+    {
+      "epoch": 0.3428757319453481,
+      "grad_norm": 0.1787755936384201,
+      "learning_rate": 0.0001,
+      "loss": 1.8942,
+      "step": 527
+    },
+    {
+      "epoch": 0.3435263500325309,
+      "grad_norm": 0.20091751217842102,
+      "learning_rate": 0.0001,
+      "loss": 2.1576,
+      "step": 528
+    },
+    {
+      "epoch": 0.3441769681197137,
+      "grad_norm": 0.21869762241840363,
+      "learning_rate": 0.0001,
+      "loss": 2.1938,
+      "step": 529
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "grad_norm": 0.26101449131965637,
+      "learning_rate": 0.0001,
+      "loss": 2.3642,
+      "step": 530
+    },
+    {
+      "epoch": 0.3454782042940794,
+      "grad_norm": 0.21874766051769257,
+      "learning_rate": 0.0001,
+      "loss": 2.4553,
+      "step": 531
+    },
+    {
+      "epoch": 0.3461288223812622,
+      "grad_norm": 0.224325492978096,
+      "learning_rate": 0.0001,
+      "loss": 2.2959,
+      "step": 532
+    },
+    {
+      "epoch": 0.346779440468445,
+      "grad_norm": 0.21268363296985626,
+      "learning_rate": 0.0001,
+      "loss": 2.1021,
+      "step": 533
+    },
+    {
+      "epoch": 0.34743005855562786,
+      "grad_norm": 0.20979231595993042,
+      "learning_rate": 0.0001,
+      "loss": 2.0304,
+      "step": 534
+    },
+    {
+      "epoch": 0.34808067664281067,
+      "grad_norm": 0.19552691280841827,
+      "learning_rate": 0.0001,
+      "loss": 1.9747,
+      "step": 535
+    },
+    {
+      "epoch": 0.3487312947299935,
+      "grad_norm": 0.27929842472076416,
+      "learning_rate": 0.0001,
+      "loss": 2.445,
+      "step": 536
+    },
+    {
+      "epoch": 0.34938191281717634,
+      "grad_norm": 0.19953188300132751,
+      "learning_rate": 0.0001,
+      "loss": 1.9766,
+      "step": 537
+    },
+    {
+      "epoch": 0.35003253090435915,
+      "grad_norm": 0.29898926615715027,
+      "learning_rate": 0.0001,
+      "loss": 2.4818,
+      "step": 538
+    },
+    {
+      "epoch": 0.35068314899154196,
+      "grad_norm": 0.18719644844532013,
+      "learning_rate": 0.0001,
+      "loss": 1.9046,
+      "step": 539
+    },
+    {
+      "epoch": 0.35133376707872477,
+      "grad_norm": 0.2602563798427582,
+      "learning_rate": 0.0001,
+      "loss": 2.1539,
+      "step": 540
+    },
+    {
+      "epoch": 0.35198438516590763,
+      "grad_norm": 0.23460406064987183,
+      "learning_rate": 0.0001,
+      "loss": 2.3826,
+      "step": 541
+    },
+    {
+      "epoch": 0.35263500325309044,
+      "grad_norm": 0.2821134328842163,
+      "learning_rate": 0.0001,
+      "loss": 2.223,
+      "step": 542
+    },
+    {
+      "epoch": 0.35328562134027325,
+      "grad_norm": 0.2641044557094574,
+      "learning_rate": 0.0001,
+      "loss": 2.2402,
+      "step": 543
+    },
+    {
+      "epoch": 0.35393623942745606,
+      "grad_norm": 0.21963565051555634,
+      "learning_rate": 0.0001,
+      "loss": 2.3988,
+      "step": 544
+    },
+    {
+      "epoch": 0.3545868575146389,
+      "grad_norm": 0.26475685834884644,
+      "learning_rate": 0.0001,
+      "loss": 2.3046,
+      "step": 545
+    },
+    {
+      "epoch": 0.35523747560182173,
+      "grad_norm": 0.27148157358169556,
+      "learning_rate": 0.0001,
+      "loss": 2.5076,
+      "step": 546
+    },
+    {
+      "epoch": 0.35588809368900454,
+      "grad_norm": 0.28925588726997375,
+      "learning_rate": 0.0001,
+      "loss": 2.8395,
+      "step": 547
+    },
+    {
+      "epoch": 0.3565387117761874,
+      "grad_norm": 0.22953632473945618,
+      "learning_rate": 0.0001,
+      "loss": 2.1198,
+      "step": 548
+    },
+    {
+      "epoch": 0.3571893298633702,
+      "grad_norm": 0.23960557579994202,
+      "learning_rate": 0.0001,
+      "loss": 2.3064,
+      "step": 549
+    },
+    {
+      "epoch": 0.357839947950553,
+      "grad_norm": 0.3133333921432495,
+      "learning_rate": 0.0001,
+      "loss": 2.6034,
+      "step": 550
+    },
+    {
+      "epoch": 0.3584905660377358,
+      "grad_norm": 0.21745215356349945,
+      "learning_rate": 0.0001,
+      "loss": 2.4553,
+      "step": 551
+    },
+    {
+      "epoch": 0.3591411841249187,
+      "grad_norm": 0.23547130823135376,
+      "learning_rate": 0.0001,
+      "loss": 2.0469,
+      "step": 552
+    },
+    {
+      "epoch": 0.3597918022121015,
+      "grad_norm": 0.2646094262599945,
+      "learning_rate": 0.0001,
+      "loss": 1.9016,
+      "step": 553
+    },
+    {
+      "epoch": 0.3604424202992843,
+      "grad_norm": 0.3079530596733093,
+      "learning_rate": 0.0001,
+      "loss": 2.8979,
+      "step": 554
+    },
+    {
+      "epoch": 0.36109303838646717,
+      "grad_norm": 0.38223740458488464,
+      "learning_rate": 0.0001,
+      "loss": 3.066,
+      "step": 555
+    },
+    {
+      "epoch": 0.36174365647365,
+      "grad_norm": 0.2535337209701538,
+      "learning_rate": 0.0001,
+      "loss": 2.1327,
+      "step": 556
+    },
+    {
+      "epoch": 0.3623942745608328,
+      "grad_norm": 0.2373637855052948,
+      "learning_rate": 0.0001,
+      "loss": 2.1141,
+      "step": 557
+    },
+    {
+      "epoch": 0.3630448926480156,
+      "grad_norm": 0.19437271356582642,
+      "learning_rate": 0.0001,
+      "loss": 1.9753,
+      "step": 558
+    },
+    {
+      "epoch": 0.36369551073519846,
+      "grad_norm": 0.20236878097057343,
+      "learning_rate": 0.0001,
+      "loss": 2.2516,
+      "step": 559
+    },
+    {
+      "epoch": 0.36434612882238127,
+      "grad_norm": 0.21252363920211792,
+      "learning_rate": 0.0001,
+      "loss": 2.3645,
+      "step": 560
+    },
+    {
+      "epoch": 0.3649967469095641,
+      "grad_norm": 0.21689258515834808,
+      "learning_rate": 0.0001,
+      "loss": 2.1145,
+      "step": 561
+    },
+    {
+      "epoch": 0.3656473649967469,
+      "grad_norm": 0.22365228831768036,
+      "learning_rate": 0.0001,
+      "loss": 2.3083,
+      "step": 562
+    },
+    {
+      "epoch": 0.36629798308392975,
+      "grad_norm": 0.21607807278633118,
+      "learning_rate": 0.0001,
+      "loss": 2.3199,
+      "step": 563
+    },
+    {
+      "epoch": 0.36694860117111255,
+      "grad_norm": 0.1885683536529541,
+      "learning_rate": 0.0001,
+      "loss": 1.9303,
+      "step": 564
+    },
+    {
+      "epoch": 0.36759921925829536,
+      "grad_norm": 0.20064905285835266,
+      "learning_rate": 0.0001,
+      "loss": 2.0661,
+      "step": 565
+    },
+    {
+      "epoch": 0.3682498373454782,
+      "grad_norm": 0.23532240092754364,
+      "learning_rate": 0.0001,
+      "loss": 2.6942,
+      "step": 566
+    },
+    {
+      "epoch": 0.36890045543266103,
+      "grad_norm": 0.22937807440757751,
+      "learning_rate": 0.0001,
+      "loss": 2.1962,
+      "step": 567
+    },
+    {
+      "epoch": 0.36955107351984384,
+      "grad_norm": 0.2540866732597351,
+      "learning_rate": 0.0001,
+      "loss": 2.5012,
+      "step": 568
+    },
+    {
+      "epoch": 0.37020169160702665,
+      "grad_norm": 0.23405294120311737,
+      "learning_rate": 0.0001,
+      "loss": 2.2439,
+      "step": 569
+    },
+    {
+      "epoch": 0.3708523096942095,
+      "grad_norm": 0.24394820630550385,
+      "learning_rate": 0.0001,
+      "loss": 2.0741,
+      "step": 570
+    },
+    {
+      "epoch": 0.3715029277813923,
+      "grad_norm": 0.2063736468553543,
+      "learning_rate": 0.0001,
+      "loss": 2.0864,
+      "step": 571
+    },
+    {
+      "epoch": 0.37215354586857513,
+      "grad_norm": 0.3300686180591583,
+      "learning_rate": 0.0001,
+      "loss": 2.4983,
+      "step": 572
+    },
+    {
+      "epoch": 0.372804163955758,
+      "grad_norm": 0.21294772624969482,
+      "learning_rate": 0.0001,
+      "loss": 2.2273,
+      "step": 573
+    },
+    {
+      "epoch": 0.3734547820429408,
+      "grad_norm": 0.2629190981388092,
+      "learning_rate": 0.0001,
+      "loss": 2.1732,
+      "step": 574
+    },
+    {
+      "epoch": 0.3741054001301236,
+      "grad_norm": 0.2141999751329422,
+      "learning_rate": 0.0001,
+      "loss": 2.3038,
+      "step": 575
+    },
+    {
+      "epoch": 0.3747560182173064,
+      "grad_norm": 0.3467566668987274,
+      "learning_rate": 0.0001,
+      "loss": 2.7748,
+      "step": 576
+    },
+    {
+      "epoch": 0.3754066363044893,
+      "grad_norm": 0.3112248182296753,
+      "learning_rate": 0.0001,
+      "loss": 2.2376,
+      "step": 577
+    },
+    {
+      "epoch": 0.3760572543916721,
+      "grad_norm": 0.21217738091945648,
+      "learning_rate": 0.0001,
+      "loss": 1.9146,
+      "step": 578
+    },
+    {
+      "epoch": 0.3767078724788549,
+      "grad_norm": 0.19359458982944489,
+      "learning_rate": 0.0001,
+      "loss": 2.0913,
+      "step": 579
+    },
+    {
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.27635738253593445,
+      "learning_rate": 0.0001,
+      "loss": 2.2855,
+      "step": 580
+    },
+    {
+      "epoch": 0.37800910865322057,
+      "grad_norm": 0.19366882741451263,
+      "learning_rate": 0.0001,
+      "loss": 2.0194,
+      "step": 581
+    },
+    {
+      "epoch": 0.3786597267404034,
+      "grad_norm": 0.2016839236021042,
+      "learning_rate": 0.0001,
+      "loss": 2.1519,
+      "step": 582
+    },
+    {
+      "epoch": 0.3793103448275862,
+      "grad_norm": 0.22154097259044647,
+      "learning_rate": 0.0001,
+      "loss": 1.9849,
+      "step": 583
+    },
+    {
+      "epoch": 0.37996096291476905,
+      "grad_norm": 0.2089187502861023,
+      "learning_rate": 0.0001,
+      "loss": 2.3624,
+      "step": 584
+    },
+    {
+      "epoch": 0.38061158100195186,
+      "grad_norm": 0.25050756335258484,
+      "learning_rate": 0.0001,
+      "loss": 2.1773,
+      "step": 585
+    },
+    {
+      "epoch": 0.38126219908913467,
+      "grad_norm": 0.23007918894290924,
+      "learning_rate": 0.0001,
+      "loss": 2.2054,
+      "step": 586
+    },
+    {
+      "epoch": 0.3819128171763175,
+      "grad_norm": 0.25022968649864197,
+      "learning_rate": 0.0001,
+      "loss": 2.219,
+      "step": 587
+    },
+    {
+      "epoch": 0.38256343526350034,
+      "grad_norm": 0.2205193042755127,
+      "learning_rate": 0.0001,
+      "loss": 2.2049,
+      "step": 588
+    },
+    {
+      "epoch": 0.38321405335068315,
+      "grad_norm": 0.21454961597919464,
+      "learning_rate": 0.0001,
+      "loss": 2.0683,
+      "step": 589
+    },
+    {
+      "epoch": 0.38386467143786596,
+      "grad_norm": 0.2088347226381302,
+      "learning_rate": 0.0001,
+      "loss": 2.1301,
+      "step": 590
+    },
+    {
+      "epoch": 0.3845152895250488,
+      "grad_norm": 0.20322394371032715,
+      "learning_rate": 0.0001,
+      "loss": 2.2098,
+      "step": 591
+    },
+    {
+      "epoch": 0.38516590761223163,
+      "grad_norm": 0.231514111161232,
+      "learning_rate": 0.0001,
+      "loss": 2.5523,
+      "step": 592
+    },
+    {
+      "epoch": 0.38581652569941444,
+      "grad_norm": 0.24791982769966125,
+      "learning_rate": 0.0001,
+      "loss": 2.2259,
+      "step": 593
+    },
+    {
+      "epoch": 0.38646714378659724,
+      "grad_norm": 0.21148578822612762,
+      "learning_rate": 0.0001,
+      "loss": 2.0834,
+      "step": 594
+    },
+    {
+      "epoch": 0.3871177618737801,
+      "grad_norm": 0.263713538646698,
+      "learning_rate": 0.0001,
+      "loss": 2.3101,
+      "step": 595
+    },
+    {
+      "epoch": 0.3877683799609629,
+      "grad_norm": 0.22197774052619934,
+      "learning_rate": 0.0001,
+      "loss": 2.1173,
+      "step": 596
+    },
+    {
+      "epoch": 0.3884189980481457,
+      "grad_norm": 0.2237439900636673,
+      "learning_rate": 0.0001,
+      "loss": 2.1109,
+      "step": 597
+    },
+    {
+      "epoch": 0.3890696161353286,
+      "grad_norm": 0.27451419830322266,
+      "learning_rate": 0.0001,
+      "loss": 2.5311,
+      "step": 598
+    },
+    {
+      "epoch": 0.3897202342225114,
+      "grad_norm": 0.18475750088691711,
+      "learning_rate": 0.0001,
+      "loss": 1.9241,
+      "step": 599
+    },
+    {
+      "epoch": 0.3903708523096942,
+      "grad_norm": 0.20120149850845337,
+      "learning_rate": 0.0001,
+      "loss": 2.1033,
+      "step": 600
+    },
+    {
+      "epoch": 0.391021470396877,
+      "grad_norm": 0.19626259803771973,
+      "learning_rate": 0.0001,
+      "loss": 2.1223,
+      "step": 601
+    },
+    {
+      "epoch": 0.3916720884840599,
+      "grad_norm": 0.22795897722244263,
+      "learning_rate": 0.0001,
+      "loss": 2.2021,
+      "step": 602
+    },
+    {
+      "epoch": 0.3923227065712427,
+      "grad_norm": 0.5195867419242859,
+      "learning_rate": 0.0001,
+      "loss": 3.1849,
+      "step": 603
+    },
+    {
+      "epoch": 0.3929733246584255,
+      "grad_norm": 0.2636241614818573,
+      "learning_rate": 0.0001,
+      "loss": 2.0739,
+      "step": 604
+    },
+    {
+      "epoch": 0.3936239427456083,
+      "grad_norm": 0.33922895789146423,
+      "learning_rate": 0.0001,
+      "loss": 2.31,
+      "step": 605
+    },
+    {
+      "epoch": 0.39427456083279117,
+      "grad_norm": 0.17467042803764343,
+      "learning_rate": 0.0001,
+      "loss": 1.9201,
+      "step": 606
+    },
+    {
+      "epoch": 0.394925178919974,
+      "grad_norm": 0.22457371652126312,
+      "learning_rate": 0.0001,
+      "loss": 1.9783,
+      "step": 607
+    },
+    {
+      "epoch": 0.3955757970071568,
+      "grad_norm": 0.5104444026947021,
+      "learning_rate": 0.0001,
+      "loss": 2.3777,
+      "step": 608
+    },
+    {
+      "epoch": 0.39622641509433965,
+      "grad_norm": 0.4531616270542145,
+      "learning_rate": 0.0001,
+      "loss": 2.8208,
+      "step": 609
+    },
+    {
+      "epoch": 0.39687703318152245,
+      "grad_norm": 0.20649151504039764,
+      "learning_rate": 0.0001,
+      "loss": 2.1377,
+      "step": 610
+    },
+    {
+      "epoch": 0.39752765126870526,
+      "grad_norm": 0.39769667387008667,
+      "learning_rate": 0.0001,
+      "loss": 2.2228,
+      "step": 611
+    },
+    {
+      "epoch": 0.39817826935588807,
+      "grad_norm": 0.2832731008529663,
+      "learning_rate": 0.0001,
+      "loss": 1.9664,
+      "step": 612
+    },
+    {
+      "epoch": 0.39882888744307093,
+      "grad_norm": 0.2754386067390442,
+      "learning_rate": 0.0001,
+      "loss": 2.5595,
+      "step": 613
+    },
+    {
+      "epoch": 0.39947950553025374,
+      "grad_norm": 0.404364675283432,
+      "learning_rate": 0.0001,
+      "loss": 2.8133,
+      "step": 614
+    },
+    {
+      "epoch": 0.40013012361743655,
+      "grad_norm": 0.30304789543151855,
+      "learning_rate": 0.0001,
+      "loss": 2.2729,
+      "step": 615
+    },
+    {
+      "epoch": 0.4007807417046194,
+      "grad_norm": 0.2519910931587219,
+      "learning_rate": 0.0001,
+      "loss": 2.3655,
+      "step": 616
+    },
+    {
+      "epoch": 0.4014313597918022,
+      "grad_norm": 0.2863995134830475,
+      "learning_rate": 0.0001,
+      "loss": 2.0774,
+      "step": 617
+    },
+    {
+      "epoch": 0.40208197787898503,
+      "grad_norm": 0.393622487783432,
+      "learning_rate": 0.0001,
+      "loss": 2.5082,
+      "step": 618
+    },
+    {
+      "epoch": 0.40273259596616784,
+      "grad_norm": 0.21836060285568237,
+      "learning_rate": 0.0001,
+      "loss": 1.9548,
+      "step": 619
+    },
+    {
+      "epoch": 0.4033832140533507,
+      "grad_norm": 0.358052521944046,
+      "learning_rate": 0.0001,
+      "loss": 2.5158,
+      "step": 620
+    },
+    {
+      "epoch": 0.4040338321405335,
+      "grad_norm": 0.237140953540802,
+      "learning_rate": 0.0001,
+      "loss": 2.2111,
+      "step": 621
+    },
+    {
+      "epoch": 0.4046844502277163,
+      "grad_norm": 0.20998883247375488,
+      "learning_rate": 0.0001,
+      "loss": 2.1351,
+      "step": 622
+    },
+    {
+      "epoch": 0.4053350683148991,
+      "grad_norm": 0.18059247732162476,
+      "learning_rate": 0.0001,
+      "loss": 1.9451,
+      "step": 623
+    },
+    {
+      "epoch": 0.405985686402082,
+      "grad_norm": 0.17532669007778168,
+      "learning_rate": 0.0001,
+      "loss": 1.8591,
+      "step": 624
+    },
+    {
+      "epoch": 0.4066363044892648,
+      "grad_norm": 0.24097976088523865,
+      "learning_rate": 0.0001,
+      "loss": 2.6534,
+      "step": 625
+    },
+    {
+      "epoch": 0.4072869225764476,
+      "grad_norm": 0.19505445659160614,
+      "learning_rate": 0.0001,
+      "loss": 1.8952,
+      "step": 626
+    },
+    {
+      "epoch": 0.40793754066363047,
+      "grad_norm": 0.232722207903862,
+      "learning_rate": 0.0001,
+      "loss": 2.2055,
+      "step": 627
+    },
+    {
+      "epoch": 0.4085881587508133,
+      "grad_norm": 0.23899732530117035,
+      "learning_rate": 0.0001,
+      "loss": 2.5848,
+      "step": 628
+    },
+    {
+      "epoch": 0.4092387768379961,
+      "grad_norm": 0.2411729097366333,
+      "learning_rate": 0.0001,
+      "loss": 2.5315,
+      "step": 629
+    },
+    {
+      "epoch": 0.4098893949251789,
+      "grad_norm": 0.25042012333869934,
+      "learning_rate": 0.0001,
+      "loss": 2.4154,
+      "step": 630
+    },
+    {
+      "epoch": 0.41054001301236176,
+      "grad_norm": 0.2764488160610199,
+      "learning_rate": 0.0001,
+      "loss": 2.0564,
+      "step": 631
+    },
+    {
+      "epoch": 0.41119063109954457,
+      "grad_norm": 0.24761155247688293,
+      "learning_rate": 0.0001,
+      "loss": 2.3245,
+      "step": 632
+    },
+    {
+      "epoch": 0.4118412491867274,
+      "grad_norm": 0.22376200556755066,
+      "learning_rate": 0.0001,
+      "loss": 2.1881,
+      "step": 633
+    },
+    {
+      "epoch": 0.41249186727391024,
+      "grad_norm": 0.19060148298740387,
+      "learning_rate": 0.0001,
+      "loss": 1.9588,
+      "step": 634
+    },
+    {
+      "epoch": 0.41314248536109305,
+      "grad_norm": 0.4157400131225586,
+      "learning_rate": 0.0001,
+      "loss": 2.9024,
+      "step": 635
+    },
+    {
+      "epoch": 0.41379310344827586,
+      "grad_norm": 0.2557002007961273,
+      "learning_rate": 0.0001,
+      "loss": 1.9819,
+      "step": 636
+    },
+    {
+      "epoch": 0.41444372153545866,
+      "grad_norm": 0.2908417880535126,
+      "learning_rate": 0.0001,
+      "loss": 2.112,
+      "step": 637
+    },
+    {
+      "epoch": 0.41509433962264153,
+      "grad_norm": 0.32937270402908325,
+      "learning_rate": 0.0001,
+      "loss": 2.4976,
+      "step": 638
+    },
+    {
+      "epoch": 0.41574495770982434,
+      "grad_norm": 0.20382268726825714,
+      "learning_rate": 0.0001,
+      "loss": 2.0448,
+      "step": 639
+    },
+    {
+      "epoch": 0.41639557579700714,
+      "grad_norm": 0.23484939336776733,
+      "learning_rate": 0.0001,
+      "loss": 1.9514,
+      "step": 640
+    },
+    {
+      "epoch": 0.41704619388418995,
+      "grad_norm": 0.23023058474063873,
+      "learning_rate": 0.0001,
+      "loss": 2.0768,
+      "step": 641
+    },
+    {
+      "epoch": 0.4176968119713728,
+      "grad_norm": 0.22951190173625946,
+      "learning_rate": 0.0001,
+      "loss": 2.0764,
+      "step": 642
+    },
+    {
+      "epoch": 0.4183474300585556,
+      "grad_norm": 0.18971513211727142,
+      "learning_rate": 0.0001,
+      "loss": 1.9693,
+      "step": 643
+    },
+    {
+      "epoch": 0.41899804814573843,
+      "grad_norm": 0.24955709278583527,
+      "learning_rate": 0.0001,
+      "loss": 2.4898,
+      "step": 644
+    },
+    {
+      "epoch": 0.4196486662329213,
+      "grad_norm": 0.3344306945800781,
+      "learning_rate": 0.0001,
+      "loss": 2.4779,
+      "step": 645
+    },
+    {
+      "epoch": 0.4202992843201041,
+      "grad_norm": 0.21661825478076935,
+      "learning_rate": 0.0001,
+      "loss": 2.0472,
+      "step": 646
+    },
+    {
+      "epoch": 0.4209499024072869,
+      "grad_norm": 0.1972419023513794,
+      "learning_rate": 0.0001,
+      "loss": 2.1712,
+      "step": 647
+    },
+    {
+      "epoch": 0.4216005204944697,
+      "grad_norm": 0.21619470417499542,
+      "learning_rate": 0.0001,
+      "loss": 2.0739,
+      "step": 648
+    },
+    {
+      "epoch": 0.4222511385816526,
+      "grad_norm": 0.2329091727733612,
+      "learning_rate": 0.0001,
+      "loss": 2.1362,
+      "step": 649
+    },
+    {
+      "epoch": 0.4229017566688354,
+      "grad_norm": 0.22971969842910767,
+      "learning_rate": 0.0001,
+      "loss": 1.9898,
+      "step": 650
+    },
+    {
+      "epoch": 0.4235523747560182,
+      "grad_norm": 0.20185063779354095,
+      "learning_rate": 0.0001,
+      "loss": 2.1008,
+      "step": 651
+    },
+    {
+      "epoch": 0.42420299284320107,
+      "grad_norm": 0.2658546566963196,
+      "learning_rate": 0.0001,
+      "loss": 2.5734,
+      "step": 652
+    },
+    {
+      "epoch": 0.4248536109303839,
+      "grad_norm": 0.23109374940395355,
+      "learning_rate": 0.0001,
+      "loss": 2.2569,
+      "step": 653
+    },
+    {
+      "epoch": 0.4255042290175667,
+      "grad_norm": 0.25115352869033813,
+      "learning_rate": 0.0001,
+      "loss": 2.5967,
+      "step": 654
+    },
+    {
+      "epoch": 0.4261548471047495,
+      "grad_norm": 0.20470669865608215,
+      "learning_rate": 0.0001,
+      "loss": 2.0302,
+      "step": 655
+    },
+    {
+      "epoch": 0.42680546519193235,
+      "grad_norm": 0.2151513546705246,
+      "learning_rate": 0.0001,
+      "loss": 2.5183,
+      "step": 656
+    },
+    {
+      "epoch": 0.42745608327911516,
+      "grad_norm": 0.2571411728858948,
+      "learning_rate": 0.0001,
+      "loss": 2.255,
+      "step": 657
+    },
+    {
+      "epoch": 0.42810670136629797,
+      "grad_norm": 0.2414022833108902,
+      "learning_rate": 0.0001,
+      "loss": 2.4076,
+      "step": 658
+    },
+    {
+      "epoch": 0.42875731945348083,
+      "grad_norm": 0.21041014790534973,
+      "learning_rate": 0.0001,
+      "loss": 2.0091,
+      "step": 659
+    },
+    {
+      "epoch": 0.42940793754066364,
+      "grad_norm": 0.21241822838783264,
+      "learning_rate": 0.0001,
+      "loss": 2.355,
+      "step": 660
+    },
+    {
+      "epoch": 0.43005855562784645,
+      "grad_norm": 0.21031403541564941,
+      "learning_rate": 0.0001,
+      "loss": 1.9887,
+      "step": 661
+    },
+    {
+      "epoch": 0.43070917371502926,
+      "grad_norm": 0.19765952229499817,
+      "learning_rate": 0.0001,
+      "loss": 2.1555,
+      "step": 662
+    },
+    {
+      "epoch": 0.4313597918022121,
+      "grad_norm": 0.24740834534168243,
+      "learning_rate": 0.0001,
+      "loss": 2.2349,
+      "step": 663
+    },
+    {
+      "epoch": 0.43201040988939493,
+      "grad_norm": 0.22086234390735626,
+      "learning_rate": 0.0001,
+      "loss": 2.0948,
+      "step": 664
+    },
+    {
+      "epoch": 0.43266102797657774,
+      "grad_norm": 0.21949239075183868,
+      "learning_rate": 0.0001,
+      "loss": 2.3905,
+      "step": 665
+    },
+    {
+      "epoch": 0.43331164606376055,
+      "grad_norm": 0.20536834001541138,
+      "learning_rate": 0.0001,
+      "loss": 2.0547,
+      "step": 666
+    },
+    {
+      "epoch": 0.4339622641509434,
+      "grad_norm": 0.2570655941963196,
+      "learning_rate": 0.0001,
+      "loss": 2.0261,
+      "step": 667
+    },
+    {
+      "epoch": 0.4346128822381262,
+      "grad_norm": 0.3293687701225281,
+      "learning_rate": 0.0001,
+      "loss": 2.344,
+      "step": 668
+    },
+    {
+      "epoch": 0.435263500325309,
+      "grad_norm": 0.22947120666503906,
+      "learning_rate": 0.0001,
+      "loss": 2.232,
+      "step": 669
+    },
+    {
+      "epoch": 0.4359141184124919,
+      "grad_norm": 0.2425599992275238,
+      "learning_rate": 0.0001,
+      "loss": 2.309,
+      "step": 670
+    },
+    {
+      "epoch": 0.4365647364996747,
+      "grad_norm": 0.2506352663040161,
+      "learning_rate": 0.0001,
+      "loss": 2.1249,
+      "step": 671
+    },
+    {
+      "epoch": 0.4372153545868575,
+      "grad_norm": 0.19457192718982697,
+      "learning_rate": 0.0001,
+      "loss": 1.9461,
+      "step": 672
+    },
+    {
+      "epoch": 0.4378659726740403,
+      "grad_norm": 0.3749271035194397,
+      "learning_rate": 0.0001,
+      "loss": 2.8532,
+      "step": 673
+    },
+    {
+      "epoch": 0.4385165907612232,
+      "grad_norm": 0.25384366512298584,
+      "learning_rate": 0.0001,
+      "loss": 2.6495,
+      "step": 674
+    },
+    {
+      "epoch": 0.439167208848406,
+      "grad_norm": 0.21413469314575195,
+      "learning_rate": 0.0001,
+      "loss": 2.084,
+      "step": 675
+    },
+    {
+      "epoch": 0.4398178269355888,
+      "grad_norm": 0.228125661611557,
+      "learning_rate": 0.0001,
+      "loss": 2.2175,
+      "step": 676
+    },
+    {
+      "epoch": 0.44046844502277166,
+      "grad_norm": 0.1948491632938385,
+      "learning_rate": 0.0001,
+      "loss": 1.9702,
+      "step": 677
+    },
+    {
+      "epoch": 0.44111906310995447,
+      "grad_norm": 0.307992547750473,
+      "learning_rate": 0.0001,
+      "loss": 2.5884,
+      "step": 678
+    },
+    {
+      "epoch": 0.4417696811971373,
+      "grad_norm": 0.23681728541851044,
+      "learning_rate": 0.0001,
+      "loss": 2.2104,
+      "step": 679
+    },
+    {
+      "epoch": 0.4424202992843201,
+      "grad_norm": 0.23185166716575623,
+      "learning_rate": 0.0001,
+      "loss": 2.0823,
+      "step": 680
+    },
+    {
+      "epoch": 0.44307091737150295,
+      "grad_norm": 0.2772667109966278,
+      "learning_rate": 0.0001,
+      "loss": 2.3729,
+      "step": 681
+    },
+    {
+      "epoch": 0.44372153545868576,
+      "grad_norm": 0.18908965587615967,
+      "learning_rate": 0.0001,
+      "loss": 2.0585,
+      "step": 682
+    },
+    {
+      "epoch": 0.44437215354586856,
+      "grad_norm": 0.2063988745212555,
+      "learning_rate": 0.0001,
+      "loss": 1.9474,
+      "step": 683
+    },
+    {
+      "epoch": 0.4450227716330514,
+      "grad_norm": 0.19444917142391205,
+      "learning_rate": 0.0001,
+      "loss": 1.9269,
+      "step": 684
+    },
+    {
+      "epoch": 0.44567338972023424,
+      "grad_norm": 0.2866727113723755,
+      "learning_rate": 0.0001,
+      "loss": 2.5145,
+      "step": 685
+    },
+    {
+      "epoch": 0.44632400780741704,
+      "grad_norm": 0.24801641702651978,
+      "learning_rate": 0.0001,
+      "loss": 2.2954,
+      "step": 686
+    },
+    {
+      "epoch": 0.44697462589459985,
+      "grad_norm": 0.2115658074617386,
+      "learning_rate": 0.0001,
+      "loss": 2.1956,
+      "step": 687
+    },
+    {
+      "epoch": 0.4476252439817827,
+      "grad_norm": 0.3155558109283447,
+      "learning_rate": 0.0001,
+      "loss": 2.7396,
+      "step": 688
+    },
+    {
+      "epoch": 0.4482758620689655,
+      "grad_norm": 0.22418133914470673,
+      "learning_rate": 0.0001,
+      "loss": 2.1066,
+      "step": 689
+    },
+    {
+      "epoch": 0.44892648015614833,
+      "grad_norm": 0.2707614600658417,
+      "learning_rate": 0.0001,
+      "loss": 2.3353,
+      "step": 690
+    },
+    {
+      "epoch": 0.44957709824333114,
+      "grad_norm": 0.22262880206108093,
+      "learning_rate": 0.0001,
+      "loss": 2.2143,
+      "step": 691
+    },
+    {
+      "epoch": 0.450227716330514,
+      "grad_norm": 0.25256767868995667,
+      "learning_rate": 0.0001,
+      "loss": 2.2786,
+      "step": 692
+    },
+    {
+      "epoch": 0.4508783344176968,
+      "grad_norm": 0.20360921323299408,
+      "learning_rate": 0.0001,
+      "loss": 2.0059,
+      "step": 693
+    },
+    {
+      "epoch": 0.4515289525048796,
+      "grad_norm": 0.20573420822620392,
+      "learning_rate": 0.0001,
+      "loss": 2.0884,
+      "step": 694
+    },
+    {
+      "epoch": 0.4521795705920625,
+      "grad_norm": 0.31812623143196106,
+      "learning_rate": 0.0001,
+      "loss": 2.5905,
+      "step": 695
+    },
+    {
+      "epoch": 0.4528301886792453,
+      "grad_norm": 0.24690969288349152,
+      "learning_rate": 0.0001,
+      "loss": 2.5157,
+      "step": 696
+    },
+    {
+      "epoch": 0.4534808067664281,
+      "grad_norm": 0.256793737411499,
+      "learning_rate": 0.0001,
+      "loss": 2.1548,
+      "step": 697
+    },
+    {
+      "epoch": 0.4541314248536109,
+      "grad_norm": 0.2659960985183716,
+      "learning_rate": 0.0001,
+      "loss": 2.2977,
+      "step": 698
+    },
+    {
+      "epoch": 0.4547820429407938,
+      "grad_norm": 0.23824195563793182,
+      "learning_rate": 0.0001,
+      "loss": 2.5946,
+      "step": 699
+    },
+    {
+      "epoch": 0.4554326610279766,
+      "grad_norm": 0.2580608129501343,
+      "learning_rate": 0.0001,
+      "loss": 2.2608,
+      "step": 700
+    },
+    {
+      "epoch": 0.4560832791151594,
+      "grad_norm": 0.270622193813324,
+      "learning_rate": 0.0001,
+      "loss": 2.5848,
+      "step": 701
+    },
+    {
+      "epoch": 0.4567338972023422,
+      "grad_norm": 0.2170489877462387,
+      "learning_rate": 0.0001,
+      "loss": 2.4315,
+      "step": 702
+    },
+    {
+      "epoch": 0.45738451528952506,
+      "grad_norm": 0.20716050267219543,
+      "learning_rate": 0.0001,
+      "loss": 2.1592,
+      "step": 703
+    },
+    {
+      "epoch": 0.45803513337670787,
+      "grad_norm": 0.24847671389579773,
+      "learning_rate": 0.0001,
+      "loss": 2.3202,
+      "step": 704
+    },
+    {
+      "epoch": 0.4586857514638907,
+      "grad_norm": 0.24049146473407745,
+      "learning_rate": 0.0001,
+      "loss": 2.1968,
+      "step": 705
+    },
+    {
+      "epoch": 0.45933636955107354,
+      "grad_norm": 0.2079533487558365,
+      "learning_rate": 0.0001,
+      "loss": 2.2966,
+      "step": 706
+    },
+    {
+      "epoch": 0.45998698763825635,
+      "grad_norm": 0.18255428969860077,
+      "learning_rate": 0.0001,
+      "loss": 1.9931,
+      "step": 707
+    },
+    {
+      "epoch": 0.46063760572543916,
+      "grad_norm": 0.28015655279159546,
+      "learning_rate": 0.0001,
+      "loss": 2.2605,
+      "step": 708
+    },
+    {
+      "epoch": 0.46128822381262197,
+      "grad_norm": 0.27453094720840454,
+      "learning_rate": 0.0001,
+      "loss": 2.2835,
+      "step": 709
+    },
+    {
+      "epoch": 0.46193884189980483,
+      "grad_norm": 0.2751506268978119,
+      "learning_rate": 0.0001,
+      "loss": 2.665,
+      "step": 710
+    },
+    {
+      "epoch": 0.46258945998698764,
+      "grad_norm": 0.2759210169315338,
+      "learning_rate": 0.0001,
+      "loss": 2.3593,
+      "step": 711
+    },
+    {
+      "epoch": 0.46324007807417045,
+      "grad_norm": 0.2902829051017761,
+      "learning_rate": 0.0001,
+      "loss": 2.7421,
+      "step": 712
+    },
+    {
+      "epoch": 0.4638906961613533,
+      "grad_norm": 0.24083854258060455,
+      "learning_rate": 0.0001,
+      "loss": 2.4644,
+      "step": 713
+    },
+    {
+      "epoch": 0.4645413142485361,
+      "grad_norm": 0.23614934086799622,
+      "learning_rate": 0.0001,
+      "loss": 2.2939,
+      "step": 714
+    },
+    {
+      "epoch": 0.4651919323357189,
+      "grad_norm": 0.1972537487745285,
+      "learning_rate": 0.0001,
+      "loss": 1.9391,
+      "step": 715
+    },
+    {
+      "epoch": 0.46584255042290174,
+      "grad_norm": 0.2227838933467865,
+      "learning_rate": 0.0001,
+      "loss": 1.9396,
+      "step": 716
+    },
+    {
+      "epoch": 0.4664931685100846,
+      "grad_norm": 0.3672918379306793,
+      "learning_rate": 0.0001,
+      "loss": 2.7508,
+      "step": 717
+    },
+    {
+      "epoch": 0.4671437865972674,
+      "grad_norm": 0.2712246775627136,
+      "learning_rate": 0.0001,
+      "loss": 2.2838,
+      "step": 718
+    },
+    {
+      "epoch": 0.4677944046844502,
+      "grad_norm": 0.2337927669286728,
+      "learning_rate": 0.0001,
+      "loss": 1.9807,
+      "step": 719
+    },
+    {
+      "epoch": 0.468445022771633,
+      "grad_norm": 0.2051180601119995,
+      "learning_rate": 0.0001,
+      "loss": 2.0311,
+      "step": 720
+    },
+    {
+      "epoch": 0.4690956408588159,
+      "grad_norm": 0.1965889185667038,
+      "learning_rate": 0.0001,
+      "loss": 2.1114,
+      "step": 721
+    },
+    {
+      "epoch": 0.4697462589459987,
+      "grad_norm": 0.2106337547302246,
+      "learning_rate": 0.0001,
+      "loss": 2.0792,
+      "step": 722
+    },
+    {
+      "epoch": 0.4703968770331815,
+      "grad_norm": 0.19918356835842133,
+      "learning_rate": 0.0001,
+      "loss": 2.1323,
+      "step": 723
+    },
+    {
+      "epoch": 0.47104749512036437,
+      "grad_norm": 0.20124401152133942,
+      "learning_rate": 0.0001,
+      "loss": 2.0008,
+      "step": 724
+    },
+    {
+      "epoch": 0.4716981132075472,
+      "grad_norm": 0.2172473967075348,
+      "learning_rate": 0.0001,
+      "loss": 2.3891,
+      "step": 725
+    },
+    {
+      "epoch": 0.47234873129473,
+      "grad_norm": 0.2524811029434204,
+      "learning_rate": 0.0001,
+      "loss": 2.3343,
+      "step": 726
+    },
+    {
+      "epoch": 0.4729993493819128,
+      "grad_norm": 0.22882957756519318,
+      "learning_rate": 0.0001,
+      "loss": 2.6723,
+      "step": 727
+    },
+    {
+      "epoch": 0.47364996746909566,
+      "grad_norm": 0.2434161901473999,
+      "learning_rate": 0.0001,
+      "loss": 1.9549,
+      "step": 728
+    },
+    {
+      "epoch": 0.47430058555627846,
+      "grad_norm": 0.19140364229679108,
+      "learning_rate": 0.0001,
+      "loss": 2.0468,
+      "step": 729
+    },
+    {
+      "epoch": 0.4749512036434613,
+      "grad_norm": 0.22166937589645386,
+      "learning_rate": 0.0001,
+      "loss": 2.3432,
+      "step": 730
+    },
+    {
+      "epoch": 0.47560182173064414,
+      "grad_norm": 0.2005748748779297,
+      "learning_rate": 0.0001,
+      "loss": 2.0616,
+      "step": 731
+    },
+    {
+      "epoch": 0.47625243981782694,
+      "grad_norm": 0.3115980923175812,
+      "learning_rate": 0.0001,
+      "loss": 2.6153,
+      "step": 732
+    },
+    {
+      "epoch": 0.47690305790500975,
+      "grad_norm": 0.27135169506073,
+      "learning_rate": 0.0001,
+      "loss": 2.3225,
+      "step": 733
+    },
+    {
+      "epoch": 0.47755367599219256,
+      "grad_norm": 0.20748727023601532,
+      "learning_rate": 0.0001,
+      "loss": 1.834,
+      "step": 734
+    },
+    {
+      "epoch": 0.4782042940793754,
+      "grad_norm": 0.4031495153903961,
+      "learning_rate": 0.0001,
+      "loss": 2.8177,
+      "step": 735
+    },
+    {
+      "epoch": 0.47885491216655823,
+      "grad_norm": 0.2978368401527405,
+      "learning_rate": 0.0001,
+      "loss": 2.6178,
+      "step": 736
+    },
+    {
+      "epoch": 0.47950553025374104,
+      "grad_norm": 0.3466270864009857,
+      "learning_rate": 0.0001,
+      "loss": 2.6031,
+      "step": 737
+    },
+    {
+      "epoch": 0.4801561483409239,
+      "grad_norm": 0.20074127614498138,
+      "learning_rate": 0.0001,
+      "loss": 2.247,
+      "step": 738
+    },
+    {
+      "epoch": 0.4808067664281067,
+      "grad_norm": 0.2393479198217392,
+      "learning_rate": 0.0001,
+      "loss": 2.1265,
+      "step": 739
+    },
+    {
+      "epoch": 0.4814573845152895,
+      "grad_norm": 0.27758634090423584,
+      "learning_rate": 0.0001,
+      "loss": 2.5025,
+      "step": 740
+    },
+    {
+      "epoch": 0.48210800260247233,
+      "grad_norm": 0.20123820006847382,
+      "learning_rate": 0.0001,
+      "loss": 2.0083,
+      "step": 741
+    },
+    {
+      "epoch": 0.4827586206896552,
+      "grad_norm": 0.19012506306171417,
+      "learning_rate": 0.0001,
+      "loss": 2.0212,
+      "step": 742
+    },
+    {
+      "epoch": 0.483409238776838,
+      "grad_norm": 0.19451047480106354,
+      "learning_rate": 0.0001,
+      "loss": 2.0295,
+      "step": 743
+    },
+    {
+      "epoch": 0.4840598568640208,
+      "grad_norm": 0.3339052200317383,
+      "learning_rate": 0.0001,
+      "loss": 2.4813,
+      "step": 744
+    },
+    {
+      "epoch": 0.4847104749512036,
+      "grad_norm": 0.2646152973175049,
+      "learning_rate": 0.0001,
+      "loss": 2.4302,
+      "step": 745
+    },
+    {
+      "epoch": 0.4853610930383865,
+      "grad_norm": 0.23590324819087982,
+      "learning_rate": 0.0001,
+      "loss": 2.1723,
+      "step": 746
+    },
+    {
+      "epoch": 0.4860117111255693,
+      "grad_norm": 0.28924039006233215,
+      "learning_rate": 0.0001,
+      "loss": 2.8005,
+      "step": 747
+    },
+    {
+      "epoch": 0.4866623292127521,
+      "grad_norm": 0.21145464479923248,
+      "learning_rate": 0.0001,
+      "loss": 2.3501,
+      "step": 748
+    },
+    {
+      "epoch": 0.48731294729993496,
+      "grad_norm": 0.22815656661987305,
+      "learning_rate": 0.0001,
+      "loss": 2.1997,
+      "step": 749
+    },
+    {
+      "epoch": 0.48796356538711777,
+      "grad_norm": 0.24325215816497803,
+      "learning_rate": 0.0001,
+      "loss": 2.039,
+      "step": 750
+    },
+    {
+      "epoch": 0.4886141834743006,
+      "grad_norm": 0.3235335052013397,
+      "learning_rate": 0.0001,
+      "loss": 2.4533,
+      "step": 751
+    },
+    {
+      "epoch": 0.4892648015614834,
+      "grad_norm": 0.25513559579849243,
+      "learning_rate": 0.0001,
+      "loss": 2.3779,
+      "step": 752
+    },
+    {
+      "epoch": 0.48991541964866625,
+      "grad_norm": 0.2905427813529968,
+      "learning_rate": 0.0001,
+      "loss": 1.9843,
+      "step": 753
+    },
+    {
+      "epoch": 0.49056603773584906,
+      "grad_norm": 0.23760183155536652,
+      "learning_rate": 0.0001,
+      "loss": 2.1825,
+      "step": 754
+    },
+    {
+      "epoch": 0.49121665582303187,
+      "grad_norm": 0.2170071303844452,
+      "learning_rate": 0.0001,
+      "loss": 1.9877,
+      "step": 755
+    },
+    {
+      "epoch": 0.49186727391021473,
+      "grad_norm": 0.2555190920829773,
+      "learning_rate": 0.0001,
+      "loss": 2.457,
+      "step": 756
+    },
+    {
+      "epoch": 0.49251789199739754,
+      "grad_norm": 0.2571033835411072,
+      "learning_rate": 0.0001,
+      "loss": 2.1152,
+      "step": 757
+    },
+    {
+      "epoch": 0.49316851008458035,
+      "grad_norm": 0.23969238996505737,
+      "learning_rate": 0.0001,
+      "loss": 2.3439,
+      "step": 758
+    },
+    {
+      "epoch": 0.49381912817176316,
+      "grad_norm": 0.1900262087583542,
+      "learning_rate": 0.0001,
+      "loss": 1.8999,
+      "step": 759
+    },
+    {
+      "epoch": 0.494469746258946,
+      "grad_norm": 0.19621430337429047,
+      "learning_rate": 0.0001,
+      "loss": 2.0658,
+      "step": 760
+    },
+    {
+      "epoch": 0.4951203643461288,
+      "grad_norm": 0.21956481039524078,
+      "learning_rate": 0.0001,
+      "loss": 2.5427,
+      "step": 761
+    },
+    {
+      "epoch": 0.49577098243331164,
+      "grad_norm": 0.22567258775234222,
+      "learning_rate": 0.0001,
+      "loss": 2.2777,
+      "step": 762
+    },
+    {
+      "epoch": 0.49642160052049444,
+      "grad_norm": 0.20233570039272308,
+      "learning_rate": 0.0001,
+      "loss": 2.0342,
+      "step": 763
+    },
+    {
+      "epoch": 0.4970722186076773,
+      "grad_norm": 0.23662947118282318,
+      "learning_rate": 0.0001,
+      "loss": 2.3668,
+      "step": 764
+    },
+    {
+      "epoch": 0.4977228366948601,
+      "grad_norm": 0.2625278830528259,
+      "learning_rate": 0.0001,
+      "loss": 2.6536,
+      "step": 765
+    },
+    {
+      "epoch": 0.4983734547820429,
+      "grad_norm": 0.23235228657722473,
+      "learning_rate": 0.0001,
+      "loss": 2.1891,
+      "step": 766
+    },
+    {
+      "epoch": 0.4990240728692258,
+      "grad_norm": 0.19439217448234558,
+      "learning_rate": 0.0001,
+      "loss": 1.9647,
+      "step": 767
+    },
+    {
+      "epoch": 0.4996746909564086,
+      "grad_norm": 0.19810114800930023,
+      "learning_rate": 0.0001,
+      "loss": 1.9965,
+      "step": 768
+    },
+    {
+      "epoch": 0.5003253090435914,
+      "grad_norm": 0.2525380253791809,
+      "learning_rate": 0.0001,
+      "loss": 2.2444,
+      "step": 769
+    },
+    {
+      "epoch": 0.5009759271307742,
+      "grad_norm": 0.2409314513206482,
+      "learning_rate": 0.0001,
+      "loss": 2.1717,
+      "step": 770
+    },
+    {
+      "epoch": 0.501626545217957,
+      "grad_norm": 0.25244686007499695,
+      "learning_rate": 0.0001,
+      "loss": 2.0126,
+      "step": 771
+    },
+    {
+      "epoch": 0.5022771633051398,
+      "grad_norm": 0.19767141342163086,
+      "learning_rate": 0.0001,
+      "loss": 2.1384,
+      "step": 772
+    },
+    {
+      "epoch": 0.5029277813923227,
+      "grad_norm": 0.39446812868118286,
+      "learning_rate": 0.0001,
+      "loss": 2.8039,
+      "step": 773
+    },
+    {
+      "epoch": 0.5035783994795056,
+      "grad_norm": 0.2643390893936157,
+      "learning_rate": 0.0001,
+      "loss": 2.1524,
+      "step": 774
+    },
+    {
+      "epoch": 0.5042290175666884,
+      "grad_norm": 0.27606508135795593,
+      "learning_rate": 0.0001,
+      "loss": 2.1802,
+      "step": 775
+    },
+    {
+      "epoch": 0.5048796356538712,
+      "grad_norm": 0.364106148481369,
+      "learning_rate": 0.0001,
+      "loss": 2.9694,
+      "step": 776
+    },
+    {
+      "epoch": 0.505530253741054,
+      "grad_norm": 0.23091645538806915,
+      "learning_rate": 0.0001,
+      "loss": 2.5471,
+      "step": 777
+    },
+    {
+      "epoch": 0.5061808718282368,
+      "grad_norm": 0.19318193197250366,
+      "learning_rate": 0.0001,
+      "loss": 2.2082,
+      "step": 778
+    },
+    {
+      "epoch": 0.5068314899154196,
+      "grad_norm": 0.28997862339019775,
+      "learning_rate": 0.0001,
+      "loss": 2.4399,
+      "step": 779
+    },
+    {
+      "epoch": 0.5074821080026025,
+      "grad_norm": 0.22487197816371918,
+      "learning_rate": 0.0001,
+      "loss": 2.1946,
+      "step": 780
+    },
+    {
+      "epoch": 0.5081327260897853,
+      "grad_norm": 0.24430596828460693,
+      "learning_rate": 0.0001,
+      "loss": 2.4456,
+      "step": 781
+    },
+    {
+      "epoch": 0.5087833441769681,
+      "grad_norm": 0.21677151322364807,
+      "learning_rate": 0.0001,
+      "loss": 2.2082,
+      "step": 782
+    },
+    {
+      "epoch": 0.5094339622641509,
+      "grad_norm": 0.47995632886886597,
+      "learning_rate": 0.0001,
+      "loss": 3.1358,
+      "step": 783
+    },
+    {
+      "epoch": 0.5100845803513337,
+      "grad_norm": 0.19044414162635803,
+      "learning_rate": 0.0001,
+      "loss": 1.8924,
+      "step": 784
+    },
+    {
+      "epoch": 0.5107351984385166,
+      "grad_norm": 0.19143608212471008,
+      "learning_rate": 0.0001,
+      "loss": 2.0459,
+      "step": 785
+    },
+    {
+      "epoch": 0.5113858165256994,
+      "grad_norm": 0.22588413953781128,
+      "learning_rate": 0.0001,
+      "loss": 2.1369,
+      "step": 786
+    },
+    {
+      "epoch": 0.5120364346128823,
+      "grad_norm": 0.2786167860031128,
+      "learning_rate": 0.0001,
+      "loss": 2.2029,
+      "step": 787
+    },
+    {
+      "epoch": 0.5126870527000651,
+      "grad_norm": 0.24471627175807953,
+      "learning_rate": 0.0001,
+      "loss": 2.1248,
+      "step": 788
+    },
+    {
+      "epoch": 0.5133376707872479,
+      "grad_norm": 0.17795225977897644,
+      "learning_rate": 0.0001,
+      "loss": 1.7926,
+      "step": 789
+    },
+    {
+      "epoch": 0.5139882888744307,
+      "grad_norm": 0.2173709124326706,
+      "learning_rate": 0.0001,
+      "loss": 2.0538,
+      "step": 790
+    },
+    {
+      "epoch": 0.5146389069616135,
+      "grad_norm": 0.2027692049741745,
+      "learning_rate": 0.0001,
+      "loss": 1.8568,
+      "step": 791
+    },
+    {
+      "epoch": 0.5152895250487963,
+      "grad_norm": 0.2013595849275589,
+      "learning_rate": 0.0001,
+      "loss": 2.0501,
+      "step": 792
+    },
+    {
+      "epoch": 0.5159401431359791,
+      "grad_norm": 0.21996662020683289,
+      "learning_rate": 0.0001,
+      "loss": 2.0374,
+      "step": 793
+    },
+    {
+      "epoch": 0.516590761223162,
+      "grad_norm": 0.21435722708702087,
+      "learning_rate": 0.0001,
+      "loss": 2.1907,
+      "step": 794
+    },
+    {
+      "epoch": 0.5172413793103449,
+      "grad_norm": 0.21512284874916077,
+      "learning_rate": 0.0001,
+      "loss": 2.315,
+      "step": 795
+    },
+    {
+      "epoch": 0.5178919973975277,
+      "grad_norm": 0.19432400166988373,
+      "learning_rate": 0.0001,
+      "loss": 2.103,
+      "step": 796
+    },
+    {
+      "epoch": 0.5185426154847105,
+      "grad_norm": 0.23112992942333221,
+      "learning_rate": 0.0001,
+      "loss": 2.328,
+      "step": 797
+    },
+    {
+      "epoch": 0.5191932335718933,
+      "grad_norm": 0.19719737768173218,
+      "learning_rate": 0.0001,
+      "loss": 1.9569,
+      "step": 798
+    },
+    {
+      "epoch": 0.5198438516590761,
+      "grad_norm": 0.2115892618894577,
+      "learning_rate": 0.0001,
+      "loss": 2.2533,
+      "step": 799
+    },
+    {
+      "epoch": 0.5204944697462589,
+      "grad_norm": 0.24321842193603516,
+      "learning_rate": 0.0001,
+      "loss": 2.6597,
+      "step": 800
+    },
+    {
+      "epoch": 0.5211450878334418,
+      "grad_norm": 0.18219350278377533,
+      "learning_rate": 0.0001,
+      "loss": 1.8709,
+      "step": 801
+    },
+    {
+      "epoch": 0.5217957059206246,
+      "grad_norm": 0.18715021014213562,
+      "learning_rate": 0.0001,
+      "loss": 2.0021,
+      "step": 802
+    },
+    {
+      "epoch": 0.5224463240078074,
+      "grad_norm": 0.25940024852752686,
+      "learning_rate": 0.0001,
+      "loss": 2.3742,
+      "step": 803
+    },
+    {
+      "epoch": 0.5230969420949902,
+      "grad_norm": 0.18714728951454163,
+      "learning_rate": 0.0001,
+      "loss": 2.211,
+      "step": 804
+    },
+    {
+      "epoch": 0.523747560182173,
+      "grad_norm": 0.20145951211452484,
+      "learning_rate": 0.0001,
+      "loss": 2.0047,
+      "step": 805
+    },
+    {
+      "epoch": 0.5243981782693559,
+      "grad_norm": 0.18992845714092255,
+      "learning_rate": 0.0001,
+      "loss": 1.8559,
+      "step": 806
+    },
+    {
+      "epoch": 0.5250487963565387,
+      "grad_norm": 0.2682324945926666,
+      "learning_rate": 0.0001,
+      "loss": 2.4791,
+      "step": 807
+    },
+    {
+      "epoch": 0.5256994144437215,
+      "grad_norm": 0.33034664392471313,
+      "learning_rate": 0.0001,
+      "loss": 2.3089,
+      "step": 808
+    },
+    {
+      "epoch": 0.5263500325309044,
+      "grad_norm": 0.18838956952095032,
+      "learning_rate": 0.0001,
+      "loss": 1.9462,
+      "step": 809
+    },
+    {
+      "epoch": 0.5270006506180872,
+      "grad_norm": 0.42872169613838196,
+      "learning_rate": 0.0001,
+      "loss": 2.6874,
+      "step": 810
+    },
+    {
+      "epoch": 0.52765126870527,
+      "grad_norm": 0.2108643501996994,
+      "learning_rate": 0.0001,
+      "loss": 2.3627,
+      "step": 811
+    },
+    {
+      "epoch": 0.5283018867924528,
+      "grad_norm": 0.21745599806308746,
+      "learning_rate": 0.0001,
+      "loss": 2.1204,
+      "step": 812
+    },
+    {
+      "epoch": 0.5289525048796356,
+      "grad_norm": 0.2577585279941559,
+      "learning_rate": 0.0001,
+      "loss": 1.9746,
+      "step": 813
+    },
+    {
+      "epoch": 0.5296031229668184,
+      "grad_norm": 0.372471421957016,
+      "learning_rate": 0.0001,
+      "loss": 2.688,
+      "step": 814
+    },
+    {
+      "epoch": 0.5302537410540012,
+      "grad_norm": 0.2425181120634079,
+      "learning_rate": 0.0001,
+      "loss": 2.1377,
+      "step": 815
+    },
+    {
+      "epoch": 0.5309043591411842,
+      "grad_norm": 0.2638307511806488,
+      "learning_rate": 0.0001,
+      "loss": 2.1088,
+      "step": 816
+    },
+    {
+      "epoch": 0.531554977228367,
+      "grad_norm": 0.2356933355331421,
+      "learning_rate": 0.0001,
+      "loss": 2.2291,
+      "step": 817
+    },
+    {
+      "epoch": 0.5322055953155498,
+      "grad_norm": 0.23714864253997803,
+      "learning_rate": 0.0001,
+      "loss": 2.0929,
+      "step": 818
+    },
+    {
+      "epoch": 0.5328562134027326,
+      "grad_norm": 0.19541950523853302,
+      "learning_rate": 0.0001,
+      "loss": 2.0883,
+      "step": 819
+    },
+    {
+      "epoch": 0.5335068314899154,
+      "grad_norm": 0.3091617822647095,
+      "learning_rate": 0.0001,
+      "loss": 3.0127,
+      "step": 820
+    },
+    {
+      "epoch": 0.5341574495770982,
+      "grad_norm": 0.2592740058898926,
+      "learning_rate": 0.0001,
+      "loss": 1.8307,
+      "step": 821
+    },
+    {
+      "epoch": 0.534808067664281,
+      "grad_norm": 0.22505807876586914,
+      "learning_rate": 0.0001,
+      "loss": 2.462,
+      "step": 822
+    },
+    {
+      "epoch": 0.5354586857514639,
+      "grad_norm": 0.22032824158668518,
+      "learning_rate": 0.0001,
+      "loss": 2.2718,
+      "step": 823
+    },
+    {
+      "epoch": 0.5361093038386467,
+      "grad_norm": 0.2457459270954132,
+      "learning_rate": 0.0001,
+      "loss": 2.4213,
+      "step": 824
+    },
+    {
+      "epoch": 0.5367599219258296,
+      "grad_norm": 0.24181683361530304,
+      "learning_rate": 0.0001,
+      "loss": 1.9347,
+      "step": 825
+    },
+    {
+      "epoch": 0.5374105400130124,
+      "grad_norm": 0.29988738894462585,
+      "learning_rate": 0.0001,
+      "loss": 2.7697,
+      "step": 826
+    },
+    {
+      "epoch": 0.5380611581001952,
+      "grad_norm": 0.24946388602256775,
+      "learning_rate": 0.0001,
+      "loss": 2.2117,
+      "step": 827
+    },
+    {
+      "epoch": 0.538711776187378,
+      "grad_norm": 0.20339331030845642,
+      "learning_rate": 0.0001,
+      "loss": 1.9936,
+      "step": 828
+    },
+    {
+      "epoch": 0.5393623942745608,
+      "grad_norm": 0.22250457108020782,
+      "learning_rate": 0.0001,
+      "loss": 2.0785,
+      "step": 829
+    },
+    {
+      "epoch": 0.5400130123617437,
+      "grad_norm": 0.1869298666715622,
+      "learning_rate": 0.0001,
+      "loss": 2.0406,
+      "step": 830
+    },
+    {
+      "epoch": 0.5406636304489265,
+      "grad_norm": 0.1873755156993866,
+      "learning_rate": 0.0001,
+      "loss": 1.9126,
+      "step": 831
+    },
+    {
+      "epoch": 0.5413142485361093,
+      "grad_norm": 0.3135535418987274,
+      "learning_rate": 0.0001,
+      "loss": 2.2881,
+      "step": 832
+    },
+    {
+      "epoch": 0.5419648666232921,
+      "grad_norm": 0.20596185326576233,
+      "learning_rate": 0.0001,
+      "loss": 2.0682,
+      "step": 833
+    },
+    {
+      "epoch": 0.5426154847104749,
+      "grad_norm": 0.25786712765693665,
+      "learning_rate": 0.0001,
+      "loss": 2.0591,
+      "step": 834
+    },
+    {
+      "epoch": 0.5432661027976577,
+      "grad_norm": 0.2592066824436188,
+      "learning_rate": 0.0001,
+      "loss": 2.052,
+      "step": 835
+    },
+    {
+      "epoch": 0.5439167208848406,
+      "grad_norm": 0.20738951861858368,
+      "learning_rate": 0.0001,
+      "loss": 1.9726,
+      "step": 836
+    },
+    {
+      "epoch": 0.5445673389720235,
+      "grad_norm": 0.21384763717651367,
+      "learning_rate": 0.0001,
+      "loss": 2.1897,
+      "step": 837
+    },
+    {
+      "epoch": 0.5452179570592063,
+      "grad_norm": 0.22050943970680237,
+      "learning_rate": 0.0001,
+      "loss": 2.3597,
+      "step": 838
+    },
+    {
+      "epoch": 0.5458685751463891,
+      "grad_norm": 0.1996280699968338,
+      "learning_rate": 0.0001,
+      "loss": 2.0492,
+      "step": 839
+    },
+    {
+      "epoch": 0.5465191932335719,
+      "grad_norm": 0.2430533468723297,
+      "learning_rate": 0.0001,
+      "loss": 2.2774,
+      "step": 840
+    },
+    {
+      "epoch": 0.5471698113207547,
+      "grad_norm": 0.22777177393436432,
+      "learning_rate": 0.0001,
+      "loss": 2.0779,
+      "step": 841
+    },
+    {
+      "epoch": 0.5478204294079375,
+      "grad_norm": 0.22464539110660553,
+      "learning_rate": 0.0001,
+      "loss": 2.3316,
+      "step": 842
+    },
+    {
+      "epoch": 0.5484710474951203,
+      "grad_norm": 0.17759400606155396,
+      "learning_rate": 0.0001,
+      "loss": 1.8407,
+      "step": 843
+    },
+    {
+      "epoch": 0.5491216655823032,
+      "grad_norm": 0.22264355421066284,
+      "learning_rate": 0.0001,
+      "loss": 2.2869,
+      "step": 844
+    },
+    {
+      "epoch": 0.549772283669486,
+      "grad_norm": 0.20819737017154694,
+      "learning_rate": 0.0001,
+      "loss": 2.1209,
+      "step": 845
+    },
+    {
+      "epoch": 0.5504229017566689,
+      "grad_norm": 0.2194463461637497,
+      "learning_rate": 0.0001,
+      "loss": 2.1457,
+      "step": 846
+    },
+    {
+      "epoch": 0.5510735198438517,
+      "grad_norm": 0.19314661622047424,
+      "learning_rate": 0.0001,
+      "loss": 2.1063,
+      "step": 847
+    },
+    {
+      "epoch": 0.5517241379310345,
+      "grad_norm": 0.186354860663414,
+      "learning_rate": 0.0001,
+      "loss": 2.0833,
+      "step": 848
+    },
+    {
+      "epoch": 0.5523747560182173,
+      "grad_norm": 0.1862732619047165,
+      "learning_rate": 0.0001,
+      "loss": 1.9441,
+      "step": 849
+    },
+    {
+      "epoch": 0.5530253741054001,
+      "grad_norm": 0.24664181470870972,
+      "learning_rate": 0.0001,
+      "loss": 2.3277,
+      "step": 850
+    },
+    {
+      "epoch": 0.5536759921925829,
+      "grad_norm": 0.20182165503501892,
+      "learning_rate": 0.0001,
+      "loss": 2.1902,
+      "step": 851
+    },
+    {
+      "epoch": 0.5543266102797658,
+      "grad_norm": 0.2108999788761139,
+      "learning_rate": 0.0001,
+      "loss": 2.0826,
+      "step": 852
+    },
+    {
+      "epoch": 0.5549772283669486,
+      "grad_norm": 0.25388890504837036,
+      "learning_rate": 0.0001,
+      "loss": 2.5149,
+      "step": 853
+    },
+    {
+      "epoch": 0.5556278464541314,
+      "grad_norm": 0.2074718177318573,
+      "learning_rate": 0.0001,
+      "loss": 1.9135,
+      "step": 854
+    },
+    {
+      "epoch": 0.5562784645413142,
+      "grad_norm": 0.1992723047733307,
+      "learning_rate": 0.0001,
+      "loss": 2.186,
+      "step": 855
+    },
+    {
+      "epoch": 0.556929082628497,
+      "grad_norm": 0.18721085786819458,
+      "learning_rate": 0.0001,
+      "loss": 1.9453,
+      "step": 856
+    },
+    {
+      "epoch": 0.5575797007156799,
+      "grad_norm": 0.21606992185115814,
+      "learning_rate": 0.0001,
+      "loss": 2.1703,
+      "step": 857
+    },
+    {
+      "epoch": 0.5582303188028627,
+      "grad_norm": 0.2854723334312439,
+      "learning_rate": 0.0001,
+      "loss": 2.9538,
+      "step": 858
+    },
+    {
+      "epoch": 0.5588809368900456,
+      "grad_norm": 0.21503040194511414,
+      "learning_rate": 0.0001,
+      "loss": 2.0194,
+      "step": 859
+    },
+    {
+      "epoch": 0.5595315549772284,
+      "grad_norm": 0.2690679430961609,
+      "learning_rate": 0.0001,
+      "loss": 2.1562,
+      "step": 860
+    },
+    {
+      "epoch": 0.5601821730644112,
+      "grad_norm": 0.2811613976955414,
+      "learning_rate": 0.0001,
+      "loss": 2.2475,
+      "step": 861
+    },
+    {
+      "epoch": 0.560832791151594,
+      "grad_norm": 0.2551681697368622,
+      "learning_rate": 0.0001,
+      "loss": 2.5585,
+      "step": 862
+    },
+    {
+      "epoch": 0.5614834092387768,
+      "grad_norm": 0.21423856914043427,
+      "learning_rate": 0.0001,
+      "loss": 2.1194,
+      "step": 863
+    },
+    {
+      "epoch": 0.5621340273259596,
+      "grad_norm": 0.22121264040470123,
+      "learning_rate": 0.0001,
+      "loss": 1.9257,
+      "step": 864
+    },
+    {
+      "epoch": 0.5627846454131424,
+      "grad_norm": 0.38684332370758057,
+      "learning_rate": 0.0001,
+      "loss": 2.5203,
+      "step": 865
+    },
+    {
+      "epoch": 0.5634352635003254,
+      "grad_norm": 0.20299634337425232,
+      "learning_rate": 0.0001,
+      "loss": 2.0868,
+      "step": 866
+    },
+    {
+      "epoch": 0.5640858815875082,
+      "grad_norm": 0.33485493063926697,
+      "learning_rate": 0.0001,
+      "loss": 2.457,
+      "step": 867
+    },
+    {
+      "epoch": 0.564736499674691,
+      "grad_norm": 0.23778866231441498,
+      "learning_rate": 0.0001,
+      "loss": 1.9863,
+      "step": 868
+    },
+    {
+      "epoch": 0.5653871177618738,
+      "grad_norm": 0.18562458455562592,
+      "learning_rate": 0.0001,
+      "loss": 1.915,
+      "step": 869
+    },
+    {
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.3780176341533661,
+      "learning_rate": 0.0001,
+      "loss": 2.5518,
+      "step": 870
+    },
+    {
+      "epoch": 0.5666883539362394,
+      "grad_norm": 0.1924014538526535,
+      "learning_rate": 0.0001,
+      "loss": 2.0665,
+      "step": 871
+    },
+    {
+      "epoch": 0.5673389720234222,
+      "grad_norm": 0.19788160920143127,
+      "learning_rate": 0.0001,
+      "loss": 1.9408,
+      "step": 872
+    },
+    {
+      "epoch": 0.5679895901106051,
+      "grad_norm": 0.2435147911310196,
+      "learning_rate": 0.0001,
+      "loss": 2.3716,
+      "step": 873
+    },
+    {
+      "epoch": 0.5686402081977879,
+      "grad_norm": 0.2023211270570755,
+      "learning_rate": 0.0001,
+      "loss": 2.2786,
+      "step": 874
+    },
+    {
+      "epoch": 0.5692908262849707,
+      "grad_norm": 0.29936715960502625,
+      "learning_rate": 0.0001,
+      "loss": 2.6689,
+      "step": 875
+    },
+    {
+      "epoch": 0.5699414443721535,
+      "grad_norm": 0.18846483528614044,
+      "learning_rate": 0.0001,
+      "loss": 1.9436,
+      "step": 876
+    },
+    {
+      "epoch": 0.5705920624593364,
+      "grad_norm": 0.44592785835266113,
+      "learning_rate": 0.0001,
+      "loss": 2.8648,
+      "step": 877
+    },
+    {
+      "epoch": 0.5712426805465192,
+      "grad_norm": 0.221640944480896,
+      "learning_rate": 0.0001,
+      "loss": 2.1613,
+      "step": 878
+    },
+    {
+      "epoch": 0.571893298633702,
+      "grad_norm": 0.22345726191997528,
+      "learning_rate": 0.0001,
+      "loss": 2.076,
+      "step": 879
+    },
+    {
+      "epoch": 0.5725439167208849,
+      "grad_norm": 0.20094214379787445,
+      "learning_rate": 0.0001,
+      "loss": 2.0474,
+      "step": 880
+    },
+    {
+      "epoch": 0.5731945348080677,
+      "grad_norm": 0.1997043937444687,
+      "learning_rate": 0.0001,
+      "loss": 1.9812,
+      "step": 881
+    },
+    {
+      "epoch": 0.5738451528952505,
+      "grad_norm": 0.3758605420589447,
+      "learning_rate": 0.0001,
+      "loss": 2.8357,
+      "step": 882
+    },
+    {
+      "epoch": 0.5744957709824333,
+      "grad_norm": 0.2940578758716583,
+      "learning_rate": 0.0001,
+      "loss": 2.4955,
+      "step": 883
+    },
+    {
+      "epoch": 0.5751463890696161,
+      "grad_norm": 0.2434762865304947,
+      "learning_rate": 0.0001,
+      "loss": 2.0011,
+      "step": 884
+    },
+    {
+      "epoch": 0.5757970071567989,
+      "grad_norm": 0.24335308372974396,
+      "learning_rate": 0.0001,
+      "loss": 2.5458,
+      "step": 885
+    },
+    {
+      "epoch": 0.5764476252439817,
+      "grad_norm": 0.2063351422548294,
+      "learning_rate": 0.0001,
+      "loss": 1.9801,
+      "step": 886
+    },
+    {
+      "epoch": 0.5770982433311646,
+      "grad_norm": 0.35102301836013794,
+      "learning_rate": 0.0001,
+      "loss": 2.5647,
+      "step": 887
+    },
+    {
+      "epoch": 0.5777488614183475,
+      "grad_norm": 0.22332875430583954,
+      "learning_rate": 0.0001,
+      "loss": 2.0542,
+      "step": 888
+    },
+    {
+      "epoch": 0.5783994795055303,
+      "grad_norm": 0.2073124796152115,
+      "learning_rate": 0.0001,
+      "loss": 1.9348,
+      "step": 889
+    },
+    {
+      "epoch": 0.5790500975927131,
+      "grad_norm": 0.21079733967781067,
+      "learning_rate": 0.0001,
+      "loss": 1.9829,
+      "step": 890
+    },
+    {
+      "epoch": 0.5797007156798959,
+      "grad_norm": 0.2842913866043091,
+      "learning_rate": 0.0001,
+      "loss": 2.7215,
+      "step": 891
+    },
+    {
+      "epoch": 0.5803513337670787,
+      "grad_norm": 0.2807595133781433,
+      "learning_rate": 0.0001,
+      "loss": 2.1827,
+      "step": 892
+    },
+    {
+      "epoch": 0.5810019518542615,
+      "grad_norm": 0.24955599009990692,
+      "learning_rate": 0.0001,
+      "loss": 2.6246,
+      "step": 893
+    },
+    {
+      "epoch": 0.5816525699414443,
+      "grad_norm": 0.23281241953372955,
+      "learning_rate": 0.0001,
+      "loss": 2.3944,
+      "step": 894
+    },
+    {
+      "epoch": 0.5823031880286272,
+      "grad_norm": 0.2617682218551636,
+      "learning_rate": 0.0001,
+      "loss": 2.6147,
+      "step": 895
+    },
+    {
+      "epoch": 0.58295380611581,
+      "grad_norm": 0.1915360391139984,
+      "learning_rate": 0.0001,
+      "loss": 2.0095,
+      "step": 896
+    },
+    {
+      "epoch": 0.5836044242029929,
+      "grad_norm": 0.20270249247550964,
+      "learning_rate": 0.0001,
+      "loss": 1.8983,
+      "step": 897
+    },
+    {
+      "epoch": 0.5842550422901757,
+      "grad_norm": 0.21804624795913696,
+      "learning_rate": 0.0001,
+      "loss": 2.0425,
+      "step": 898
+    },
+    {
+      "epoch": 0.5849056603773585,
+      "grad_norm": 0.25326576828956604,
+      "learning_rate": 0.0001,
+      "loss": 2.4875,
+      "step": 899
+    },
+    {
+      "epoch": 0.5855562784645413,
+      "grad_norm": 0.21714434027671814,
+      "learning_rate": 0.0001,
+      "loss": 2.269,
+      "step": 900
+    },
+    {
+      "epoch": 0.5862068965517241,
+      "grad_norm": 0.22771766781806946,
+      "learning_rate": 0.0001,
+      "loss": 2.3039,
+      "step": 901
+    },
+    {
+      "epoch": 0.586857514638907,
+      "grad_norm": 0.3638748824596405,
+      "learning_rate": 0.0001,
+      "loss": 2.7448,
+      "step": 902
+    },
+    {
+      "epoch": 0.5875081327260898,
+      "grad_norm": 0.20194686949253082,
+      "learning_rate": 0.0001,
+      "loss": 2.0141,
+      "step": 903
+    },
+    {
+      "epoch": 0.5881587508132726,
+      "grad_norm": 0.187494158744812,
+      "learning_rate": 0.0001,
+      "loss": 2.1188,
+      "step": 904
+    },
+    {
+      "epoch": 0.5888093689004554,
+      "grad_norm": 0.23371635377407074,
+      "learning_rate": 0.0001,
+      "loss": 2.6014,
+      "step": 905
+    },
+    {
+      "epoch": 0.5894599869876382,
+      "grad_norm": 0.2642146050930023,
+      "learning_rate": 0.0001,
+      "loss": 2.2053,
+      "step": 906
+    },
+    {
+      "epoch": 0.590110605074821,
+      "grad_norm": 0.20045514404773712,
+      "learning_rate": 0.0001,
+      "loss": 2.1828,
+      "step": 907
+    },
+    {
+      "epoch": 0.5907612231620039,
+      "grad_norm": 0.22904321551322937,
+      "learning_rate": 0.0001,
+      "loss": 2.3128,
+      "step": 908
+    },
+    {
+      "epoch": 0.5914118412491868,
+      "grad_norm": 0.36857542395591736,
+      "learning_rate": 0.0001,
+      "loss": 3.3891,
+      "step": 909
+    },
+    {
+      "epoch": 0.5920624593363696,
+      "grad_norm": 0.3417764902114868,
+      "learning_rate": 0.0001,
+      "loss": 2.6737,
+      "step": 910
+    },
+    {
+      "epoch": 0.5927130774235524,
+      "grad_norm": 0.46861669421195984,
+      "learning_rate": 0.0001,
+      "loss": 2.5329,
+      "step": 911
+    },
+    {
+      "epoch": 0.5933636955107352,
+      "grad_norm": 0.32909440994262695,
+      "learning_rate": 0.0001,
+      "loss": 2.4894,
+      "step": 912
+    },
+    {
+      "epoch": 0.594014313597918,
+      "grad_norm": 0.2176060974597931,
+      "learning_rate": 0.0001,
+      "loss": 1.9696,
+      "step": 913
+    },
+    {
+      "epoch": 0.5946649316851008,
+      "grad_norm": 0.27317941188812256,
+      "learning_rate": 0.0001,
+      "loss": 2.2179,
+      "step": 914
+    },
+    {
+      "epoch": 0.5953155497722836,
+      "grad_norm": 0.267123281955719,
+      "learning_rate": 0.0001,
+      "loss": 2.5464,
+      "step": 915
+    },
+    {
+      "epoch": 0.5959661678594665,
+      "grad_norm": 0.320402055978775,
+      "learning_rate": 0.0001,
+      "loss": 2.5021,
+      "step": 916
+    },
+    {
+      "epoch": 0.5966167859466494,
+      "grad_norm": 0.20610998570919037,
+      "learning_rate": 0.0001,
+      "loss": 2.0586,
+      "step": 917
+    },
+    {
+      "epoch": 0.5972674040338322,
+      "grad_norm": 0.2108345478773117,
+      "learning_rate": 0.0001,
+      "loss": 2.3278,
+      "step": 918
+    },
+    {
+      "epoch": 0.597918022121015,
+      "grad_norm": 0.18368126451969147,
+      "learning_rate": 0.0001,
+      "loss": 2.1026,
+      "step": 919
+    },
+    {
+      "epoch": 0.5985686402081978,
+      "grad_norm": 0.20730890333652496,
+      "learning_rate": 0.0001,
+      "loss": 2.1936,
+      "step": 920
+    },
+    {
+      "epoch": 0.5992192582953806,
+      "grad_norm": 0.2921161651611328,
+      "learning_rate": 0.0001,
+      "loss": 2.5618,
+      "step": 921
+    },
+    {
+      "epoch": 0.5998698763825634,
+      "grad_norm": 0.23977220058441162,
+      "learning_rate": 0.0001,
+      "loss": 2.533,
+      "step": 922
+    },
+    {
+      "epoch": 0.6005204944697463,
+      "grad_norm": 0.25839105248451233,
+      "learning_rate": 0.0001,
+      "loss": 2.7033,
+      "step": 923
+    },
+    {
+      "epoch": 0.6011711125569291,
+      "grad_norm": 0.214335098862648,
+      "learning_rate": 0.0001,
+      "loss": 1.9153,
+      "step": 924
+    },
+    {
+      "epoch": 0.6018217306441119,
+      "grad_norm": 0.19577006995677948,
+      "learning_rate": 0.0001,
+      "loss": 1.8612,
+      "step": 925
+    },
+    {
+      "epoch": 0.6024723487312947,
+      "grad_norm": 0.22480078041553497,
+      "learning_rate": 0.0001,
+      "loss": 2.2383,
+      "step": 926
+    },
+    {
+      "epoch": 0.6031229668184775,
+      "grad_norm": 0.2090427577495575,
+      "learning_rate": 0.0001,
+      "loss": 1.9532,
+      "step": 927
+    },
+    {
+      "epoch": 0.6037735849056604,
+      "grad_norm": 0.21045666933059692,
+      "learning_rate": 0.0001,
+      "loss": 2.1285,
+      "step": 928
+    },
+    {
+      "epoch": 0.6044242029928432,
+      "grad_norm": 0.2302238792181015,
+      "learning_rate": 0.0001,
+      "loss": 2.5368,
+      "step": 929
+    },
+    {
+      "epoch": 0.605074821080026,
+      "grad_norm": 0.22230245172977448,
+      "learning_rate": 0.0001,
+      "loss": 2.0551,
+      "step": 930
+    },
+    {
+      "epoch": 0.6057254391672089,
+      "grad_norm": 0.2619292140007019,
+      "learning_rate": 0.0001,
+      "loss": 2.5149,
+      "step": 931
+    },
+    {
+      "epoch": 0.6063760572543917,
+      "grad_norm": 0.20247308909893036,
+      "learning_rate": 0.0001,
+      "loss": 2.0032,
+      "step": 932
+    },
+    {
+      "epoch": 0.6070266753415745,
+      "grad_norm": 0.19772449135780334,
+      "learning_rate": 0.0001,
+      "loss": 1.9627,
+      "step": 933
+    },
+    {
+      "epoch": 0.6076772934287573,
+      "grad_norm": 0.1917680948972702,
+      "learning_rate": 0.0001,
+      "loss": 1.9659,
+      "step": 934
+    },
+    {
+      "epoch": 0.6083279115159401,
+      "grad_norm": 0.3457018733024597,
+      "learning_rate": 0.0001,
+      "loss": 2.4537,
+      "step": 935
+    },
+    {
+      "epoch": 0.6089785296031229,
+      "grad_norm": 0.2027028501033783,
+      "learning_rate": 0.0001,
+      "loss": 2.1681,
+      "step": 936
+    },
+    {
+      "epoch": 0.6096291476903057,
+      "grad_norm": 0.24525637924671173,
+      "learning_rate": 0.0001,
+      "loss": 2.0816,
+      "step": 937
+    },
+    {
+      "epoch": 0.6102797657774887,
+      "grad_norm": 0.2690584659576416,
+      "learning_rate": 0.0001,
+      "loss": 2.7011,
+      "step": 938
+    },
+    {
+      "epoch": 0.6109303838646715,
+      "grad_norm": 0.20961976051330566,
+      "learning_rate": 0.0001,
+      "loss": 2.576,
+      "step": 939
+    },
+    {
+      "epoch": 0.6115810019518543,
+      "grad_norm": 0.21827319264411926,
+      "learning_rate": 0.0001,
+      "loss": 2.2605,
+      "step": 940
+    },
+    {
+      "epoch": 0.6122316200390371,
+      "grad_norm": 0.20448362827301025,
+      "learning_rate": 0.0001,
+      "loss": 1.9963,
+      "step": 941
+    },
+    {
+      "epoch": 0.6128822381262199,
+      "grad_norm": 0.2513864040374756,
+      "learning_rate": 0.0001,
+      "loss": 2.4111,
+      "step": 942
+    },
+    {
+      "epoch": 0.6135328562134027,
+      "grad_norm": 0.28347763419151306,
+      "learning_rate": 0.0001,
+      "loss": 2.3459,
+      "step": 943
+    },
+    {
+      "epoch": 0.6141834743005855,
+      "grad_norm": 0.20679716765880585,
+      "learning_rate": 0.0001,
+      "loss": 1.9423,
+      "step": 944
+    },
+    {
+      "epoch": 0.6148340923877684,
+      "grad_norm": 0.20072445273399353,
+      "learning_rate": 0.0001,
+      "loss": 2.2,
+      "step": 945
+    },
+    {
+      "epoch": 0.6154847104749512,
+      "grad_norm": 0.2190425843000412,
+      "learning_rate": 0.0001,
+      "loss": 2.358,
+      "step": 946
+    },
+    {
+      "epoch": 0.616135328562134,
+      "grad_norm": 0.2672726511955261,
+      "learning_rate": 0.0001,
+      "loss": 2.5034,
+      "step": 947
+    },
+    {
+      "epoch": 0.6167859466493169,
+      "grad_norm": 0.20329232513904572,
+      "learning_rate": 0.0001,
+      "loss": 2.2972,
+      "step": 948
+    },
+    {
+      "epoch": 0.6174365647364997,
+      "grad_norm": 0.21593444049358368,
+      "learning_rate": 0.0001,
+      "loss": 2.8221,
+      "step": 949
+    },
+    {
+      "epoch": 0.6180871828236825,
+      "grad_norm": 0.22062361240386963,
+      "learning_rate": 0.0001,
+      "loss": 2.2051,
+      "step": 950
+    },
+    {
+      "epoch": 0.6187378009108653,
+      "grad_norm": 0.20640413463115692,
+      "learning_rate": 0.0001,
+      "loss": 2.1973,
+      "step": 951
+    },
+    {
+      "epoch": 0.6193884189980482,
+      "grad_norm": 0.18919388949871063,
+      "learning_rate": 0.0001,
+      "loss": 2.1166,
+      "step": 952
+    },
+    {
+      "epoch": 0.620039037085231,
+      "grad_norm": 0.18566597998142242,
+      "learning_rate": 0.0001,
+      "loss": 1.9342,
+      "step": 953
+    },
+    {
+      "epoch": 0.6206896551724138,
+      "grad_norm": 0.3724953234195709,
+      "learning_rate": 0.0001,
+      "loss": 3.0303,
+      "step": 954
+    },
+    {
+      "epoch": 0.6213402732595966,
+      "grad_norm": 0.24559584259986877,
+      "learning_rate": 0.0001,
+      "loss": 2.387,
+      "step": 955
+    },
+    {
+      "epoch": 0.6219908913467794,
+      "grad_norm": 0.20384235680103302,
+      "learning_rate": 0.0001,
+      "loss": 2.1224,
+      "step": 956
+    },
+    {
+      "epoch": 0.6226415094339622,
+      "grad_norm": 0.3225831687450409,
+      "learning_rate": 0.0001,
+      "loss": 2.4856,
+      "step": 957
+    },
+    {
+      "epoch": 0.623292127521145,
+      "grad_norm": 0.21676267683506012,
+      "learning_rate": 0.0001,
+      "loss": 2.3457,
+      "step": 958
+    },
+    {
+      "epoch": 0.623942745608328,
+      "grad_norm": 0.21707187592983246,
+      "learning_rate": 0.0001,
+      "loss": 2.3985,
+      "step": 959
+    },
+    {
+      "epoch": 0.6245933636955108,
+      "grad_norm": 0.311277836561203,
+      "learning_rate": 0.0001,
+      "loss": 2.3087,
+      "step": 960
+    },
+    {
+      "epoch": 0.6252439817826936,
+      "grad_norm": 0.18904085457324982,
+      "learning_rate": 0.0001,
+      "loss": 1.9421,
+      "step": 961
+    },
+    {
+      "epoch": 0.6258945998698764,
+      "grad_norm": 0.39046210050582886,
+      "learning_rate": 0.0001,
+      "loss": 2.7524,
+      "step": 962
+    },
+    {
+      "epoch": 0.6265452179570592,
+      "grad_norm": 0.18455897271633148,
+      "learning_rate": 0.0001,
+      "loss": 1.7536,
+      "step": 963
+    },
+    {
+      "epoch": 0.627195836044242,
+      "grad_norm": 0.1874053180217743,
+      "learning_rate": 0.0001,
+      "loss": 2.0853,
+      "step": 964
+    },
+    {
+      "epoch": 0.6278464541314248,
+      "grad_norm": 0.24766068160533905,
+      "learning_rate": 0.0001,
+      "loss": 2.8099,
+      "step": 965
+    },
+    {
+      "epoch": 0.6284970722186076,
+      "grad_norm": 0.20977729558944702,
+      "learning_rate": 0.0001,
+      "loss": 2.0339,
+      "step": 966
+    },
+    {
+      "epoch": 0.6291476903057905,
+      "grad_norm": 0.2659202516078949,
+      "learning_rate": 0.0001,
+      "loss": 2.1282,
+      "step": 967
+    },
+    {
+      "epoch": 0.6297983083929733,
+      "grad_norm": 0.23760046064853668,
+      "learning_rate": 0.0001,
+      "loss": 2.4225,
+      "step": 968
+    },
+    {
+      "epoch": 0.6304489264801562,
+      "grad_norm": 0.1884511113166809,
+      "learning_rate": 0.0001,
+      "loss": 1.972,
+      "step": 969
+    },
+    {
+      "epoch": 0.631099544567339,
+      "grad_norm": 0.2816404402256012,
+      "learning_rate": 0.0001,
+      "loss": 2.6831,
+      "step": 970
+    },
+    {
+      "epoch": 0.6317501626545218,
+      "grad_norm": 0.1874386966228485,
+      "learning_rate": 0.0001,
+      "loss": 2.0042,
+      "step": 971
+    },
+    {
+      "epoch": 0.6324007807417046,
+      "grad_norm": 0.21592558920383453,
+      "learning_rate": 0.0001,
+      "loss": 2.338,
+      "step": 972
+    },
+    {
+      "epoch": 0.6330513988288874,
+      "grad_norm": 0.22190915048122406,
+      "learning_rate": 0.0001,
+      "loss": 2.23,
+      "step": 973
+    },
+    {
+      "epoch": 0.6337020169160703,
+      "grad_norm": 0.23270365595817566,
+      "learning_rate": 0.0001,
+      "loss": 2.1849,
+      "step": 974
+    },
+    {
+      "epoch": 0.6343526350032531,
+      "grad_norm": 0.20524165034294128,
+      "learning_rate": 0.0001,
+      "loss": 1.8509,
+      "step": 975
+    },
+    {
+      "epoch": 0.6350032530904359,
+      "grad_norm": 0.27826493978500366,
+      "learning_rate": 0.0001,
+      "loss": 2.6736,
+      "step": 976
+    },
+    {
+      "epoch": 0.6356538711776187,
+      "grad_norm": 0.19887575507164001,
+      "learning_rate": 0.0001,
+      "loss": 2.1369,
+      "step": 977
+    },
+    {
+      "epoch": 0.6363044892648015,
+      "grad_norm": 0.3760605752468109,
+      "learning_rate": 0.0001,
+      "loss": 2.7617,
+      "step": 978
+    },
+    {
+      "epoch": 0.6369551073519844,
+      "grad_norm": 0.2116486132144928,
+      "learning_rate": 0.0001,
+      "loss": 2.1353,
+      "step": 979
+    },
+    {
+      "epoch": 0.6376057254391672,
+      "grad_norm": 0.20685400068759918,
+      "learning_rate": 0.0001,
+      "loss": 2.2221,
+      "step": 980
+    },
+    {
+      "epoch": 0.6382563435263501,
+      "grad_norm": 0.25631460547447205,
+      "learning_rate": 0.0001,
+      "loss": 2.2755,
+      "step": 981
+    },
+    {
+      "epoch": 0.6389069616135329,
+      "grad_norm": 0.2831932604312897,
+      "learning_rate": 0.0001,
+      "loss": 2.2544,
+      "step": 982
+    },
+    {
+      "epoch": 0.6395575797007157,
+      "grad_norm": 0.19301310181617737,
+      "learning_rate": 0.0001,
+      "loss": 2.1736,
+      "step": 983
+    },
+    {
+      "epoch": 0.6402081977878985,
+      "grad_norm": 0.18511143326759338,
+      "learning_rate": 0.0001,
+      "loss": 1.8847,
+      "step": 984
+    },
+    {
+      "epoch": 0.6408588158750813,
+      "grad_norm": 0.23753167688846588,
+      "learning_rate": 0.0001,
+      "loss": 2.131,
+      "step": 985
+    },
+    {
+      "epoch": 0.6415094339622641,
+      "grad_norm": 0.24566152691841125,
+      "learning_rate": 0.0001,
+      "loss": 2.2071,
+      "step": 986
+    },
+    {
+      "epoch": 0.6421600520494469,
+      "grad_norm": 0.21481812000274658,
+      "learning_rate": 0.0001,
+      "loss": 2.0292,
+      "step": 987
+    },
+    {
+      "epoch": 0.6428106701366298,
+      "grad_norm": 0.3042278587818146,
+      "learning_rate": 0.0001,
+      "loss": 2.6444,
+      "step": 988
+    },
+    {
+      "epoch": 0.6434612882238127,
+      "grad_norm": 0.30741778016090393,
+      "learning_rate": 0.0001,
+      "loss": 2.5146,
+      "step": 989
+    },
+    {
+      "epoch": 0.6441119063109955,
+      "grad_norm": 0.40835896134376526,
+      "learning_rate": 0.0001,
+      "loss": 2.9053,
+      "step": 990
+    },
+    {
+      "epoch": 0.6447625243981783,
+      "grad_norm": 0.21121574938297272,
+      "learning_rate": 0.0001,
+      "loss": 2.4513,
+      "step": 991
+    },
+    {
+      "epoch": 0.6454131424853611,
+      "grad_norm": 0.2634606659412384,
+      "learning_rate": 0.0001,
+      "loss": 2.3141,
+      "step": 992
+    },
+    {
+      "epoch": 0.6460637605725439,
+      "grad_norm": 0.2463708072900772,
+      "learning_rate": 0.0001,
+      "loss": 2.4421,
+      "step": 993
+    },
+    {
+      "epoch": 0.6467143786597267,
+      "grad_norm": 0.25485244393348694,
+      "learning_rate": 0.0001,
+      "loss": 2.3788,
+      "step": 994
+    },
+    {
+      "epoch": 0.6473649967469096,
+      "grad_norm": 0.20773370563983917,
+      "learning_rate": 0.0001,
+      "loss": 1.9861,
+      "step": 995
+    },
+    {
+      "epoch": 0.6480156148340924,
+      "grad_norm": 0.20728078484535217,
+      "learning_rate": 0.0001,
+      "loss": 2.3341,
+      "step": 996
+    },
+    {
+      "epoch": 0.6486662329212752,
+      "grad_norm": 0.26925981044769287,
+      "learning_rate": 0.0001,
+      "loss": 2.9172,
+      "step": 997
+    },
+    {
+      "epoch": 0.649316851008458,
+      "grad_norm": 0.21403877437114716,
+      "learning_rate": 0.0001,
+      "loss": 2.1318,
+      "step": 998
+    },
+    {
+      "epoch": 0.6499674690956408,
+      "grad_norm": 0.2597064673900604,
+      "learning_rate": 0.0001,
+      "loss": 2.4316,
+      "step": 999
+    },
+    {
+      "epoch": 0.6506180871828237,
+      "grad_norm": 0.26858747005462646,
+      "learning_rate": 0.0001,
+      "loss": 2.2716,
+      "step": 1000
+    },
+    {
+      "epoch": 0.6512687052700065,
+      "grad_norm": 0.5603036880493164,
+      "learning_rate": 0.0001,
+      "loss": 3.1137,
+      "step": 1001
+    },
+    {
+      "epoch": 0.6519193233571894,
+      "grad_norm": 0.2423018366098404,
+      "learning_rate": 0.0001,
+      "loss": 2.2346,
+      "step": 1002
+    },
+    {
+      "epoch": 0.6525699414443722,
+      "grad_norm": 0.22914621233940125,
+      "learning_rate": 0.0001,
+      "loss": 2.2852,
+      "step": 1003
+    },
+    {
+      "epoch": 0.653220559531555,
+      "grad_norm": 0.22781658172607422,
+      "learning_rate": 0.0001,
+      "loss": 2.1961,
+      "step": 1004
+    },
+    {
+      "epoch": 0.6538711776187378,
+      "grad_norm": 0.2614092528820038,
+      "learning_rate": 0.0001,
+      "loss": 2.0631,
+      "step": 1005
+    },
+    {
+      "epoch": 0.6545217957059206,
+      "grad_norm": 0.23658867180347443,
+      "learning_rate": 0.0001,
+      "loss": 2.0379,
+      "step": 1006
+    },
+    {
+      "epoch": 0.6551724137931034,
+      "grad_norm": 0.20862211287021637,
+      "learning_rate": 0.0001,
+      "loss": 2.2786,
+      "step": 1007
+    },
+    {
+      "epoch": 0.6558230318802862,
+      "grad_norm": 0.2251960188150406,
+      "learning_rate": 0.0001,
+      "loss": 2.06,
+      "step": 1008
+    },
+    {
+      "epoch": 0.656473649967469,
+      "grad_norm": 0.2885074317455292,
+      "learning_rate": 0.0001,
+      "loss": 2.2583,
+      "step": 1009
+    },
+    {
+      "epoch": 0.657124268054652,
+      "grad_norm": 0.20309656858444214,
+      "learning_rate": 0.0001,
+      "loss": 2.1557,
+      "step": 1010
+    },
+    {
+      "epoch": 0.6577748861418348,
+      "grad_norm": 0.20139531791210175,
+      "learning_rate": 0.0001,
+      "loss": 2.3419,
+      "step": 1011
+    },
+    {
+      "epoch": 0.6584255042290176,
+      "grad_norm": 0.2853332757949829,
+      "learning_rate": 0.0001,
+      "loss": 2.1415,
+      "step": 1012
+    },
+    {
+      "epoch": 0.6590761223162004,
+      "grad_norm": 0.2907620966434479,
+      "learning_rate": 0.0001,
+      "loss": 2.4452,
+      "step": 1013
+    },
+    {
+      "epoch": 0.6597267404033832,
+      "grad_norm": 0.18982461094856262,
+      "learning_rate": 0.0001,
+      "loss": 2.0215,
+      "step": 1014
+    },
+    {
+      "epoch": 0.660377358490566,
+      "grad_norm": 0.20890061557292938,
+      "learning_rate": 0.0001,
+      "loss": 2.0383,
+      "step": 1015
+    },
+    {
+      "epoch": 0.6610279765777488,
+      "grad_norm": 0.21294118463993073,
+      "learning_rate": 0.0001,
+      "loss": 1.7722,
+      "step": 1016
+    },
+    {
+      "epoch": 0.6616785946649317,
+      "grad_norm": 0.22494040429592133,
+      "learning_rate": 0.0001,
+      "loss": 2.034,
+      "step": 1017
+    },
+    {
+      "epoch": 0.6623292127521145,
+      "grad_norm": 0.25089555978775024,
+      "learning_rate": 0.0001,
+      "loss": 2.3322,
+      "step": 1018
+    },
+    {
+      "epoch": 0.6629798308392973,
+      "grad_norm": 0.18898023664951324,
+      "learning_rate": 0.0001,
+      "loss": 1.9914,
+      "step": 1019
+    },
+    {
+      "epoch": 0.6636304489264802,
+      "grad_norm": 0.221091166138649,
+      "learning_rate": 0.0001,
+      "loss": 2.1613,
+      "step": 1020
+    },
+    {
+      "epoch": 0.664281067013663,
+      "grad_norm": 0.22317297756671906,
+      "learning_rate": 0.0001,
+      "loss": 2.3438,
+      "step": 1021
+    },
+    {
+      "epoch": 0.6649316851008458,
+      "grad_norm": 0.18826670944690704,
+      "learning_rate": 0.0001,
+      "loss": 2.0218,
+      "step": 1022
+    },
+    {
+      "epoch": 0.6655823031880286,
+      "grad_norm": 0.22612391412258148,
+      "learning_rate": 0.0001,
+      "loss": 2.2931,
+      "step": 1023
+    },
+    {
+      "epoch": 0.6662329212752115,
+      "grad_norm": 0.3006114959716797,
+      "learning_rate": 0.0001,
+      "loss": 2.4949,
+      "step": 1024
+    },
+    {
+      "epoch": 0.6668835393623943,
+      "grad_norm": 0.1835569143295288,
+      "learning_rate": 0.0001,
+      "loss": 1.9396,
+      "step": 1025
+    },
+    {
+      "epoch": 0.6675341574495771,
+      "grad_norm": 0.19352416694164276,
+      "learning_rate": 0.0001,
+      "loss": 2.0038,
+      "step": 1026
+    },
+    {
+      "epoch": 0.6681847755367599,
+      "grad_norm": 0.2259102463722229,
+      "learning_rate": 0.0001,
+      "loss": 2.1818,
+      "step": 1027
+    },
+    {
+      "epoch": 0.6688353936239427,
+      "grad_norm": 0.20237034559249878,
+      "learning_rate": 0.0001,
+      "loss": 2.3196,
+      "step": 1028
+    },
+    {
+      "epoch": 0.6694860117111255,
+      "grad_norm": 0.1844060719013214,
+      "learning_rate": 0.0001,
+      "loss": 2.1389,
+      "step": 1029
+    },
+    {
+      "epoch": 0.6701366297983083,
+      "grad_norm": 0.21057841181755066,
+      "learning_rate": 0.0001,
+      "loss": 2.0058,
+      "step": 1030
+    },
+    {
+      "epoch": 0.6707872478854913,
+      "grad_norm": 0.20054426789283752,
+      "learning_rate": 0.0001,
+      "loss": 2.2874,
+      "step": 1031
+    },
+    {
+      "epoch": 0.6714378659726741,
+      "grad_norm": 0.2507307529449463,
+      "learning_rate": 0.0001,
+      "loss": 2.4245,
+      "step": 1032
+    },
+    {
+      "epoch": 0.6720884840598569,
+      "grad_norm": 0.21066251397132874,
+      "learning_rate": 0.0001,
+      "loss": 2.1688,
+      "step": 1033
+    },
+    {
+      "epoch": 0.6727391021470397,
+      "grad_norm": 0.22210632264614105,
+      "learning_rate": 0.0001,
+      "loss": 2.1985,
+      "step": 1034
+    },
+    {
+      "epoch": 0.6733897202342225,
+      "grad_norm": 0.21617744863033295,
+      "learning_rate": 0.0001,
+      "loss": 2.5918,
+      "step": 1035
+    },
+    {
+      "epoch": 0.6740403383214053,
+      "grad_norm": 0.46473971009254456,
+      "learning_rate": 0.0001,
+      "loss": 2.9341,
+      "step": 1036
+    },
+    {
+      "epoch": 0.6746909564085881,
+      "grad_norm": 0.20464558899402618,
+      "learning_rate": 0.0001,
+      "loss": 2.1654,
+      "step": 1037
+    },
+    {
+      "epoch": 0.675341574495771,
+      "grad_norm": 0.212956503033638,
+      "learning_rate": 0.0001,
+      "loss": 2.1959,
+      "step": 1038
+    },
+    {
+      "epoch": 0.6759921925829538,
+      "grad_norm": 0.2572340667247772,
+      "learning_rate": 0.0001,
+      "loss": 2.4918,
+      "step": 1039
+    },
+    {
+      "epoch": 0.6766428106701367,
+      "grad_norm": 0.3264685273170471,
+      "learning_rate": 0.0001,
+      "loss": 2.8708,
+      "step": 1040
+    },
+    {
+      "epoch": 0.6772934287573195,
+      "grad_norm": 0.22119931876659393,
+      "learning_rate": 0.0001,
+      "loss": 2.2222,
+      "step": 1041
+    },
+    {
+      "epoch": 0.6779440468445023,
+      "grad_norm": 0.24374569952487946,
+      "learning_rate": 0.0001,
+      "loss": 2.2457,
+      "step": 1042
+    },
+    {
+      "epoch": 0.6785946649316851,
+      "grad_norm": 0.2548108696937561,
+      "learning_rate": 0.0001,
+      "loss": 2.485,
+      "step": 1043
+    },
+    {
+      "epoch": 0.6792452830188679,
+      "grad_norm": 0.20976418256759644,
+      "learning_rate": 0.0001,
+      "loss": 2.3068,
+      "step": 1044
+    },
+    {
+      "epoch": 0.6798959011060507,
+      "grad_norm": 0.25135618448257446,
+      "learning_rate": 0.0001,
+      "loss": 2.1083,
+      "step": 1045
+    },
+    {
+      "epoch": 0.6805465191932336,
+      "grad_norm": 0.2677728831768036,
+      "learning_rate": 0.0001,
+      "loss": 2.4257,
+      "step": 1046
+    },
+    {
+      "epoch": 0.6811971372804164,
+      "grad_norm": 0.20250125229358673,
+      "learning_rate": 0.0001,
+      "loss": 2.0643,
+      "step": 1047
+    },
+    {
+      "epoch": 0.6818477553675992,
+      "grad_norm": 0.20850299298763275,
+      "learning_rate": 0.0001,
+      "loss": 2.0383,
+      "step": 1048
+    },
+    {
+      "epoch": 0.682498373454782,
+      "grad_norm": 0.21116970479488373,
+      "learning_rate": 0.0001,
+      "loss": 2.0259,
+      "step": 1049
+    },
+    {
+      "epoch": 0.6831489915419648,
+      "grad_norm": 0.2572707235813141,
+      "learning_rate": 0.0001,
+      "loss": 2.1982,
+      "step": 1050
+    },
+    {
+      "epoch": 0.6837996096291477,
+      "grad_norm": 0.2010831981897354,
+      "learning_rate": 0.0001,
+      "loss": 2.0687,
+      "step": 1051
+    },
+    {
+      "epoch": 0.6844502277163305,
+      "grad_norm": 0.23995356261730194,
+      "learning_rate": 0.0001,
+      "loss": 2.1938,
+      "step": 1052
+    },
+    {
+      "epoch": 0.6851008458035134,
+      "grad_norm": 0.21428103744983673,
+      "learning_rate": 0.0001,
+      "loss": 2.2514,
+      "step": 1053
+    },
+    {
+      "epoch": 0.6857514638906962,
+      "grad_norm": 0.21370433270931244,
+      "learning_rate": 0.0001,
+      "loss": 2.2523,
+      "step": 1054
+    },
+    {
+      "epoch": 0.686402081977879,
+      "grad_norm": 0.2131800800561905,
+      "learning_rate": 0.0001,
+      "loss": 2.2413,
+      "step": 1055
+    },
+    {
+      "epoch": 0.6870527000650618,
+      "grad_norm": 0.20007681846618652,
+      "learning_rate": 0.0001,
+      "loss": 2.176,
+      "step": 1056
+    },
+    {
+      "epoch": 0.6877033181522446,
+      "grad_norm": 0.2108153998851776,
+      "learning_rate": 0.0001,
+      "loss": 2.1081,
+      "step": 1057
+    },
+    {
+      "epoch": 0.6883539362394274,
+      "grad_norm": 0.19952858984470367,
+      "learning_rate": 0.0001,
+      "loss": 2.0249,
+      "step": 1058
+    },
+    {
+      "epoch": 0.6890045543266102,
+      "grad_norm": 0.20590882003307343,
+      "learning_rate": 0.0001,
+      "loss": 2.1949,
+      "step": 1059
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "grad_norm": 0.2126530408859253,
+      "learning_rate": 0.0001,
+      "loss": 2.2726,
+      "step": 1060
+    },
+    {
+      "epoch": 0.690305790500976,
+      "grad_norm": 0.30162468552589417,
+      "learning_rate": 0.0001,
+      "loss": 2.5032,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6909564085881588,
+      "grad_norm": 0.24452462792396545,
+      "learning_rate": 0.0001,
+      "loss": 2.3021,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6916070266753416,
+      "grad_norm": 0.17819760739803314,
+      "learning_rate": 0.0001,
+      "loss": 1.9628,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6922576447625244,
+      "grad_norm": 0.17437471449375153,
+      "learning_rate": 0.0001,
+      "loss": 1.879,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6929082628497072,
+      "grad_norm": 0.3003963232040405,
+      "learning_rate": 0.0001,
+      "loss": 2.4695,
+      "step": 1065
+    },
+    {
+      "epoch": 0.69355888093689,
+      "grad_norm": 0.2007562667131424,
+      "learning_rate": 0.0001,
+      "loss": 1.9754,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6942094990240729,
+      "grad_norm": 0.21425336599349976,
+      "learning_rate": 0.0001,
+      "loss": 2.1767,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6948601171112557,
+      "grad_norm": 0.20287302136421204,
+      "learning_rate": 0.0001,
+      "loss": 1.9933,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6955107351984385,
+      "grad_norm": 0.2762700021266937,
+      "learning_rate": 0.0001,
+      "loss": 2.1079,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6961613532856213,
+      "grad_norm": 0.18358288705348969,
+      "learning_rate": 0.0001,
+      "loss": 1.9445,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6968119713728042,
+      "grad_norm": 0.21157526969909668,
+      "learning_rate": 0.0001,
+      "loss": 2.169,
+      "step": 1071
+    },
+    {
+      "epoch": 0.697462589459987,
+      "grad_norm": 0.1847715675830841,
+      "learning_rate": 0.0001,
+      "loss": 2.0757,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6981132075471698,
+      "grad_norm": 0.1923181712627411,
+      "learning_rate": 0.0001,
+      "loss": 2.2365,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6987638256343527,
+      "grad_norm": 0.26491835713386536,
+      "learning_rate": 0.0001,
+      "loss": 2.4613,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6994144437215355,
+      "grad_norm": 0.17674419283866882,
+      "learning_rate": 0.0001,
+      "loss": 1.9706,
+      "step": 1075
+    },
+    {
+      "epoch": 0.7000650618087183,
+      "grad_norm": 0.19894379377365112,
+      "learning_rate": 0.0001,
+      "loss": 1.9227,
+      "step": 1076
+    },
+    {
+      "epoch": 0.7007156798959011,
+      "grad_norm": 0.19496971368789673,
+      "learning_rate": 0.0001,
+      "loss": 2.1783,
+      "step": 1077
+    },
+    {
+      "epoch": 0.7013662979830839,
+      "grad_norm": 0.20685461163520813,
+      "learning_rate": 0.0001,
+      "loss": 2.1542,
+      "step": 1078
+    },
+    {
+      "epoch": 0.7020169160702667,
+      "grad_norm": 0.23061524331569672,
+      "learning_rate": 0.0001,
+      "loss": 2.3346,
+      "step": 1079
+    },
+    {
+      "epoch": 0.7026675341574495,
+      "grad_norm": 0.2044321447610855,
+      "learning_rate": 0.0001,
+      "loss": 2.0157,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7033181522446325,
+      "grad_norm": 0.18851466476917267,
+      "learning_rate": 0.0001,
+      "loss": 2.2045,
+      "step": 1081
+    },
+    {
+      "epoch": 0.7039687703318153,
+      "grad_norm": 0.18530018627643585,
+      "learning_rate": 0.0001,
+      "loss": 2.0695,
+      "step": 1082
+    },
+    {
+      "epoch": 0.7046193884189981,
+      "grad_norm": 0.23562023043632507,
+      "learning_rate": 0.0001,
+      "loss": 2.3919,
+      "step": 1083
+    },
+    {
+      "epoch": 0.7052700065061809,
+      "grad_norm": 0.22246116399765015,
+      "learning_rate": 0.0001,
+      "loss": 2.5821,
+      "step": 1084
+    },
+    {
+      "epoch": 0.7059206245933637,
+      "grad_norm": 0.2134729027748108,
+      "learning_rate": 0.0001,
+      "loss": 2.2181,
+      "step": 1085
+    },
+    {
+      "epoch": 0.7065712426805465,
+      "grad_norm": 0.29674917459487915,
+      "learning_rate": 0.0001,
+      "loss": 2.5069,
+      "step": 1086
+    },
+    {
+      "epoch": 0.7072218607677293,
+      "grad_norm": 0.2098974883556366,
+      "learning_rate": 0.0001,
+      "loss": 2.3307,
+      "step": 1087
+    },
+    {
+      "epoch": 0.7078724788549121,
+      "grad_norm": 0.27041876316070557,
+      "learning_rate": 0.0001,
+      "loss": 2.8081,
+      "step": 1088
+    },
+    {
+      "epoch": 0.708523096942095,
+      "grad_norm": 0.19734299182891846,
+      "learning_rate": 0.0001,
+      "loss": 2.0588,
+      "step": 1089
+    },
+    {
+      "epoch": 0.7091737150292778,
+      "grad_norm": 0.22952257096767426,
+      "learning_rate": 0.0001,
+      "loss": 2.2607,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7098243331164606,
+      "grad_norm": 0.20846691727638245,
+      "learning_rate": 0.0001,
+      "loss": 2.1657,
+      "step": 1091
+    },
+    {
+      "epoch": 0.7104749512036435,
+      "grad_norm": 0.19664259254932404,
+      "learning_rate": 0.0001,
+      "loss": 2.1256,
+      "step": 1092
+    },
+    {
+      "epoch": 0.7111255692908263,
+      "grad_norm": 0.23994791507720947,
+      "learning_rate": 0.0001,
+      "loss": 2.5377,
+      "step": 1093
+    },
+    {
+      "epoch": 0.7117761873780091,
+      "grad_norm": 0.22439789772033691,
+      "learning_rate": 0.0001,
+      "loss": 2.6225,
+      "step": 1094
+    },
+    {
+      "epoch": 0.7124268054651919,
+      "grad_norm": 0.20211316645145416,
+      "learning_rate": 0.0001,
+      "loss": 2.0582,
+      "step": 1095
+    },
+    {
+      "epoch": 0.7130774235523748,
+      "grad_norm": 0.23308198153972626,
+      "learning_rate": 0.0001,
+      "loss": 2.4341,
+      "step": 1096
+    },
+    {
+      "epoch": 0.7137280416395576,
+      "grad_norm": 0.17806245386600494,
+      "learning_rate": 0.0001,
+      "loss": 2.0211,
+      "step": 1097
+    },
+    {
+      "epoch": 0.7143786597267404,
+      "grad_norm": 0.20525243878364563,
+      "learning_rate": 0.0001,
+      "loss": 2.1248,
+      "step": 1098
+    },
+    {
+      "epoch": 0.7150292778139232,
+      "grad_norm": 0.22835716605186462,
+      "learning_rate": 0.0001,
+      "loss": 2.2993,
+      "step": 1099
+    },
+    {
+      "epoch": 0.715679895901106,
+      "grad_norm": 0.37078213691711426,
+      "learning_rate": 0.0001,
+      "loss": 3.1289,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7163305139882888,
+      "grad_norm": 0.22253082692623138,
+      "learning_rate": 0.0001,
+      "loss": 2.2304,
+      "step": 1101
+    },
+    {
+      "epoch": 0.7169811320754716,
+      "grad_norm": 0.20494401454925537,
+      "learning_rate": 0.0001,
+      "loss": 1.9473,
+      "step": 1102
+    },
+    {
+      "epoch": 0.7176317501626546,
+      "grad_norm": 0.22128112614154816,
+      "learning_rate": 0.0001,
+      "loss": 1.993,
+      "step": 1103
+    },
+    {
+      "epoch": 0.7182823682498374,
+      "grad_norm": 0.20786182582378387,
+      "learning_rate": 0.0001,
+      "loss": 2.0048,
+      "step": 1104
+    },
+    {
+      "epoch": 0.7189329863370202,
+      "grad_norm": 0.27697819471359253,
+      "learning_rate": 0.0001,
+      "loss": 2.372,
+      "step": 1105
+    },
+    {
+      "epoch": 0.719583604424203,
+      "grad_norm": 0.26237788796424866,
+      "learning_rate": 0.0001,
+      "loss": 1.9573,
+      "step": 1106
+    },
+    {
+      "epoch": 0.7202342225113858,
+      "grad_norm": 0.2544906437397003,
+      "learning_rate": 0.0001,
+      "loss": 2.2805,
+      "step": 1107
+    },
+    {
+      "epoch": 0.7208848405985686,
+      "grad_norm": 0.2175043374300003,
+      "learning_rate": 0.0001,
+      "loss": 2.3201,
+      "step": 1108
+    },
+    {
+      "epoch": 0.7215354586857514,
+      "grad_norm": 0.19637277722358704,
+      "learning_rate": 0.0001,
+      "loss": 1.8868,
+      "step": 1109
+    },
+    {
+      "epoch": 0.7221860767729343,
+      "grad_norm": 0.19888024032115936,
+      "learning_rate": 0.0001,
+      "loss": 2.0324,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7228366948601171,
+      "grad_norm": 0.20008981227874756,
+      "learning_rate": 0.0001,
+      "loss": 2.2898,
+      "step": 1111
+    },
+    {
+      "epoch": 0.7234873129473,
+      "grad_norm": 0.25185343623161316,
+      "learning_rate": 0.0001,
+      "loss": 2.2424,
+      "step": 1112
+    },
+    {
+      "epoch": 0.7241379310344828,
+      "grad_norm": 0.2434062957763672,
+      "learning_rate": 0.0001,
+      "loss": 2.2884,
+      "step": 1113
+    },
+    {
+      "epoch": 0.7247885491216656,
+      "grad_norm": 0.2278825044631958,
+      "learning_rate": 0.0001,
+      "loss": 2.1751,
+      "step": 1114
+    },
+    {
+      "epoch": 0.7254391672088484,
+      "grad_norm": 0.23180316388607025,
+      "learning_rate": 0.0001,
+      "loss": 2.6033,
+      "step": 1115
+    },
+    {
+      "epoch": 0.7260897852960312,
+      "grad_norm": 0.18574117124080658,
+      "learning_rate": 0.0001,
+      "loss": 2.3172,
+      "step": 1116
+    },
+    {
+      "epoch": 0.7267404033832141,
+      "grad_norm": 0.286155641078949,
+      "learning_rate": 0.0001,
+      "loss": 2.0482,
+      "step": 1117
+    },
+    {
+      "epoch": 0.7273910214703969,
+      "grad_norm": 0.1757357120513916,
+      "learning_rate": 0.0001,
+      "loss": 1.8881,
+      "step": 1118
+    },
+    {
+      "epoch": 0.7280416395575797,
+      "grad_norm": 0.25008201599121094,
+      "learning_rate": 0.0001,
+      "loss": 2.3797,
+      "step": 1119
+    },
+    {
+      "epoch": 0.7286922576447625,
+      "grad_norm": 0.29816892743110657,
+      "learning_rate": 0.0001,
+      "loss": 2.9163,
+      "step": 1120
+    },
+    {
+      "epoch": 0.7293428757319453,
+      "grad_norm": 0.1951293647289276,
+      "learning_rate": 0.0001,
+      "loss": 2.0613,
+      "step": 1121
+    },
+    {
+      "epoch": 0.7299934938191281,
+      "grad_norm": 0.23593062162399292,
+      "learning_rate": 0.0001,
+      "loss": 2.2103,
+      "step": 1122
+    },
+    {
+      "epoch": 0.730644111906311,
+      "grad_norm": 0.18619036674499512,
+      "learning_rate": 0.0001,
+      "loss": 1.9223,
+      "step": 1123
+    },
+    {
+      "epoch": 0.7312947299934938,
+      "grad_norm": 0.20853224396705627,
+      "learning_rate": 0.0001,
+      "loss": 2.2651,
+      "step": 1124
+    },
+    {
+      "epoch": 0.7319453480806767,
+      "grad_norm": 0.27427271008491516,
+      "learning_rate": 0.0001,
+      "loss": 2.3866,
+      "step": 1125
+    },
+    {
+      "epoch": 0.7325959661678595,
+      "grad_norm": 0.35531318187713623,
+      "learning_rate": 0.0001,
+      "loss": 2.8333,
+      "step": 1126
+    },
+    {
+      "epoch": 0.7332465842550423,
+      "grad_norm": 0.21375155448913574,
+      "learning_rate": 0.0001,
+      "loss": 2.0703,
+      "step": 1127
+    },
+    {
+      "epoch": 0.7338972023422251,
+      "grad_norm": 0.24240247905254364,
+      "learning_rate": 0.0001,
+      "loss": 2.3032,
+      "step": 1128
+    },
+    {
+      "epoch": 0.7345478204294079,
+      "grad_norm": 0.2277136594057083,
+      "learning_rate": 0.0001,
+      "loss": 2.585,
+      "step": 1129
+    },
+    {
+      "epoch": 0.7351984385165907,
+      "grad_norm": 0.20665140450000763,
+      "learning_rate": 0.0001,
+      "loss": 2.1351,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7358490566037735,
+      "grad_norm": 0.2534540891647339,
+      "learning_rate": 0.0001,
+      "loss": 2.5023,
+      "step": 1131
+    },
+    {
+      "epoch": 0.7364996746909565,
+      "grad_norm": 0.19695554673671722,
+      "learning_rate": 0.0001,
+      "loss": 1.9286,
+      "step": 1132
+    },
+    {
+      "epoch": 0.7371502927781393,
+      "grad_norm": 0.18500645458698273,
+      "learning_rate": 0.0001,
+      "loss": 2.0609,
+      "step": 1133
+    },
+    {
+      "epoch": 0.7378009108653221,
+      "grad_norm": 0.2103162556886673,
+      "learning_rate": 0.0001,
+      "loss": 2.2247,
+      "step": 1134
+    },
+    {
+      "epoch": 0.7384515289525049,
+      "grad_norm": 0.20303300023078918,
+      "learning_rate": 0.0001,
+      "loss": 2.1164,
+      "step": 1135
+    },
+    {
+      "epoch": 0.7391021470396877,
+      "grad_norm": 0.23574739694595337,
+      "learning_rate": 0.0001,
+      "loss": 2.6325,
+      "step": 1136
+    },
+    {
+      "epoch": 0.7397527651268705,
+      "grad_norm": 0.2764929234981537,
+      "learning_rate": 0.0001,
+      "loss": 2.3049,
+      "step": 1137
+    },
+    {
+      "epoch": 0.7404033832140533,
+      "grad_norm": 0.23995018005371094,
+      "learning_rate": 0.0001,
+      "loss": 2.3196,
+      "step": 1138
+    },
+    {
+      "epoch": 0.7410540013012362,
+      "grad_norm": 0.19074063003063202,
+      "learning_rate": 0.0001,
+      "loss": 2.1566,
+      "step": 1139
+    },
+    {
+      "epoch": 0.741704619388419,
+      "grad_norm": 0.18186306953430176,
+      "learning_rate": 0.0001,
+      "loss": 1.9629,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7423552374756018,
+      "grad_norm": 0.23841345310211182,
+      "learning_rate": 0.0001,
+      "loss": 2.1942,
+      "step": 1141
+    },
+    {
+      "epoch": 0.7430058555627846,
+      "grad_norm": 0.19697019457817078,
+      "learning_rate": 0.0001,
+      "loss": 2.0186,
+      "step": 1142
+    },
+    {
+      "epoch": 0.7436564736499675,
+      "grad_norm": 0.2117876410484314,
+      "learning_rate": 0.0001,
+      "loss": 2.4395,
+      "step": 1143
+    },
+    {
+      "epoch": 0.7443070917371503,
+      "grad_norm": 0.26921918988227844,
+      "learning_rate": 0.0001,
+      "loss": 2.4332,
+      "step": 1144
+    },
+    {
+      "epoch": 0.7449577098243331,
+      "grad_norm": 0.18999671936035156,
+      "learning_rate": 0.0001,
+      "loss": 2.0209,
+      "step": 1145
+    },
+    {
+      "epoch": 0.745608327911516,
+      "grad_norm": 0.22686484456062317,
+      "learning_rate": 0.0001,
+      "loss": 2.4369,
+      "step": 1146
+    },
+    {
+      "epoch": 0.7462589459986988,
+      "grad_norm": 0.22974656522274017,
+      "learning_rate": 0.0001,
+      "loss": 2.3737,
+      "step": 1147
+    },
+    {
+      "epoch": 0.7469095640858816,
+      "grad_norm": 0.19007977843284607,
+      "learning_rate": 0.0001,
+      "loss": 2.145,
+      "step": 1148
+    },
+    {
+      "epoch": 0.7475601821730644,
+      "grad_norm": 0.23000845313072205,
+      "learning_rate": 0.0001,
+      "loss": 2.0555,
+      "step": 1149
+    },
+    {
+      "epoch": 0.7482108002602472,
+      "grad_norm": 0.33339783549308777,
+      "learning_rate": 0.0001,
+      "loss": 2.7318,
+      "step": 1150
+    },
+    {
+      "epoch": 0.74886141834743,
+      "grad_norm": 0.18458595871925354,
+      "learning_rate": 0.0001,
+      "loss": 1.7868,
+      "step": 1151
+    },
+    {
+      "epoch": 0.7495120364346128,
+      "grad_norm": 0.2283509373664856,
+      "learning_rate": 0.0001,
+      "loss": 2.2609,
+      "step": 1152
+    },
+    {
+      "epoch": 0.7501626545217958,
+      "grad_norm": 0.31175729632377625,
+      "learning_rate": 0.0001,
+      "loss": 2.5524,
+      "step": 1153
+    },
+    {
+      "epoch": 0.7508132726089786,
+      "grad_norm": 0.18617112934589386,
+      "learning_rate": 0.0001,
+      "loss": 2.2029,
+      "step": 1154
+    },
+    {
+      "epoch": 0.7514638906961614,
+      "grad_norm": 0.28690317273139954,
+      "learning_rate": 0.0001,
+      "loss": 2.4705,
+      "step": 1155
+    },
+    {
+      "epoch": 0.7521145087833442,
+      "grad_norm": 0.2267671674489975,
+      "learning_rate": 0.0001,
+      "loss": 2.1093,
+      "step": 1156
+    },
+    {
+      "epoch": 0.752765126870527,
+      "grad_norm": 0.21956512331962585,
+      "learning_rate": 0.0001,
+      "loss": 2.0962,
+      "step": 1157
+    },
+    {
+      "epoch": 0.7534157449577098,
+      "grad_norm": 0.2681393027305603,
+      "learning_rate": 0.0001,
+      "loss": 2.35,
+      "step": 1158
+    },
+    {
+      "epoch": 0.7540663630448926,
+      "grad_norm": 0.23306699097156525,
+      "learning_rate": 0.0001,
+      "loss": 2.4911,
+      "step": 1159
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.3148876428604126,
+      "learning_rate": 0.0001,
+      "loss": 2.8802,
+      "step": 1160
+    },
+    {
+      "epoch": 0.7553675992192583,
+      "grad_norm": 0.2260347157716751,
+      "learning_rate": 0.0001,
+      "loss": 1.9286,
+      "step": 1161
+    },
+    {
+      "epoch": 0.7560182173064411,
+      "grad_norm": 0.24939195811748505,
+      "learning_rate": 0.0001,
+      "loss": 2.3544,
+      "step": 1162
+    },
+    {
+      "epoch": 0.756668835393624,
+      "grad_norm": 0.21007601916790009,
+      "learning_rate": 0.0001,
+      "loss": 2.0132,
+      "step": 1163
+    },
+    {
+      "epoch": 0.7573194534808068,
+      "grad_norm": 0.2570975720882416,
+      "learning_rate": 0.0001,
+      "loss": 1.9665,
+      "step": 1164
+    },
+    {
+      "epoch": 0.7579700715679896,
+      "grad_norm": 0.2818357050418854,
+      "learning_rate": 0.0001,
+      "loss": 2.2252,
+      "step": 1165
+    },
+    {
+      "epoch": 0.7586206896551724,
+      "grad_norm": 0.22388941049575806,
+      "learning_rate": 0.0001,
+      "loss": 2.4553,
+      "step": 1166
+    },
+    {
+      "epoch": 0.7592713077423552,
+      "grad_norm": 0.22799374163150787,
+      "learning_rate": 0.0001,
+      "loss": 2.4447,
+      "step": 1167
+    },
+    {
+      "epoch": 0.7599219258295381,
+      "grad_norm": 0.2610357105731964,
+      "learning_rate": 0.0001,
+      "loss": 2.4024,
+      "step": 1168
+    },
+    {
+      "epoch": 0.7605725439167209,
+      "grad_norm": 0.39793217182159424,
+      "learning_rate": 0.0001,
+      "loss": 3.1529,
+      "step": 1169
+    },
+    {
+      "epoch": 0.7612231620039037,
+      "grad_norm": 0.19805116951465607,
+      "learning_rate": 0.0001,
+      "loss": 1.9483,
+      "step": 1170
+    },
+    {
+      "epoch": 0.7618737800910865,
+      "grad_norm": 0.208368182182312,
+      "learning_rate": 0.0001,
+      "loss": 2.1785,
+      "step": 1171
+    },
+    {
+      "epoch": 0.7625243981782693,
+      "grad_norm": 0.25101637840270996,
+      "learning_rate": 0.0001,
+      "loss": 2.2517,
+      "step": 1172
+    },
+    {
+      "epoch": 0.7631750162654521,
+      "grad_norm": 0.27432793378829956,
+      "learning_rate": 0.0001,
+      "loss": 2.4759,
+      "step": 1173
+    },
+    {
+      "epoch": 0.763825634352635,
+      "grad_norm": 0.18746371567249298,
+      "learning_rate": 0.0001,
+      "loss": 2.0188,
+      "step": 1174
+    },
+    {
+      "epoch": 0.7644762524398179,
+      "grad_norm": 0.2882263958454132,
+      "learning_rate": 0.0001,
+      "loss": 2.2948,
+      "step": 1175
+    },
+    {
+      "epoch": 0.7651268705270007,
+      "grad_norm": 0.22075092792510986,
+      "learning_rate": 0.0001,
+      "loss": 2.4894,
+      "step": 1176
+    },
+    {
+      "epoch": 0.7657774886141835,
+      "grad_norm": 0.20792776346206665,
+      "learning_rate": 0.0001,
+      "loss": 1.8502,
+      "step": 1177
+    },
+    {
+      "epoch": 0.7664281067013663,
+      "grad_norm": 0.2436477392911911,
+      "learning_rate": 0.0001,
+      "loss": 2.1296,
+      "step": 1178
+    },
+    {
+      "epoch": 0.7670787247885491,
+      "grad_norm": 0.2839182913303375,
+      "learning_rate": 0.0001,
+      "loss": 2.8409,
+      "step": 1179
+    },
+    {
+      "epoch": 0.7677293428757319,
+      "grad_norm": 0.1826743334531784,
+      "learning_rate": 0.0001,
+      "loss": 1.941,
+      "step": 1180
+    },
+    {
+      "epoch": 0.7683799609629147,
+      "grad_norm": 0.2757255434989929,
+      "learning_rate": 0.0001,
+      "loss": 2.7297,
+      "step": 1181
+    },
+    {
+      "epoch": 0.7690305790500976,
+      "grad_norm": 0.23313826322555542,
+      "learning_rate": 0.0001,
+      "loss": 2.8796,
+      "step": 1182
+    },
+    {
+      "epoch": 0.7696811971372804,
+      "grad_norm": 0.28900882601737976,
+      "learning_rate": 0.0001,
+      "loss": 2.313,
+      "step": 1183
+    },
+    {
+      "epoch": 0.7703318152244633,
+      "grad_norm": 0.32883039116859436,
+      "learning_rate": 0.0001,
+      "loss": 3.041,
+      "step": 1184
+    },
+    {
+      "epoch": 0.7709824333116461,
+      "grad_norm": 0.2116912454366684,
+      "learning_rate": 0.0001,
+      "loss": 1.9891,
+      "step": 1185
+    },
+    {
+      "epoch": 0.7716330513988289,
+      "grad_norm": 0.2055017203092575,
+      "learning_rate": 0.0001,
+      "loss": 1.9567,
+      "step": 1186
+    },
+    {
+      "epoch": 0.7722836694860117,
+      "grad_norm": 0.2978801131248474,
+      "learning_rate": 0.0001,
+      "loss": 2.3322,
+      "step": 1187
+    },
+    {
+      "epoch": 0.7729342875731945,
+      "grad_norm": 0.21910034120082855,
+      "learning_rate": 0.0001,
+      "loss": 2.0262,
+      "step": 1188
+    },
+    {
+      "epoch": 0.7735849056603774,
+      "grad_norm": 0.19952894747257233,
+      "learning_rate": 0.0001,
+      "loss": 2.0621,
+      "step": 1189
+    },
+    {
+      "epoch": 0.7742355237475602,
+      "grad_norm": 0.20744554698467255,
+      "learning_rate": 0.0001,
+      "loss": 2.1154,
+      "step": 1190
+    },
+    {
+      "epoch": 0.774886141834743,
+      "grad_norm": 0.23886847496032715,
+      "learning_rate": 0.0001,
+      "loss": 2.3023,
+      "step": 1191
+    },
+    {
+      "epoch": 0.7755367599219258,
+      "grad_norm": 0.20722374320030212,
+      "learning_rate": 0.0001,
+      "loss": 2.2384,
+      "step": 1192
+    },
+    {
+      "epoch": 0.7761873780091086,
+      "grad_norm": 0.23317816853523254,
+      "learning_rate": 0.0001,
+      "loss": 2.6381,
+      "step": 1193
+    },
+    {
+      "epoch": 0.7768379960962914,
+      "grad_norm": 0.2527480125427246,
+      "learning_rate": 0.0001,
+      "loss": 2.1711,
+      "step": 1194
+    },
+    {
+      "epoch": 0.7774886141834743,
+      "grad_norm": 0.23817451298236847,
+      "learning_rate": 0.0001,
+      "loss": 2.6561,
+      "step": 1195
+    },
+    {
+      "epoch": 0.7781392322706572,
+      "grad_norm": 0.2609005570411682,
+      "learning_rate": 0.0001,
+      "loss": 2.5488,
+      "step": 1196
+    },
+    {
+      "epoch": 0.77878985035784,
+      "grad_norm": 0.19870908558368683,
+      "learning_rate": 0.0001,
+      "loss": 2.0435,
+      "step": 1197
+    },
+    {
+      "epoch": 0.7794404684450228,
+      "grad_norm": 0.20385386049747467,
+      "learning_rate": 0.0001,
+      "loss": 1.9711,
+      "step": 1198
+    },
+    {
+      "epoch": 0.7800910865322056,
+      "grad_norm": 0.20179738104343414,
+      "learning_rate": 0.0001,
+      "loss": 2.0247,
+      "step": 1199
+    },
+    {
+      "epoch": 0.7807417046193884,
+      "grad_norm": 0.40090981125831604,
+      "learning_rate": 0.0001,
+      "loss": 2.795,
+      "step": 1200
+    },
+    {
+      "epoch": 0.7813923227065712,
+      "grad_norm": 0.1885748654603958,
+      "learning_rate": 0.0001,
+      "loss": 2.1588,
+      "step": 1201
+    },
+    {
+      "epoch": 0.782042940793754,
+      "grad_norm": 0.21952667832374573,
+      "learning_rate": 0.0001,
+      "loss": 2.0901,
+      "step": 1202
+    },
+    {
+      "epoch": 0.7826935588809368,
+      "grad_norm": 0.2344968616962433,
+      "learning_rate": 0.0001,
+      "loss": 1.9943,
+      "step": 1203
+    },
+    {
+      "epoch": 0.7833441769681198,
+      "grad_norm": 0.3153589069843292,
+      "learning_rate": 0.0001,
+      "loss": 2.59,
+      "step": 1204
+    },
+    {
+      "epoch": 0.7839947950553026,
+      "grad_norm": 0.1870599389076233,
+      "learning_rate": 0.0001,
+      "loss": 1.9435,
+      "step": 1205
+    },
+    {
+      "epoch": 0.7846454131424854,
+      "grad_norm": 0.189214825630188,
+      "learning_rate": 0.0001,
+      "loss": 2.128,
+      "step": 1206
+    },
+    {
+      "epoch": 0.7852960312296682,
+      "grad_norm": 0.22551633417606354,
+      "learning_rate": 0.0001,
+      "loss": 2.3913,
+      "step": 1207
+    },
+    {
+      "epoch": 0.785946649316851,
+      "grad_norm": 0.19963033497333527,
+      "learning_rate": 0.0001,
+      "loss": 2.1456,
+      "step": 1208
+    },
+    {
+      "epoch": 0.7865972674040338,
+      "grad_norm": 0.2087828814983368,
+      "learning_rate": 0.0001,
+      "loss": 2.3486,
+      "step": 1209
+    },
+    {
+      "epoch": 0.7872478854912166,
+      "grad_norm": 0.19814416766166687,
+      "learning_rate": 0.0001,
+      "loss": 2.0208,
+      "step": 1210
+    },
+    {
+      "epoch": 0.7878985035783995,
+      "grad_norm": 0.20670342445373535,
+      "learning_rate": 0.0001,
+      "loss": 2.1276,
+      "step": 1211
+    },
+    {
+      "epoch": 0.7885491216655823,
+      "grad_norm": 0.1881658136844635,
+      "learning_rate": 0.0001,
+      "loss": 2.0502,
+      "step": 1212
+    },
+    {
+      "epoch": 0.7891997397527651,
+      "grad_norm": 0.2015887349843979,
+      "learning_rate": 0.0001,
+      "loss": 2.2935,
+      "step": 1213
+    },
+    {
+      "epoch": 0.789850357839948,
+      "grad_norm": 0.23532694578170776,
+      "learning_rate": 0.0001,
+      "loss": 2.8046,
+      "step": 1214
+    },
+    {
+      "epoch": 0.7905009759271308,
+      "grad_norm": 0.18583200871944427,
+      "learning_rate": 0.0001,
+      "loss": 1.7999,
+      "step": 1215
+    },
+    {
+      "epoch": 0.7911515940143136,
+      "grad_norm": 0.23056970536708832,
+      "learning_rate": 0.0001,
+      "loss": 2.126,
+      "step": 1216
+    },
+    {
+      "epoch": 0.7918022121014964,
+      "grad_norm": 0.3166569471359253,
+      "learning_rate": 0.0001,
+      "loss": 3.0332,
+      "step": 1217
+    },
+    {
+      "epoch": 0.7924528301886793,
+      "grad_norm": 0.273381769657135,
+      "learning_rate": 0.0001,
+      "loss": 2.2258,
+      "step": 1218
+    },
+    {
+      "epoch": 0.7931034482758621,
+      "grad_norm": 0.3166522979736328,
+      "learning_rate": 0.0001,
+      "loss": 2.35,
+      "step": 1219
+    },
+    {
+      "epoch": 0.7937540663630449,
+      "grad_norm": 0.1906355321407318,
+      "learning_rate": 0.0001,
+      "loss": 1.9739,
+      "step": 1220
+    },
+    {
+      "epoch": 0.7944046844502277,
+      "grad_norm": 0.2339126616716385,
+      "learning_rate": 0.0001,
+      "loss": 2.3575,
+      "step": 1221
+    },
+    {
+      "epoch": 0.7950553025374105,
+      "grad_norm": 0.2760171592235565,
+      "learning_rate": 0.0001,
+      "loss": 2.4708,
+      "step": 1222
+    },
+    {
+      "epoch": 0.7957059206245933,
+      "grad_norm": 0.17487159371376038,
+      "learning_rate": 0.0001,
+      "loss": 1.7924,
+      "step": 1223
+    },
+    {
+      "epoch": 0.7963565387117761,
+      "grad_norm": 0.19386877119541168,
+      "learning_rate": 0.0001,
+      "loss": 2.3044,
+      "step": 1224
+    },
+    {
+      "epoch": 0.7970071567989591,
+      "grad_norm": 0.18056143820285797,
+      "learning_rate": 0.0001,
+      "loss": 1.9543,
+      "step": 1225
+    },
+    {
+      "epoch": 0.7976577748861419,
+      "grad_norm": 0.3085278868675232,
+      "learning_rate": 0.0001,
+      "loss": 2.2131,
+      "step": 1226
+    },
+    {
+      "epoch": 0.7983083929733247,
+      "grad_norm": 0.1960904896259308,
+      "learning_rate": 0.0001,
+      "loss": 2.0918,
+      "step": 1227
+    },
+    {
+      "epoch": 0.7989590110605075,
+      "grad_norm": 0.19437837600708008,
+      "learning_rate": 0.0001,
+      "loss": 2.2241,
+      "step": 1228
+    },
+    {
+      "epoch": 0.7996096291476903,
+      "grad_norm": 0.2129238396883011,
+      "learning_rate": 0.0001,
+      "loss": 2.1891,
+      "step": 1229
+    },
+    {
+      "epoch": 0.8002602472348731,
+      "grad_norm": 0.20101650059223175,
+      "learning_rate": 0.0001,
+      "loss": 2.1341,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8009108653220559,
+      "grad_norm": 0.20897014439105988,
+      "learning_rate": 0.0001,
+      "loss": 2.0937,
+      "step": 1231
+    },
+    {
+      "epoch": 0.8015614834092388,
+      "grad_norm": 0.2693694829940796,
+      "learning_rate": 0.0001,
+      "loss": 2.7406,
+      "step": 1232
+    },
+    {
+      "epoch": 0.8022121014964216,
+      "grad_norm": 0.2322738617658615,
+      "learning_rate": 0.0001,
+      "loss": 2.8483,
+      "step": 1233
+    },
+    {
+      "epoch": 0.8028627195836044,
+      "grad_norm": 0.21177823841571808,
+      "learning_rate": 0.0001,
+      "loss": 2.2315,
+      "step": 1234
+    },
+    {
+      "epoch": 0.8035133376707873,
+      "grad_norm": 0.2920454442501068,
+      "learning_rate": 0.0001,
+      "loss": 3.0264,
+      "step": 1235
+    },
+    {
+      "epoch": 0.8041639557579701,
+      "grad_norm": 0.2331319898366928,
+      "learning_rate": 0.0001,
+      "loss": 2.4574,
+      "step": 1236
+    },
+    {
+      "epoch": 0.8048145738451529,
+      "grad_norm": 0.2339990735054016,
+      "learning_rate": 0.0001,
+      "loss": 2.2752,
+      "step": 1237
+    },
+    {
+      "epoch": 0.8054651919323357,
+      "grad_norm": 0.22823981940746307,
+      "learning_rate": 0.0001,
+      "loss": 1.9615,
+      "step": 1238
+    },
+    {
+      "epoch": 0.8061158100195186,
+      "grad_norm": 0.20435038208961487,
+      "learning_rate": 0.0001,
+      "loss": 1.9989,
+      "step": 1239
+    },
+    {
+      "epoch": 0.8067664281067014,
+      "grad_norm": 0.32488611340522766,
+      "learning_rate": 0.0001,
+      "loss": 2.4791,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8074170461938842,
+      "grad_norm": 0.27227675914764404,
+      "learning_rate": 0.0001,
+      "loss": 2.6443,
+      "step": 1241
+    },
+    {
+      "epoch": 0.808067664281067,
+      "grad_norm": 0.20864960551261902,
+      "learning_rate": 0.0001,
+      "loss": 2.2324,
+      "step": 1242
+    },
+    {
+      "epoch": 0.8087182823682498,
+      "grad_norm": 0.22645455598831177,
+      "learning_rate": 0.0001,
+      "loss": 2.0199,
+      "step": 1243
+    },
+    {
+      "epoch": 0.8093689004554326,
+      "grad_norm": 0.22091244161128998,
+      "learning_rate": 0.0001,
+      "loss": 2.1145,
+      "step": 1244
+    },
+    {
+      "epoch": 0.8100195185426154,
+      "grad_norm": 0.20442111790180206,
+      "learning_rate": 0.0001,
+      "loss": 2.1277,
+      "step": 1245
+    },
+    {
+      "epoch": 0.8106701366297983,
+      "grad_norm": 0.19400720298290253,
+      "learning_rate": 0.0001,
+      "loss": 1.951,
+      "step": 1246
+    },
+    {
+      "epoch": 0.8113207547169812,
+      "grad_norm": 0.474490225315094,
+      "learning_rate": 0.0001,
+      "loss": 3.0206,
+      "step": 1247
+    },
+    {
+      "epoch": 0.811971372804164,
+      "grad_norm": 0.23634073138237,
+      "learning_rate": 0.0001,
+      "loss": 2.2556,
+      "step": 1248
+    },
+    {
+      "epoch": 0.8126219908913468,
+      "grad_norm": 0.23998601734638214,
+      "learning_rate": 0.0001,
+      "loss": 2.3201,
+      "step": 1249
+    },
+    {
+      "epoch": 0.8132726089785296,
+      "grad_norm": 0.19258932769298553,
+      "learning_rate": 0.0001,
+      "loss": 1.9719,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8139232270657124,
+      "grad_norm": 0.21039240062236786,
+      "learning_rate": 0.0001,
+      "loss": 2.3617,
+      "step": 1251
+    },
+    {
+      "epoch": 0.8145738451528952,
+      "grad_norm": 0.37176814675331116,
+      "learning_rate": 0.0001,
+      "loss": 2.7183,
+      "step": 1252
+    },
+    {
+      "epoch": 0.815224463240078,
+      "grad_norm": 0.24739331007003784,
+      "learning_rate": 0.0001,
+      "loss": 2.0098,
+      "step": 1253
+    },
+    {
+      "epoch": 0.8158750813272609,
+      "grad_norm": 0.32313254475593567,
+      "learning_rate": 0.0001,
+      "loss": 2.062,
+      "step": 1254
+    },
+    {
+      "epoch": 0.8165256994144438,
+      "grad_norm": 0.2571156322956085,
+      "learning_rate": 0.0001,
+      "loss": 2.3973,
+      "step": 1255
+    },
+    {
+      "epoch": 0.8171763175016266,
+      "grad_norm": 0.266369491815567,
+      "learning_rate": 0.0001,
+      "loss": 2.6019,
+      "step": 1256
+    },
+    {
+      "epoch": 0.8178269355888094,
+      "grad_norm": 0.3770993649959564,
+      "learning_rate": 0.0001,
+      "loss": 2.5413,
+      "step": 1257
+    },
+    {
+      "epoch": 0.8184775536759922,
+      "grad_norm": 0.24964609742164612,
+      "learning_rate": 0.0001,
+      "loss": 1.7407,
+      "step": 1258
+    },
+    {
+      "epoch": 0.819128171763175,
+      "grad_norm": 0.208835169672966,
+      "learning_rate": 0.0001,
+      "loss": 2.36,
+      "step": 1259
+    },
+    {
+      "epoch": 0.8197787898503578,
+      "grad_norm": 0.19789732992649078,
+      "learning_rate": 0.0001,
+      "loss": 2.0967,
+      "step": 1260
+    },
+    {
+      "epoch": 0.8204294079375407,
+      "grad_norm": 0.4847930669784546,
+      "learning_rate": 0.0001,
+      "loss": 2.9673,
+      "step": 1261
+    },
+    {
+      "epoch": 0.8210800260247235,
+      "grad_norm": 0.277960866689682,
+      "learning_rate": 0.0001,
+      "loss": 2.2165,
+      "step": 1262
+    },
+    {
+      "epoch": 0.8217306441119063,
+      "grad_norm": 0.20278669893741608,
+      "learning_rate": 0.0001,
+      "loss": 2.5098,
+      "step": 1263
+    },
+    {
+      "epoch": 0.8223812621990891,
+      "grad_norm": 0.3295345604419708,
+      "learning_rate": 0.0001,
+      "loss": 2.4451,
+      "step": 1264
+    },
+    {
+      "epoch": 0.8230318802862719,
+      "grad_norm": 0.25482621788978577,
+      "learning_rate": 0.0001,
+      "loss": 2.3178,
+      "step": 1265
+    },
+    {
+      "epoch": 0.8236824983734548,
+      "grad_norm": 0.21955101191997528,
+      "learning_rate": 0.0001,
+      "loss": 2.3245,
+      "step": 1266
+    },
+    {
+      "epoch": 0.8243331164606376,
+      "grad_norm": 0.19811898469924927,
+      "learning_rate": 0.0001,
+      "loss": 2.1608,
+      "step": 1267
+    },
+    {
+      "epoch": 0.8249837345478205,
+      "grad_norm": 0.20357833802700043,
+      "learning_rate": 0.0001,
+      "loss": 2.0502,
+      "step": 1268
+    },
+    {
+      "epoch": 0.8256343526350033,
+      "grad_norm": 0.25111669301986694,
+      "learning_rate": 0.0001,
+      "loss": 2.9059,
+      "step": 1269
+    },
+    {
+      "epoch": 0.8262849707221861,
+      "grad_norm": 0.20970256626605988,
+      "learning_rate": 0.0001,
+      "loss": 2.3496,
+      "step": 1270
+    },
+    {
+      "epoch": 0.8269355888093689,
+      "grad_norm": 0.19146494567394257,
+      "learning_rate": 0.0001,
+      "loss": 2.0773,
+      "step": 1271
+    },
+    {
+      "epoch": 0.8275862068965517,
+      "grad_norm": 0.2083313763141632,
+      "learning_rate": 0.0001,
+      "loss": 2.0031,
+      "step": 1272
+    },
+    {
+      "epoch": 0.8282368249837345,
+      "grad_norm": 0.19460196793079376,
+      "learning_rate": 0.0001,
+      "loss": 2.0411,
+      "step": 1273
+    },
+    {
+      "epoch": 0.8288874430709173,
+      "grad_norm": 0.1900896281003952,
+      "learning_rate": 0.0001,
+      "loss": 1.9517,
+      "step": 1274
+    },
+    {
+      "epoch": 0.8295380611581002,
+      "grad_norm": 0.20020513236522675,
+      "learning_rate": 0.0001,
+      "loss": 2.2062,
+      "step": 1275
+    },
+    {
+      "epoch": 0.8301886792452831,
+      "grad_norm": 0.21990856528282166,
+      "learning_rate": 0.0001,
+      "loss": 2.0837,
+      "step": 1276
+    },
+    {
+      "epoch": 0.8308392973324659,
+      "grad_norm": 0.1966349482536316,
+      "learning_rate": 0.0001,
+      "loss": 2.0407,
+      "step": 1277
+    },
+    {
+      "epoch": 0.8314899154196487,
+      "grad_norm": 0.19897864758968353,
+      "learning_rate": 0.0001,
+      "loss": 2.1639,
+      "step": 1278
+    },
+    {
+      "epoch": 0.8321405335068315,
+      "grad_norm": 0.21094024181365967,
+      "learning_rate": 0.0001,
+      "loss": 2.2158,
+      "step": 1279
+    },
+    {
+      "epoch": 0.8327911515940143,
+      "grad_norm": 0.1989631950855255,
+      "learning_rate": 0.0001,
+      "loss": 1.9578,
+      "step": 1280
+    },
+    {
+      "epoch": 0.8334417696811971,
+      "grad_norm": 0.1953240931034088,
+      "learning_rate": 0.0001,
+      "loss": 2.0365,
+      "step": 1281
+    },
+    {
+      "epoch": 0.8340923877683799,
+      "grad_norm": 0.33914485573768616,
+      "learning_rate": 0.0001,
+      "loss": 2.3676,
+      "step": 1282
+    },
+    {
+      "epoch": 0.8347430058555628,
+      "grad_norm": 0.17135807871818542,
+      "learning_rate": 0.0001,
+      "loss": 1.821,
+      "step": 1283
+    },
+    {
+      "epoch": 0.8353936239427456,
+      "grad_norm": 0.1993912309408188,
+      "learning_rate": 0.0001,
+      "loss": 2.4103,
+      "step": 1284
+    },
+    {
+      "epoch": 0.8360442420299284,
+      "grad_norm": 0.21222157776355743,
+      "learning_rate": 0.0001,
+      "loss": 2.3443,
+      "step": 1285
+    },
+    {
+      "epoch": 0.8366948601171112,
+      "grad_norm": 0.22162573039531708,
+      "learning_rate": 0.0001,
+      "loss": 2.1757,
+      "step": 1286
+    },
+    {
+      "epoch": 0.8373454782042941,
+      "grad_norm": 0.22677986323833466,
+      "learning_rate": 0.0001,
+      "loss": 2.0542,
+      "step": 1287
+    },
+    {
+      "epoch": 0.8379960962914769,
+      "grad_norm": 0.1974060982465744,
+      "learning_rate": 0.0001,
+      "loss": 2.1686,
+      "step": 1288
+    },
+    {
+      "epoch": 0.8386467143786597,
+      "grad_norm": 0.30552592873573303,
+      "learning_rate": 0.0001,
+      "loss": 2.5467,
+      "step": 1289
+    },
+    {
+      "epoch": 0.8392973324658426,
+      "grad_norm": 0.24357165396213531,
+      "learning_rate": 0.0001,
+      "loss": 2.3276,
+      "step": 1290
+    },
+    {
+      "epoch": 0.8399479505530254,
+      "grad_norm": 0.1960456818342209,
+      "learning_rate": 0.0001,
+      "loss": 2.0956,
+      "step": 1291
+    },
+    {
+      "epoch": 0.8405985686402082,
+      "grad_norm": 0.24264569580554962,
+      "learning_rate": 0.0001,
+      "loss": 2.0666,
+      "step": 1292
+    },
+    {
+      "epoch": 0.841249186727391,
+      "grad_norm": 0.25320202112197876,
+      "learning_rate": 0.0001,
+      "loss": 2.033,
+      "step": 1293
+    },
+    {
+      "epoch": 0.8418998048145738,
+      "grad_norm": 0.2313191145658493,
+      "learning_rate": 0.0001,
+      "loss": 2.0571,
+      "step": 1294
+    },
+    {
+      "epoch": 0.8425504229017566,
+      "grad_norm": 0.42846229672431946,
+      "learning_rate": 0.0001,
+      "loss": 2.0875,
+      "step": 1295
+    },
+    {
+      "epoch": 0.8432010409889394,
+      "grad_norm": 0.19277000427246094,
+      "learning_rate": 0.0001,
+      "loss": 1.9303,
+      "step": 1296
+    },
+    {
+      "epoch": 0.8438516590761224,
+      "grad_norm": 0.1947111338376999,
+      "learning_rate": 0.0001,
+      "loss": 1.9482,
+      "step": 1297
+    },
+    {
+      "epoch": 0.8445022771633052,
+      "grad_norm": 0.30196627974510193,
+      "learning_rate": 0.0001,
+      "loss": 2.3238,
+      "step": 1298
+    },
+    {
+      "epoch": 0.845152895250488,
+      "grad_norm": 0.21137486398220062,
+      "learning_rate": 0.0001,
+      "loss": 2.1962,
+      "step": 1299
+    },
+    {
+      "epoch": 0.8458035133376708,
+      "grad_norm": 0.2568284571170807,
+      "learning_rate": 0.0001,
+      "loss": 2.3231,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8464541314248536,
+      "grad_norm": 0.2092464715242386,
+      "learning_rate": 0.0001,
+      "loss": 1.8074,
+      "step": 1301
+    },
+    {
+      "epoch": 0.8471047495120364,
+      "grad_norm": 0.2112191617488861,
+      "learning_rate": 0.0001,
+      "loss": 2.169,
+      "step": 1302
+    },
+    {
+      "epoch": 0.8477553675992192,
+      "grad_norm": 0.17425194382667542,
+      "learning_rate": 0.0001,
+      "loss": 1.8025,
+      "step": 1303
+    },
+    {
+      "epoch": 0.8484059856864021,
+      "grad_norm": 0.20808906853199005,
+      "learning_rate": 0.0001,
+      "loss": 2.0869,
+      "step": 1304
+    },
+    {
+      "epoch": 0.8490566037735849,
+      "grad_norm": 0.25200703740119934,
+      "learning_rate": 0.0001,
+      "loss": 2.4963,
+      "step": 1305
+    },
+    {
+      "epoch": 0.8497072218607677,
+      "grad_norm": 0.23948469758033752,
+      "learning_rate": 0.0001,
+      "loss": 2.3028,
+      "step": 1306
+    },
+    {
+      "epoch": 0.8503578399479506,
+      "grad_norm": 0.185250923037529,
+      "learning_rate": 0.0001,
+      "loss": 1.7409,
+      "step": 1307
+    },
+    {
+      "epoch": 0.8510084580351334,
+      "grad_norm": 0.18948182463645935,
+      "learning_rate": 0.0001,
+      "loss": 1.8922,
+      "step": 1308
+    },
+    {
+      "epoch": 0.8516590761223162,
+      "grad_norm": 0.2027200311422348,
+      "learning_rate": 0.0001,
+      "loss": 2.0922,
+      "step": 1309
+    },
+    {
+      "epoch": 0.852309694209499,
+      "grad_norm": 0.28325602412223816,
+      "learning_rate": 0.0001,
+      "loss": 2.0428,
+      "step": 1310
+    },
+    {
+      "epoch": 0.8529603122966819,
+      "grad_norm": 0.1829916387796402,
+      "learning_rate": 0.0001,
+      "loss": 1.9518,
+      "step": 1311
+    },
+    {
+      "epoch": 0.8536109303838647,
+      "grad_norm": 0.1982378512620926,
+      "learning_rate": 0.0001,
+      "loss": 2.0209,
+      "step": 1312
+    },
+    {
+      "epoch": 0.8542615484710475,
+      "grad_norm": 0.18915079534053802,
+      "learning_rate": 0.0001,
+      "loss": 1.9291,
+      "step": 1313
+    },
+    {
+      "epoch": 0.8549121665582303,
+      "grad_norm": 0.1832190752029419,
+      "learning_rate": 0.0001,
+      "loss": 1.9818,
+      "step": 1314
+    },
+    {
+      "epoch": 0.8555627846454131,
+      "grad_norm": 0.2646237313747406,
+      "learning_rate": 0.0001,
+      "loss": 2.4418,
+      "step": 1315
+    },
+    {
+      "epoch": 0.8562134027325959,
+      "grad_norm": 0.2831929326057434,
+      "learning_rate": 0.0001,
+      "loss": 2.8355,
+      "step": 1316
+    },
+    {
+      "epoch": 0.8568640208197787,
+      "grad_norm": 0.2711881995201111,
+      "learning_rate": 0.0001,
+      "loss": 2.1963,
+      "step": 1317
+    },
+    {
+      "epoch": 0.8575146389069617,
+      "grad_norm": 0.25786513090133667,
+      "learning_rate": 0.0001,
+      "loss": 3.002,
+      "step": 1318
+    },
+    {
+      "epoch": 0.8581652569941445,
+      "grad_norm": 0.26838061213493347,
+      "learning_rate": 0.0001,
+      "loss": 3.1155,
+      "step": 1319
+    },
+    {
+      "epoch": 0.8588158750813273,
+      "grad_norm": 0.2220889776945114,
+      "learning_rate": 0.0001,
+      "loss": 2.0535,
+      "step": 1320
+    },
+    {
+      "epoch": 0.8594664931685101,
+      "grad_norm": 0.2008647471666336,
+      "learning_rate": 0.0001,
+      "loss": 2.0515,
+      "step": 1321
+    },
+    {
+      "epoch": 0.8601171112556929,
+      "grad_norm": 0.22017711400985718,
+      "learning_rate": 0.0001,
+      "loss": 2.289,
+      "step": 1322
+    },
+    {
+      "epoch": 0.8607677293428757,
+      "grad_norm": 0.19674621522426605,
+      "learning_rate": 0.0001,
+      "loss": 1.9414,
+      "step": 1323
+    },
+    {
+      "epoch": 0.8614183474300585,
+      "grad_norm": 0.191552072763443,
+      "learning_rate": 0.0001,
+      "loss": 1.9939,
+      "step": 1324
+    },
+    {
+      "epoch": 0.8620689655172413,
+      "grad_norm": 0.20212143659591675,
+      "learning_rate": 0.0001,
+      "loss": 1.8938,
+      "step": 1325
+    },
+    {
+      "epoch": 0.8627195836044242,
+      "grad_norm": 0.22502020001411438,
+      "learning_rate": 0.0001,
+      "loss": 2.13,
+      "step": 1326
+    },
+    {
+      "epoch": 0.863370201691607,
+      "grad_norm": 0.2504305839538574,
+      "learning_rate": 0.0001,
+      "loss": 2.7666,
+      "step": 1327
+    },
+    {
+      "epoch": 0.8640208197787899,
+      "grad_norm": 0.19481819868087769,
+      "learning_rate": 0.0001,
+      "loss": 2.1141,
+      "step": 1328
+    },
+    {
+      "epoch": 0.8646714378659727,
+      "grad_norm": 0.21994583308696747,
+      "learning_rate": 0.0001,
+      "loss": 2.7615,
+      "step": 1329
+    },
+    {
+      "epoch": 0.8653220559531555,
+      "grad_norm": 0.19281654059886932,
+      "learning_rate": 0.0001,
+      "loss": 2.0864,
+      "step": 1330
+    },
+    {
+      "epoch": 0.8659726740403383,
+      "grad_norm": 0.20329228043556213,
+      "learning_rate": 0.0001,
+      "loss": 2.1002,
+      "step": 1331
+    },
+    {
+      "epoch": 0.8666232921275211,
+      "grad_norm": 0.19484490156173706,
+      "learning_rate": 0.0001,
+      "loss": 2.0519,
+      "step": 1332
+    },
+    {
+      "epoch": 0.867273910214704,
+      "grad_norm": 0.1867295801639557,
+      "learning_rate": 0.0001,
+      "loss": 1.9208,
+      "step": 1333
+    },
+    {
+      "epoch": 0.8679245283018868,
+      "grad_norm": 0.30128392577171326,
+      "learning_rate": 0.0001,
+      "loss": 2.7527,
+      "step": 1334
+    },
+    {
+      "epoch": 0.8685751463890696,
+      "grad_norm": 0.22880543768405914,
+      "learning_rate": 0.0001,
+      "loss": 2.449,
+      "step": 1335
+    },
+    {
+      "epoch": 0.8692257644762524,
+      "grad_norm": 0.23333753645420074,
+      "learning_rate": 0.0001,
+      "loss": 2.0425,
+      "step": 1336
+    },
+    {
+      "epoch": 0.8698763825634352,
+      "grad_norm": 0.34176793694496155,
+      "learning_rate": 0.0001,
+      "loss": 2.8857,
+      "step": 1337
+    },
+    {
+      "epoch": 0.870527000650618,
+      "grad_norm": 0.19983690977096558,
+      "learning_rate": 0.0001,
+      "loss": 2.2466,
+      "step": 1338
+    },
+    {
+      "epoch": 0.8711776187378009,
+      "grad_norm": 0.21883231401443481,
+      "learning_rate": 0.0001,
+      "loss": 2.1262,
+      "step": 1339
+    },
+    {
+      "epoch": 0.8718282368249838,
+      "grad_norm": 0.19143971800804138,
+      "learning_rate": 0.0001,
+      "loss": 2.0119,
+      "step": 1340
+    },
+    {
+      "epoch": 0.8724788549121666,
+      "grad_norm": 0.25845617055892944,
+      "learning_rate": 0.0001,
+      "loss": 2.6315,
+      "step": 1341
+    },
+    {
+      "epoch": 0.8731294729993494,
+      "grad_norm": 0.1914021521806717,
+      "learning_rate": 0.0001,
+      "loss": 1.8571,
+      "step": 1342
+    },
+    {
+      "epoch": 0.8737800910865322,
+      "grad_norm": 0.2742185592651367,
+      "learning_rate": 0.0001,
+      "loss": 2.1467,
+      "step": 1343
+    },
+    {
+      "epoch": 0.874430709173715,
+      "grad_norm": 0.19927754998207092,
+      "learning_rate": 0.0001,
+      "loss": 1.9877,
+      "step": 1344
+    },
+    {
+      "epoch": 0.8750813272608978,
+      "grad_norm": 0.2340778261423111,
+      "learning_rate": 0.0001,
+      "loss": 2.2476,
+      "step": 1345
+    },
+    {
+      "epoch": 0.8757319453480806,
+      "grad_norm": 0.2931828498840332,
+      "learning_rate": 0.0001,
+      "loss": 2.4643,
+      "step": 1346
+    },
+    {
+      "epoch": 0.8763825634352636,
+      "grad_norm": 0.18637506663799286,
+      "learning_rate": 0.0001,
+      "loss": 1.7933,
+      "step": 1347
+    },
+    {
+      "epoch": 0.8770331815224464,
+      "grad_norm": 0.1898747682571411,
+      "learning_rate": 0.0001,
+      "loss": 1.9781,
+      "step": 1348
+    },
+    {
+      "epoch": 0.8776837996096292,
+      "grad_norm": 0.229608952999115,
+      "learning_rate": 0.0001,
+      "loss": 2.1293,
+      "step": 1349
+    },
+    {
+      "epoch": 0.878334417696812,
+      "grad_norm": 0.31374409794807434,
+      "learning_rate": 0.0001,
+      "loss": 2.5436,
+      "step": 1350
+    },
+    {
+      "epoch": 0.8789850357839948,
+      "grad_norm": 0.22544679045677185,
+      "learning_rate": 0.0001,
+      "loss": 2.0882,
+      "step": 1351
+    },
+    {
+      "epoch": 0.8796356538711776,
+      "grad_norm": 0.2415180653333664,
+      "learning_rate": 0.0001,
+      "loss": 2.3193,
+      "step": 1352
+    },
+    {
+      "epoch": 0.8802862719583604,
+      "grad_norm": 0.28355568647384644,
+      "learning_rate": 0.0001,
+      "loss": 2.5994,
+      "step": 1353
+    },
+    {
+      "epoch": 0.8809368900455433,
+      "grad_norm": 0.19143925607204437,
+      "learning_rate": 0.0001,
+      "loss": 2.0546,
+      "step": 1354
+    },
+    {
+      "epoch": 0.8815875081327261,
+      "grad_norm": 0.2990890443325043,
+      "learning_rate": 0.0001,
+      "loss": 2.7388,
+      "step": 1355
+    },
+    {
+      "epoch": 0.8822381262199089,
+      "grad_norm": 0.28672561049461365,
+      "learning_rate": 0.0001,
+      "loss": 1.915,
+      "step": 1356
+    },
+    {
+      "epoch": 0.8828887443070917,
+      "grad_norm": 0.20137082040309906,
+      "learning_rate": 0.0001,
+      "loss": 2.5376,
+      "step": 1357
+    },
+    {
+      "epoch": 0.8835393623942746,
+      "grad_norm": 0.2175220251083374,
+      "learning_rate": 0.0001,
+      "loss": 1.9055,
+      "step": 1358
+    },
+    {
+      "epoch": 0.8841899804814574,
+      "grad_norm": 0.2790168523788452,
+      "learning_rate": 0.0001,
+      "loss": 2.0223,
+      "step": 1359
+    },
+    {
+      "epoch": 0.8848405985686402,
+      "grad_norm": 0.22070975601673126,
+      "learning_rate": 0.0001,
+      "loss": 2.4071,
+      "step": 1360
+    },
+    {
+      "epoch": 0.885491216655823,
+      "grad_norm": 0.22505122423171997,
+      "learning_rate": 0.0001,
+      "loss": 2.2988,
+      "step": 1361
+    },
+    {
+      "epoch": 0.8861418347430059,
+      "grad_norm": 0.2231319099664688,
+      "learning_rate": 0.0001,
+      "loss": 2.0156,
+      "step": 1362
+    },
+    {
+      "epoch": 0.8867924528301887,
+      "grad_norm": 0.2921566665172577,
+      "learning_rate": 0.0001,
+      "loss": 2.7166,
+      "step": 1363
+    },
+    {
+      "epoch": 0.8874430709173715,
+      "grad_norm": 0.19267822802066803,
+      "learning_rate": 0.0001,
+      "loss": 2.0485,
+      "step": 1364
+    },
+    {
+      "epoch": 0.8880936890045543,
+      "grad_norm": 0.28789597749710083,
+      "learning_rate": 0.0001,
+      "loss": 2.7656,
+      "step": 1365
+    },
+    {
+      "epoch": 0.8887443070917371,
+      "grad_norm": 0.3205803334712982,
+      "learning_rate": 0.0001,
+      "loss": 2.5545,
+      "step": 1366
+    },
+    {
+      "epoch": 0.8893949251789199,
+      "grad_norm": 0.20888707041740417,
+      "learning_rate": 0.0001,
+      "loss": 1.8906,
+      "step": 1367
+    },
+    {
+      "epoch": 0.8900455432661027,
+      "grad_norm": 0.18200016021728516,
+      "learning_rate": 0.0001,
+      "loss": 1.8483,
+      "step": 1368
+    },
+    {
+      "epoch": 0.8906961613532857,
+      "grad_norm": 0.2367328256368637,
+      "learning_rate": 0.0001,
+      "loss": 2.3351,
+      "step": 1369
+    },
+    {
+      "epoch": 0.8913467794404685,
+      "grad_norm": 0.28111082315444946,
+      "learning_rate": 0.0001,
+      "loss": 2.5511,
+      "step": 1370
+    },
+    {
+      "epoch": 0.8919973975276513,
+      "grad_norm": 0.19744041562080383,
+      "learning_rate": 0.0001,
+      "loss": 1.9521,
+      "step": 1371
+    },
+    {
+      "epoch": 0.8926480156148341,
+      "grad_norm": 0.2166965901851654,
+      "learning_rate": 0.0001,
+      "loss": 2.1205,
+      "step": 1372
+    },
+    {
+      "epoch": 0.8932986337020169,
+      "grad_norm": 0.20931009948253632,
+      "learning_rate": 0.0001,
+      "loss": 2.1394,
+      "step": 1373
+    },
+    {
+      "epoch": 0.8939492517891997,
+      "grad_norm": 0.2102230191230774,
+      "learning_rate": 0.0001,
+      "loss": 1.9695,
+      "step": 1374
+    },
+    {
+      "epoch": 0.8945998698763825,
+      "grad_norm": 0.22161559760570526,
+      "learning_rate": 0.0001,
+      "loss": 2.4084,
+      "step": 1375
+    },
+    {
+      "epoch": 0.8952504879635654,
+      "grad_norm": 0.22104842960834503,
+      "learning_rate": 0.0001,
+      "loss": 2.6029,
+      "step": 1376
+    },
+    {
+      "epoch": 0.8959011060507482,
+      "grad_norm": 0.2125016152858734,
+      "learning_rate": 0.0001,
+      "loss": 2.0576,
+      "step": 1377
+    },
+    {
+      "epoch": 0.896551724137931,
+      "grad_norm": 0.2626838684082031,
+      "learning_rate": 0.0001,
+      "loss": 2.5907,
+      "step": 1378
+    },
+    {
+      "epoch": 0.8972023422251139,
+      "grad_norm": 0.19114330410957336,
+      "learning_rate": 0.0001,
+      "loss": 2.2824,
+      "step": 1379
+    },
+    {
+      "epoch": 0.8978529603122967,
+      "grad_norm": 0.24731865525245667,
+      "learning_rate": 0.0001,
+      "loss": 2.5292,
+      "step": 1380
+    },
+    {
+      "epoch": 0.8985035783994795,
+      "grad_norm": 0.23787495493888855,
+      "learning_rate": 0.0001,
+      "loss": 2.1433,
+      "step": 1381
+    },
+    {
+      "epoch": 0.8991541964866623,
+      "grad_norm": 0.2028874158859253,
+      "learning_rate": 0.0001,
+      "loss": 2.2726,
+      "step": 1382
+    },
+    {
+      "epoch": 0.8998048145738452,
+      "grad_norm": 0.22940067946910858,
+      "learning_rate": 0.0001,
+      "loss": 2.3222,
+      "step": 1383
+    },
+    {
+      "epoch": 0.900455432661028,
+      "grad_norm": 0.20267997682094574,
+      "learning_rate": 0.0001,
+      "loss": 2.2875,
+      "step": 1384
+    },
+    {
+      "epoch": 0.9011060507482108,
+      "grad_norm": 0.21694517135620117,
+      "learning_rate": 0.0001,
+      "loss": 2.3674,
+      "step": 1385
+    },
+    {
+      "epoch": 0.9017566688353936,
+      "grad_norm": 0.1904231160879135,
+      "learning_rate": 0.0001,
+      "loss": 1.996,
+      "step": 1386
+    },
+    {
+      "epoch": 0.9024072869225764,
+      "grad_norm": 0.2630701959133148,
+      "learning_rate": 0.0001,
+      "loss": 2.4881,
+      "step": 1387
+    },
+    {
+      "epoch": 0.9030579050097592,
+      "grad_norm": 0.19993318617343903,
+      "learning_rate": 0.0001,
+      "loss": 1.9409,
+      "step": 1388
+    },
+    {
+      "epoch": 0.903708523096942,
+      "grad_norm": 0.19389230012893677,
+      "learning_rate": 0.0001,
+      "loss": 2.1121,
+      "step": 1389
+    },
+    {
+      "epoch": 0.904359141184125,
+      "grad_norm": 0.20352298021316528,
+      "learning_rate": 0.0001,
+      "loss": 1.9887,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9050097592713078,
+      "grad_norm": 0.17967310547828674,
+      "learning_rate": 0.0001,
+      "loss": 1.8068,
+      "step": 1391
+    },
+    {
+      "epoch": 0.9056603773584906,
+      "grad_norm": 0.2310938984155655,
+      "learning_rate": 0.0001,
+      "loss": 2.2666,
+      "step": 1392
+    },
+    {
+      "epoch": 0.9063109954456734,
+      "grad_norm": 0.18979041278362274,
+      "learning_rate": 0.0001,
+      "loss": 2.0004,
+      "step": 1393
+    },
+    {
+      "epoch": 0.9069616135328562,
+      "grad_norm": 0.26813068985939026,
+      "learning_rate": 0.0001,
+      "loss": 2.4142,
+      "step": 1394
+    },
+    {
+      "epoch": 0.907612231620039,
+      "grad_norm": 0.23549699783325195,
+      "learning_rate": 0.0001,
+      "loss": 2.3059,
+      "step": 1395
+    },
+    {
+      "epoch": 0.9082628497072218,
+      "grad_norm": 0.2435377985239029,
+      "learning_rate": 0.0001,
+      "loss": 2.1919,
+      "step": 1396
+    },
+    {
+      "epoch": 0.9089134677944047,
+      "grad_norm": 0.21723680198192596,
+      "learning_rate": 0.0001,
+      "loss": 2.2244,
+      "step": 1397
+    },
+    {
+      "epoch": 0.9095640858815875,
+      "grad_norm": 0.20665475726127625,
+      "learning_rate": 0.0001,
+      "loss": 2.1907,
+      "step": 1398
+    },
+    {
+      "epoch": 0.9102147039687704,
+      "grad_norm": 0.26172783970832825,
+      "learning_rate": 0.0001,
+      "loss": 2.5632,
+      "step": 1399
+    },
+    {
+      "epoch": 0.9108653220559532,
+      "grad_norm": 0.22065763175487518,
+      "learning_rate": 0.0001,
+      "loss": 2.287,
+      "step": 1400
+    },
+    {
+      "epoch": 0.911515940143136,
+      "grad_norm": 0.260623574256897,
+      "learning_rate": 0.0001,
+      "loss": 2.7247,
+      "step": 1401
+    },
+    {
+      "epoch": 0.9121665582303188,
+      "grad_norm": 0.1967797726392746,
+      "learning_rate": 0.0001,
+      "loss": 2.3431,
+      "step": 1402
+    },
+    {
+      "epoch": 0.9128171763175016,
+      "grad_norm": 0.19779254496097565,
+      "learning_rate": 0.0001,
+      "loss": 1.8389,
+      "step": 1403
+    },
+    {
+      "epoch": 0.9134677944046844,
+      "grad_norm": 0.20970992743968964,
+      "learning_rate": 0.0001,
+      "loss": 2.1884,
+      "step": 1404
+    },
+    {
+      "epoch": 0.9141184124918673,
+      "grad_norm": 0.22229008376598358,
+      "learning_rate": 0.0001,
+      "loss": 2.2673,
+      "step": 1405
+    },
+    {
+      "epoch": 0.9147690305790501,
+      "grad_norm": 0.2208055853843689,
+      "learning_rate": 0.0001,
+      "loss": 2.1967,
+      "step": 1406
+    },
+    {
+      "epoch": 0.9154196486662329,
+      "grad_norm": 0.2209876924753189,
+      "learning_rate": 0.0001,
+      "loss": 2.2027,
+      "step": 1407
+    },
+    {
+      "epoch": 0.9160702667534157,
+      "grad_norm": 0.19158391654491425,
+      "learning_rate": 0.0001,
+      "loss": 1.9069,
+      "step": 1408
+    },
+    {
+      "epoch": 0.9167208848405985,
+      "grad_norm": 0.2156110256910324,
+      "learning_rate": 0.0001,
+      "loss": 2.2712,
+      "step": 1409
+    },
+    {
+      "epoch": 0.9173715029277814,
+      "grad_norm": 0.2610962390899658,
+      "learning_rate": 0.0001,
+      "loss": 1.8294,
+      "step": 1410
+    },
+    {
+      "epoch": 0.9180221210149642,
+      "grad_norm": 0.18197974562644958,
+      "learning_rate": 0.0001,
+      "loss": 1.9715,
+      "step": 1411
+    },
+    {
+      "epoch": 0.9186727391021471,
+      "grad_norm": 0.19082801043987274,
+      "learning_rate": 0.0001,
+      "loss": 2.1091,
+      "step": 1412
+    },
+    {
+      "epoch": 0.9193233571893299,
+      "grad_norm": 0.26832160353660583,
+      "learning_rate": 0.0001,
+      "loss": 2.7021,
+      "step": 1413
+    },
+    {
+      "epoch": 0.9199739752765127,
+      "grad_norm": 0.3070698082447052,
+      "learning_rate": 0.0001,
+      "loss": 2.4547,
+      "step": 1414
+    },
+    {
+      "epoch": 0.9206245933636955,
+      "grad_norm": 0.25139206647872925,
+      "learning_rate": 0.0001,
+      "loss": 2.5873,
+      "step": 1415
+    },
+    {
+      "epoch": 0.9212752114508783,
+      "grad_norm": 0.2131306529045105,
+      "learning_rate": 0.0001,
+      "loss": 2.3841,
+      "step": 1416
+    },
+    {
+      "epoch": 0.9219258295380611,
+      "grad_norm": 0.24531540274620056,
+      "learning_rate": 0.0001,
+      "loss": 1.9666,
+      "step": 1417
+    },
+    {
+      "epoch": 0.9225764476252439,
+      "grad_norm": 0.1986437737941742,
+      "learning_rate": 0.0001,
+      "loss": 1.9241,
+      "step": 1418
+    },
+    {
+      "epoch": 0.9232270657124269,
+      "grad_norm": 0.23614904284477234,
+      "learning_rate": 0.0001,
+      "loss": 2.5824,
+      "step": 1419
+    },
+    {
+      "epoch": 0.9238776837996097,
+      "grad_norm": 0.2782133221626282,
+      "learning_rate": 0.0001,
+      "loss": 2.1812,
+      "step": 1420
+    },
+    {
+      "epoch": 0.9245283018867925,
+      "grad_norm": 0.2232246845960617,
+      "learning_rate": 0.0001,
+      "loss": 2.3204,
+      "step": 1421
+    },
+    {
+      "epoch": 0.9251789199739753,
+      "grad_norm": 0.22002846002578735,
+      "learning_rate": 0.0001,
+      "loss": 1.8228,
+      "step": 1422
+    },
+    {
+      "epoch": 0.9258295380611581,
+      "grad_norm": 0.30900144577026367,
+      "learning_rate": 0.0001,
+      "loss": 2.4824,
+      "step": 1423
+    },
+    {
+      "epoch": 0.9264801561483409,
+      "grad_norm": 0.262989342212677,
+      "learning_rate": 0.0001,
+      "loss": 2.8719,
+      "step": 1424
+    },
+    {
+      "epoch": 0.9271307742355237,
+      "grad_norm": 0.5406531095504761,
+      "learning_rate": 0.0001,
+      "loss": 2.6984,
+      "step": 1425
+    },
+    {
+      "epoch": 0.9277813923227066,
+      "grad_norm": 0.2415890246629715,
+      "learning_rate": 0.0001,
+      "loss": 2.2543,
+      "step": 1426
+    },
+    {
+      "epoch": 0.9284320104098894,
+      "grad_norm": 0.21261392533779144,
+      "learning_rate": 0.0001,
+      "loss": 1.9761,
+      "step": 1427
+    },
+    {
+      "epoch": 0.9290826284970722,
+      "grad_norm": 0.23223569989204407,
+      "learning_rate": 0.0001,
+      "loss": 1.821,
+      "step": 1428
+    },
+    {
+      "epoch": 0.929733246584255,
+      "grad_norm": 0.2846924960613251,
+      "learning_rate": 0.0001,
+      "loss": 1.9886,
+      "step": 1429
+    },
+    {
+      "epoch": 0.9303838646714379,
+      "grad_norm": 0.2527627646923065,
+      "learning_rate": 0.0001,
+      "loss": 2.373,
+      "step": 1430
+    },
+    {
+      "epoch": 0.9310344827586207,
+      "grad_norm": 0.19917793571949005,
+      "learning_rate": 0.0001,
+      "loss": 2.0111,
+      "step": 1431
+    },
+    {
+      "epoch": 0.9316851008458035,
+      "grad_norm": 0.19021449983119965,
+      "learning_rate": 0.0001,
+      "loss": 2.0373,
+      "step": 1432
+    },
+    {
+      "epoch": 0.9323357189329864,
+      "grad_norm": 0.24929922819137573,
+      "learning_rate": 0.0001,
+      "loss": 2.3885,
+      "step": 1433
+    },
+    {
+      "epoch": 0.9329863370201692,
+      "grad_norm": 0.2533571124076843,
+      "learning_rate": 0.0001,
+      "loss": 2.544,
+      "step": 1434
+    },
+    {
+      "epoch": 0.933636955107352,
+      "grad_norm": 0.23931783437728882,
+      "learning_rate": 0.0001,
+      "loss": 2.418,
+      "step": 1435
+    },
+    {
+      "epoch": 0.9342875731945348,
+      "grad_norm": 0.30167070031166077,
+      "learning_rate": 0.0001,
+      "loss": 2.6513,
+      "step": 1436
+    },
+    {
+      "epoch": 0.9349381912817176,
+      "grad_norm": 0.1971869319677353,
+      "learning_rate": 0.0001,
+      "loss": 2.4016,
+      "step": 1437
+    },
+    {
+      "epoch": 0.9355888093689004,
+      "grad_norm": 0.21331265568733215,
+      "learning_rate": 0.0001,
+      "loss": 2.1524,
+      "step": 1438
+    },
+    {
+      "epoch": 0.9362394274560832,
+      "grad_norm": 0.26298433542251587,
+      "learning_rate": 0.0001,
+      "loss": 2.9442,
+      "step": 1439
+    },
+    {
+      "epoch": 0.936890045543266,
+      "grad_norm": 0.245792955160141,
+      "learning_rate": 0.0001,
+      "loss": 2.2055,
+      "step": 1440
+    },
+    {
+      "epoch": 0.937540663630449,
+      "grad_norm": 0.23703397810459137,
+      "learning_rate": 0.0001,
+      "loss": 2.5616,
+      "step": 1441
+    },
+    {
+      "epoch": 0.9381912817176318,
+      "grad_norm": 0.18641355633735657,
+      "learning_rate": 0.0001,
+      "loss": 1.8982,
+      "step": 1442
+    },
+    {
+      "epoch": 0.9388418998048146,
+      "grad_norm": 0.3551875650882721,
+      "learning_rate": 0.0001,
+      "loss": 2.7802,
+      "step": 1443
+    },
+    {
+      "epoch": 0.9394925178919974,
+      "grad_norm": 0.2278834879398346,
+      "learning_rate": 0.0001,
+      "loss": 2.175,
+      "step": 1444
+    },
+    {
+      "epoch": 0.9401431359791802,
+      "grad_norm": 0.26398956775665283,
+      "learning_rate": 0.0001,
+      "loss": 2.643,
+      "step": 1445
+    },
+    {
+      "epoch": 0.940793754066363,
+      "grad_norm": 0.31316065788269043,
+      "learning_rate": 0.0001,
+      "loss": 2.5662,
+      "step": 1446
+    },
+    {
+      "epoch": 0.9414443721535458,
+      "grad_norm": 0.22769761085510254,
+      "learning_rate": 0.0001,
+      "loss": 2.8677,
+      "step": 1447
+    },
+    {
+      "epoch": 0.9420949902407287,
+      "grad_norm": 0.2069929838180542,
+      "learning_rate": 0.0001,
+      "loss": 2.4393,
+      "step": 1448
+    },
+    {
+      "epoch": 0.9427456083279115,
+      "grad_norm": 0.23500226438045502,
+      "learning_rate": 0.0001,
+      "loss": 2.0914,
+      "step": 1449
+    },
+    {
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.2312425971031189,
+      "learning_rate": 0.0001,
+      "loss": 2.1085,
+      "step": 1450
+    },
+    {
+      "epoch": 0.9440468445022772,
+      "grad_norm": 0.20859290659427643,
+      "learning_rate": 0.0001,
+      "loss": 2.0653,
+      "step": 1451
+    },
+    {
+      "epoch": 0.94469746258946,
+      "grad_norm": 0.23336270451545715,
+      "learning_rate": 0.0001,
+      "loss": 2.1047,
+      "step": 1452
+    },
+    {
+      "epoch": 0.9453480806766428,
+      "grad_norm": 0.2613270580768585,
+      "learning_rate": 0.0001,
+      "loss": 2.3179,
+      "step": 1453
+    },
+    {
+      "epoch": 0.9459986987638256,
+      "grad_norm": 0.2182740867137909,
+      "learning_rate": 0.0001,
+      "loss": 2.0625,
+      "step": 1454
+    },
+    {
+      "epoch": 0.9466493168510085,
+      "grad_norm": 0.28436079621315,
+      "learning_rate": 0.0001,
+      "loss": 1.8766,
+      "step": 1455
+    },
+    {
+      "epoch": 0.9472999349381913,
+      "grad_norm": 0.1998225450515747,
+      "learning_rate": 0.0001,
+      "loss": 2.3157,
+      "step": 1456
+    },
+    {
+      "epoch": 0.9479505530253741,
+      "grad_norm": 0.19695498049259186,
+      "learning_rate": 0.0001,
+      "loss": 1.7501,
+      "step": 1457
+    },
+    {
+      "epoch": 0.9486011711125569,
+      "grad_norm": 0.1972542405128479,
+      "learning_rate": 0.0001,
+      "loss": 1.956,
+      "step": 1458
+    },
+    {
+      "epoch": 0.9492517891997397,
+      "grad_norm": 0.18410329520702362,
+      "learning_rate": 0.0001,
+      "loss": 1.8403,
+      "step": 1459
+    },
+    {
+      "epoch": 0.9499024072869225,
+      "grad_norm": 0.3675645887851715,
+      "learning_rate": 0.0001,
+      "loss": 2.9161,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9505530253741054,
+      "grad_norm": 0.2620394229888916,
+      "learning_rate": 0.0001,
+      "loss": 2.4318,
+      "step": 1461
+    },
+    {
+      "epoch": 0.9512036434612883,
+      "grad_norm": 0.28973767161369324,
+      "learning_rate": 0.0001,
+      "loss": 2.0047,
+      "step": 1462
+    },
+    {
+      "epoch": 0.9518542615484711,
+      "grad_norm": 0.31598249077796936,
+      "learning_rate": 0.0001,
+      "loss": 2.4517,
+      "step": 1463
+    },
+    {
+      "epoch": 0.9525048796356539,
+      "grad_norm": 0.18546514213085175,
+      "learning_rate": 0.0001,
+      "loss": 1.8551,
+      "step": 1464
+    },
+    {
+      "epoch": 0.9531554977228367,
+      "grad_norm": 0.32123416662216187,
+      "learning_rate": 0.0001,
+      "loss": 2.7277,
+      "step": 1465
+    },
+    {
+      "epoch": 0.9538061158100195,
+      "grad_norm": 0.25180497765541077,
+      "learning_rate": 0.0001,
+      "loss": 1.7946,
+      "step": 1466
+    },
+    {
+      "epoch": 0.9544567338972023,
+      "grad_norm": 0.24950966238975525,
+      "learning_rate": 0.0001,
+      "loss": 2.0796,
+      "step": 1467
+    },
+    {
+      "epoch": 0.9551073519843851,
+      "grad_norm": 0.20496372878551483,
+      "learning_rate": 0.0001,
+      "loss": 2.0713,
+      "step": 1468
+    },
+    {
+      "epoch": 0.955757970071568,
+      "grad_norm": 0.20856817066669464,
+      "learning_rate": 0.0001,
+      "loss": 2.1812,
+      "step": 1469
+    },
+    {
+      "epoch": 0.9564085881587508,
+      "grad_norm": 0.26053234934806824,
+      "learning_rate": 0.0001,
+      "loss": 2.3234,
+      "step": 1470
+    },
+    {
+      "epoch": 0.9570592062459337,
+      "grad_norm": 0.3086039125919342,
+      "learning_rate": 0.0001,
+      "loss": 2.3745,
+      "step": 1471
+    },
+    {
+      "epoch": 0.9577098243331165,
+      "grad_norm": 0.19647593796253204,
+      "learning_rate": 0.0001,
+      "loss": 1.8883,
+      "step": 1472
+    },
+    {
+      "epoch": 0.9583604424202993,
+      "grad_norm": 0.20327430963516235,
+      "learning_rate": 0.0001,
+      "loss": 2.125,
+      "step": 1473
+    },
+    {
+      "epoch": 0.9590110605074821,
+      "grad_norm": 0.22550363838672638,
+      "learning_rate": 0.0001,
+      "loss": 2.1609,
+      "step": 1474
+    },
+    {
+      "epoch": 0.9596616785946649,
+      "grad_norm": 0.2369288206100464,
+      "learning_rate": 0.0001,
+      "loss": 1.9352,
+      "step": 1475
+    },
+    {
+      "epoch": 0.9603122966818478,
+      "grad_norm": 0.21195881068706512,
+      "learning_rate": 0.0001,
+      "loss": 2.0275,
+      "step": 1476
+    },
+    {
+      "epoch": 0.9609629147690306,
+      "grad_norm": 0.17060896754264832,
+      "learning_rate": 0.0001,
+      "loss": 1.9566,
+      "step": 1477
+    },
+    {
+      "epoch": 0.9616135328562134,
+      "grad_norm": 0.23335829377174377,
+      "learning_rate": 0.0001,
+      "loss": 2.296,
+      "step": 1478
+    },
+    {
+      "epoch": 0.9622641509433962,
+      "grad_norm": 0.34170275926589966,
+      "learning_rate": 0.0001,
+      "loss": 2.0079,
+      "step": 1479
+    },
+    {
+      "epoch": 0.962914769030579,
+      "grad_norm": 0.2187998741865158,
+      "learning_rate": 0.0001,
+      "loss": 2.0203,
+      "step": 1480
+    },
+    {
+      "epoch": 0.9635653871177619,
+      "grad_norm": 0.1877596378326416,
+      "learning_rate": 0.0001,
+      "loss": 1.9496,
+      "step": 1481
+    },
+    {
+      "epoch": 0.9642160052049447,
+      "grad_norm": 0.18515220284461975,
+      "learning_rate": 0.0001,
+      "loss": 2.0025,
+      "step": 1482
+    },
+    {
+      "epoch": 0.9648666232921275,
+      "grad_norm": 0.21251696348190308,
+      "learning_rate": 0.0001,
+      "loss": 1.8843,
+      "step": 1483
+    },
+    {
+      "epoch": 0.9655172413793104,
+      "grad_norm": 0.19280041754245758,
+      "learning_rate": 0.0001,
+      "loss": 2.0726,
+      "step": 1484
+    },
+    {
+      "epoch": 0.9661678594664932,
+      "grad_norm": 0.1977832317352295,
+      "learning_rate": 0.0001,
+      "loss": 2.0546,
+      "step": 1485
+    },
+    {
+      "epoch": 0.966818477553676,
+      "grad_norm": 0.19019471108913422,
+      "learning_rate": 0.0001,
+      "loss": 1.9825,
+      "step": 1486
+    },
+    {
+      "epoch": 0.9674690956408588,
+      "grad_norm": 0.20381596684455872,
+      "learning_rate": 0.0001,
+      "loss": 2.3339,
+      "step": 1487
+    },
+    {
+      "epoch": 0.9681197137280416,
+      "grad_norm": 0.1899532973766327,
+      "learning_rate": 0.0001,
+      "loss": 2.2962,
+      "step": 1488
+    },
+    {
+      "epoch": 0.9687703318152244,
+      "grad_norm": 0.20524102449417114,
+      "learning_rate": 0.0001,
+      "loss": 2.0874,
+      "step": 1489
+    },
+    {
+      "epoch": 0.9694209499024072,
+      "grad_norm": 0.179798424243927,
+      "learning_rate": 0.0001,
+      "loss": 1.8875,
+      "step": 1490
+    },
+    {
+      "epoch": 0.9700715679895902,
+      "grad_norm": 0.19358840584754944,
+      "learning_rate": 0.0001,
+      "loss": 2.1539,
+      "step": 1491
+    },
+    {
+      "epoch": 0.970722186076773,
+      "grad_norm": 0.2686682343482971,
+      "learning_rate": 0.0001,
+      "loss": 2.3412,
+      "step": 1492
+    },
+    {
+      "epoch": 0.9713728041639558,
+      "grad_norm": 0.2146061509847641,
+      "learning_rate": 0.0001,
+      "loss": 2.476,
+      "step": 1493
+    },
+    {
+      "epoch": 0.9720234222511386,
+      "grad_norm": 0.26737329363822937,
+      "learning_rate": 0.0001,
+      "loss": 2.8003,
+      "step": 1494
+    },
+    {
+      "epoch": 0.9726740403383214,
+      "grad_norm": 0.23344694077968597,
+      "learning_rate": 0.0001,
+      "loss": 2.1174,
+      "step": 1495
+    },
+    {
+      "epoch": 0.9733246584255042,
+      "grad_norm": 0.1991250365972519,
+      "learning_rate": 0.0001,
+      "loss": 2.5734,
+      "step": 1496
+    },
+    {
+      "epoch": 0.973975276512687,
+      "grad_norm": 0.21246576309204102,
+      "learning_rate": 0.0001,
+      "loss": 2.5597,
+      "step": 1497
+    },
+    {
+      "epoch": 0.9746258945998699,
+      "grad_norm": 0.1873084306716919,
+      "learning_rate": 0.0001,
+      "loss": 1.9547,
+      "step": 1498
+    },
+    {
+      "epoch": 0.9752765126870527,
+      "grad_norm": 0.17600129544734955,
+      "learning_rate": 0.0001,
+      "loss": 1.7255,
+      "step": 1499
+    },
+    {
+      "epoch": 0.9759271307742355,
+      "grad_norm": 0.19860287010669708,
+      "learning_rate": 0.0001,
+      "loss": 2.5043,
+      "step": 1500
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1537,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 300,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.262770368118784e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}