diff --git "a/checkpoint-558/trainer_state.json" "b/checkpoint-558/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-558/trainer_state.json"
@@ -0,0 +1,3940 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.5,
+  "eval_steps": 500,
+  "global_step": 558,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002688172043010753,
+      "grad_norm": 1.6433222600981285,
+      "learning_rate": 0.0,
+      "loss": 1.562,
+      "step": 1
+    },
+    {
+      "epoch": 0.005376344086021506,
+      "grad_norm": 1.6862631068558513,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.4706,
+      "step": 2
+    },
+    {
+      "epoch": 0.008064516129032258,
+      "grad_norm": 1.7423201097805276,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.5406,
+      "step": 3
+    },
+    {
+      "epoch": 0.010752688172043012,
+      "grad_norm": 1.7727625055064622,
+      "learning_rate": 3e-06,
+      "loss": 1.5182,
+      "step": 4
+    },
+    {
+      "epoch": 0.013440860215053764,
+      "grad_norm": 1.5457482765192463,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.5169,
+      "step": 5
+    },
+    {
+      "epoch": 0.016129032258064516,
+      "grad_norm": 1.5659007249743502,
+      "learning_rate": 5e-06,
+      "loss": 1.4922,
+      "step": 6
+    },
+    {
+      "epoch": 0.01881720430107527,
+      "grad_norm": 1.3878881126089677,
+      "learning_rate": 6e-06,
+      "loss": 1.4863,
+      "step": 7
+    },
+    {
+      "epoch": 0.021505376344086023,
+      "grad_norm": 1.295368020848385,
+      "learning_rate": 7e-06,
+      "loss": 1.4839,
+      "step": 8
+    },
+    {
+      "epoch": 0.024193548387096774,
+      "grad_norm": 1.589857887668944,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4303,
+      "step": 9
+    },
+    {
+      "epoch": 0.026881720430107527,
+      "grad_norm": 2.60679604894195,
+      "learning_rate": 9e-06,
+      "loss": 1.3744,
+      "step": 10
+    },
+    {
+      "epoch": 0.02956989247311828,
+      "grad_norm": 0.8410885692002656,
+      "learning_rate": 1e-05,
+      "loss": 1.3498,
+      "step": 11
+    },
+    {
+      "epoch": 0.03225806451612903,
+      "grad_norm": 0.7927855266728604,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.3179,
+      "step": 12
+    },
+    {
+      "epoch": 0.03494623655913978,
+      "grad_norm": 0.6808035050220127,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3268,
+      "step": 13
+    },
+    {
+      "epoch": 0.03763440860215054,
+      "grad_norm": 0.6602967909334083,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.2784,
+      "step": 14
+    },
+    {
+      "epoch": 0.04032258064516129,
+      "grad_norm": 0.5797556052811048,
+      "learning_rate": 1.4e-05,
+      "loss": 1.2949,
+      "step": 15
+    },
+    {
+      "epoch": 0.043010752688172046,
+      "grad_norm": 0.6000541560518325,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.288,
+      "step": 16
+    },
+    {
+      "epoch": 0.0456989247311828,
+      "grad_norm": 0.6494981992893607,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2449,
+      "step": 17
+    },
+    {
+      "epoch": 0.04838709677419355,
+      "grad_norm": 0.6723097988215474,
+      "learning_rate": 1.7e-05,
+      "loss": 1.2102,
+      "step": 18
+    },
+    {
+      "epoch": 0.051075268817204304,
+      "grad_norm": 0.6702835925568053,
+      "learning_rate": 1.8e-05,
+      "loss": 1.2025,
+      "step": 19
+    },
+    {
+      "epoch": 0.053763440860215055,
+      "grad_norm": 0.625636082792655,
+      "learning_rate": 1.9e-05,
+      "loss": 1.2777,
+      "step": 20
+    },
+    {
+      "epoch": 0.056451612903225805,
+      "grad_norm": 0.6253912624763358,
+      "learning_rate": 2e-05,
+      "loss": 1.2669,
+      "step": 21
+    },
+    {
+      "epoch": 0.05913978494623656,
+      "grad_norm": 0.5910337660829342,
+      "learning_rate": 2.1000000000000002e-05,
+      "loss": 1.2654,
+      "step": 22
+    },
+    {
+      "epoch": 0.06182795698924731,
+      "grad_norm": 0.6304908028391322,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.2413,
+      "step": 23
+    },
+    {
+      "epoch": 0.06451612903225806,
+      "grad_norm": 0.5377853121890415,
+      "learning_rate": 2.3e-05,
+      "loss": 1.2109,
+      "step": 24
+    },
+    {
+      "epoch": 0.06720430107526881,
+      "grad_norm": 0.4970873703549533,
+      "learning_rate": 2.4e-05,
+      "loss": 1.1359,
+      "step": 25
+    },
+    {
+      "epoch": 0.06989247311827956,
+      "grad_norm": 0.5292734885521813,
+      "learning_rate": 2.5e-05,
+      "loss": 1.2236,
+      "step": 26
+    },
+    {
+      "epoch": 0.07258064516129033,
+      "grad_norm": 0.5428754620149544,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.2083,
+      "step": 27
+    },
+    {
+      "epoch": 0.07526881720430108,
+      "grad_norm": 0.5711123503896314,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.2161,
+      "step": 28
+    },
+    {
+      "epoch": 0.07795698924731183,
+      "grad_norm": 0.49149041488377043,
+      "learning_rate": 2.8e-05,
+      "loss": 1.1454,
+      "step": 29
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 0.5285852530799724,
+      "learning_rate": 2.9e-05,
+      "loss": 1.1194,
+      "step": 30
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "grad_norm": 0.5295555329242986,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 1.1688,
+      "step": 31
+    },
+    {
+      "epoch": 0.08602150537634409,
+      "grad_norm": 0.465354706566009,
+      "learning_rate": 3.1e-05,
+      "loss": 1.1743,
+      "step": 32
+    },
+    {
+      "epoch": 0.08870967741935484,
+      "grad_norm": 0.4486072933924605,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.0818,
+      "step": 33
+    },
+    {
+      "epoch": 0.0913978494623656,
+      "grad_norm": 0.496727888984662,
+      "learning_rate": 3.3e-05,
+      "loss": 1.2101,
+      "step": 34
+    },
+    {
+      "epoch": 0.09408602150537634,
+      "grad_norm": 0.43899748210993167,
+      "learning_rate": 3.4e-05,
+      "loss": 1.1884,
+      "step": 35
+    },
+    {
+      "epoch": 0.0967741935483871,
+      "grad_norm": 0.4147227405541853,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 1.0814,
+      "step": 36
+    },
+    {
+      "epoch": 0.09946236559139784,
+      "grad_norm": 0.48760701758721925,
+      "learning_rate": 3.6e-05,
+      "loss": 1.1212,
+      "step": 37
+    },
+    {
+      "epoch": 0.10215053763440861,
+      "grad_norm": 0.49917378567432974,
+      "learning_rate": 3.7000000000000005e-05,
+      "loss": 1.1984,
+      "step": 38
+    },
+    {
+      "epoch": 0.10483870967741936,
+      "grad_norm": 0.5304015628409972,
+      "learning_rate": 3.8e-05,
+      "loss": 1.1274,
+      "step": 39
+    },
+    {
+      "epoch": 0.10752688172043011,
+      "grad_norm": 0.4726408598975661,
+      "learning_rate": 3.9e-05,
+      "loss": 1.1323,
+      "step": 40
+    },
+    {
+      "epoch": 0.11021505376344086,
+      "grad_norm": 0.44174146995469904,
+      "learning_rate": 4e-05,
+      "loss": 1.1898,
+      "step": 41
+    },
+    {
+      "epoch": 0.11290322580645161,
+      "grad_norm": 0.5087279682773094,
+      "learning_rate": 3.999980086219931e-05,
+      "loss": 1.1469,
+      "step": 42
+    },
+    {
+      "epoch": 0.11559139784946236,
+      "grad_norm": 0.5626510931079601,
+      "learning_rate": 3.999920345276283e-05,
+      "loss": 1.1321,
+      "step": 43
+    },
+    {
+      "epoch": 0.11827956989247312,
+      "grad_norm": 0.47565220090788773,
+      "learning_rate": 3.999820778358724e-05,
+      "loss": 1.1453,
+      "step": 44
+    },
+    {
+      "epoch": 0.12096774193548387,
+      "grad_norm": 0.4431044005508681,
+      "learning_rate": 3.999681387450007e-05,
+      "loss": 1.1408,
+      "step": 45
+    },
+    {
+      "epoch": 0.12365591397849462,
+      "grad_norm": 0.47942624390584926,
+      "learning_rate": 3.999502175325932e-05,
+      "loss": 1.168,
+      "step": 46
+    },
+    {
+      "epoch": 0.12634408602150538,
+      "grad_norm": 0.43166434321061714,
+      "learning_rate": 3.999283145555291e-05,
+      "loss": 1.1087,
+      "step": 47
+    },
+    {
+      "epoch": 0.12903225806451613,
+      "grad_norm": 0.47105749411720044,
+      "learning_rate": 3.999024302499794e-05,
+      "loss": 1.0752,
+      "step": 48
+    },
+    {
+      "epoch": 0.13172043010752688,
+      "grad_norm": 0.3959072081415341,
+      "learning_rate": 3.998725651313984e-05,
+      "loss": 1.1011,
+      "step": 49
+    },
+    {
+      "epoch": 0.13440860215053763,
+      "grad_norm": 0.4416535692834609,
+      "learning_rate": 3.998387197945135e-05,
+      "loss": 1.1306,
+      "step": 50
+    },
+    {
+      "epoch": 0.13709677419354838,
+      "grad_norm": 0.4272647809985287,
+      "learning_rate": 3.9980089491331344e-05,
+      "loss": 1.1381,
+      "step": 51
+    },
+    {
+      "epoch": 0.13978494623655913,
+      "grad_norm": 0.47769854993592265,
+      "learning_rate": 3.997590912410345e-05,
+      "loss": 1.0976,
+      "step": 52
+    },
+    {
+      "epoch": 0.1424731182795699,
+      "grad_norm": 0.3877500456630632,
+      "learning_rate": 3.997133096101458e-05,
+      "loss": 1.128,
+      "step": 53
+    },
+    {
+      "epoch": 0.14516129032258066,
+      "grad_norm": 0.3869721085588235,
+      "learning_rate": 3.996635509323327e-05,
+      "loss": 1.1225,
+      "step": 54
+    },
+    {
+      "epoch": 0.1478494623655914,
+      "grad_norm": 0.47271590281090886,
+      "learning_rate": 3.9960981619847856e-05,
+      "loss": 1.1141,
+      "step": 55
+    },
+    {
+      "epoch": 0.15053763440860216,
+      "grad_norm": 0.4368206211090345,
+      "learning_rate": 3.99552106478645e-05,
+      "loss": 1.0872,
+      "step": 56
+    },
+    {
+      "epoch": 0.1532258064516129,
+      "grad_norm": 0.3872679475185707,
+      "learning_rate": 3.994904229220507e-05,
+      "loss": 1.1514,
+      "step": 57
+    },
+    {
+      "epoch": 0.15591397849462366,
+      "grad_norm": 0.406268890860899,
+      "learning_rate": 3.9942476675704854e-05,
+      "loss": 1.0965,
+      "step": 58
+    },
+    {
+      "epoch": 0.1586021505376344,
+      "grad_norm": 0.43172418498531184,
+      "learning_rate": 3.993551392911009e-05,
+      "loss": 1.1192,
+      "step": 59
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 0.4258357918752704,
+      "learning_rate": 3.9928154191075375e-05,
+      "loss": 1.0623,
+      "step": 60
+    },
+    {
+      "epoch": 0.1639784946236559,
+      "grad_norm": 0.4585556740184179,
+      "learning_rate": 3.9920397608160925e-05,
+      "loss": 1.1076,
+      "step": 61
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 0.44452627464263844,
+      "learning_rate": 3.991224433482961e-05,
+      "loss": 1.1107,
+      "step": 62
+    },
+    {
+      "epoch": 0.1693548387096774,
+      "grad_norm": 0.4787003491624029,
+      "learning_rate": 3.990369453344394e-05,
+      "loss": 1.1165,
+      "step": 63
+    },
+    {
+      "epoch": 0.17204301075268819,
+      "grad_norm": 0.4704549745433953,
+      "learning_rate": 3.989474837426277e-05,
+      "loss": 1.1541,
+      "step": 64
+    },
+    {
+      "epoch": 0.17473118279569894,
+      "grad_norm": 0.4026214434021435,
+      "learning_rate": 3.9885406035437953e-05,
+      "loss": 1.1166,
+      "step": 65
+    },
+    {
+      "epoch": 0.1774193548387097,
+      "grad_norm": 0.40057979364796353,
+      "learning_rate": 3.987566770301076e-05,
+      "loss": 1.0626,
+      "step": 66
+    },
+    {
+      "epoch": 0.18010752688172044,
+      "grad_norm": 0.4340486368362563,
+      "learning_rate": 3.98655335709082e-05,
+      "loss": 1.104,
+      "step": 67
+    },
+    {
+      "epoch": 0.1827956989247312,
+      "grad_norm": 0.42609639195543936,
+      "learning_rate": 3.985500384093917e-05,
+      "loss": 1.0893,
+      "step": 68
+    },
+    {
+      "epoch": 0.18548387096774194,
+      "grad_norm": 0.381378569874383,
+      "learning_rate": 3.984407872279037e-05,
+      "loss": 1.0433,
+      "step": 69
+    },
+    {
+      "epoch": 0.1881720430107527,
+      "grad_norm": 0.3903976348529897,
+      "learning_rate": 3.983275843402222e-05,
+      "loss": 1.1019,
+      "step": 70
+    },
+    {
+      "epoch": 0.19086021505376344,
+      "grad_norm": 0.3648695348221521,
+      "learning_rate": 3.982104320006446e-05,
+      "loss": 1.0992,
+      "step": 71
+    },
+    {
+      "epoch": 0.1935483870967742,
+      "grad_norm": 1.8993059639660952,
+      "learning_rate": 3.9808933254211665e-05,
+      "loss": 1.1056,
+      "step": 72
+    },
+    {
+      "epoch": 0.19623655913978494,
+      "grad_norm": 0.46580843289168206,
+      "learning_rate": 3.979642883761866e-05,
+      "loss": 1.1031,
+      "step": 73
+    },
+    {
+      "epoch": 0.1989247311827957,
+      "grad_norm": 0.449285515287558,
+      "learning_rate": 3.978353019929562e-05,
+      "loss": 1.1068,
+      "step": 74
+    },
+    {
+      "epoch": 0.20161290322580644,
+      "grad_norm": 0.5567418056951845,
+      "learning_rate": 3.977023759610321e-05,
+      "loss": 1.0446,
+      "step": 75
+    },
+    {
+      "epoch": 0.20430107526881722,
+      "grad_norm": 0.38684392317210076,
+      "learning_rate": 3.9756551292747405e-05,
+      "loss": 1.0377,
+      "step": 76
+    },
+    {
+      "epoch": 0.20698924731182797,
+      "grad_norm": 0.473773440244898,
+      "learning_rate": 3.974247156177423e-05,
+      "loss": 1.1396,
+      "step": 77
+    },
+    {
+      "epoch": 0.20967741935483872,
+      "grad_norm": 0.4177520757238314,
+      "learning_rate": 3.9727998683564355e-05,
+      "loss": 1.1008,
+      "step": 78
+    },
+    {
+      "epoch": 0.21236559139784947,
+      "grad_norm": 0.39719194878309766,
+      "learning_rate": 3.9713132946327494e-05,
+      "loss": 1.0215,
+      "step": 79
+    },
+    {
+      "epoch": 0.21505376344086022,
+      "grad_norm": 0.4105085260167095,
+      "learning_rate": 3.9697874646096675e-05,
+      "loss": 1.1115,
+      "step": 80
+    },
+    {
+      "epoch": 0.21774193548387097,
+      "grad_norm": 0.4087045401288919,
+      "learning_rate": 3.968222408672232e-05,
+      "loss": 1.0579,
+      "step": 81
+    },
+    {
+      "epoch": 0.22043010752688172,
+      "grad_norm": 0.39033402258475636,
+      "learning_rate": 3.9666181579866244e-05,
+      "loss": 1.0692,
+      "step": 82
+    },
+    {
+      "epoch": 0.22311827956989247,
+      "grad_norm": 0.41439706526743936,
+      "learning_rate": 3.964974744499539e-05,
+      "loss": 1.0865,
+      "step": 83
+    },
+    {
+      "epoch": 0.22580645161290322,
+      "grad_norm": 0.38234297411695073,
+      "learning_rate": 3.963292200937551e-05,
+      "loss": 1.0173,
+      "step": 84
+    },
+    {
+      "epoch": 0.22849462365591397,
+      "grad_norm": 0.5308750280660687,
+      "learning_rate": 3.961570560806461e-05,
+      "loss": 1.067,
+      "step": 85
+    },
+    {
+      "epoch": 0.23118279569892472,
+      "grad_norm": 0.43351295582441124,
+      "learning_rate": 3.959809858390634e-05,
+      "loss": 1.086,
+      "step": 86
+    },
+    {
+      "epoch": 0.23387096774193547,
+      "grad_norm": 0.42069712201952686,
+      "learning_rate": 3.9580101287523105e-05,
+      "loss": 1.1064,
+      "step": 87
+    },
+    {
+      "epoch": 0.23655913978494625,
+      "grad_norm": 0.42821523209412365,
+      "learning_rate": 3.95617140773091e-05,
+      "loss": 1.0263,
+      "step": 88
+    },
+    {
+      "epoch": 0.239247311827957,
+      "grad_norm": 0.4114502165683399,
+      "learning_rate": 3.954293731942319e-05,
+      "loss": 1.0729,
+      "step": 89
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 0.4131919780645225,
+      "learning_rate": 3.95237713877816e-05,
+      "loss": 1.0621,
+      "step": 90
+    },
+    {
+      "epoch": 0.2446236559139785,
+      "grad_norm": 0.4433939594965718,
+      "learning_rate": 3.950421666405048e-05,
+      "loss": 1.0805,
+      "step": 91
+    },
+    {
+      "epoch": 0.24731182795698925,
+      "grad_norm": 0.4056188018789589,
+      "learning_rate": 3.948427353763829e-05,
+      "loss": 1.0784,
+      "step": 92
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4642044159391645,
+      "learning_rate": 3.946394240568807e-05,
+      "loss": 1.0406,
+      "step": 93
+    },
+    {
+      "epoch": 0.25268817204301075,
+      "grad_norm": 0.4280982724994961,
+      "learning_rate": 3.944322367306951e-05,
+      "loss": 1.1117,
+      "step": 94
+    },
+    {
+      "epoch": 0.2553763440860215,
+      "grad_norm": 0.41758547723414086,
+      "learning_rate": 3.942211775237089e-05,
+      "loss": 1.0747,
+      "step": 95
+    },
+    {
+      "epoch": 0.25806451612903225,
+      "grad_norm": 0.4344009299837567,
+      "learning_rate": 3.940062506389089e-05,
+      "loss": 1.1249,
+      "step": 96
+    },
+    {
+      "epoch": 0.260752688172043,
+      "grad_norm": 0.3847297194838658,
+      "learning_rate": 3.937874603563015e-05,
+      "loss": 1.0977,
+      "step": 97
+    },
+    {
+      "epoch": 0.26344086021505375,
+      "grad_norm": 0.4959083398122344,
+      "learning_rate": 3.935648110328285e-05,
+      "loss": 1.041,
+      "step": 98
+    },
+    {
+      "epoch": 0.2661290322580645,
+      "grad_norm": 0.46262720954521647,
+      "learning_rate": 3.933383071022795e-05,
+      "loss": 1.0926,
+      "step": 99
+    },
+    {
+      "epoch": 0.26881720430107525,
+      "grad_norm": 0.4789561041937064,
+      "learning_rate": 3.93107953075204e-05,
+      "loss": 1.0701,
+      "step": 100
+    },
+    {
+      "epoch": 0.271505376344086,
+      "grad_norm": 0.4229869803365367,
+      "learning_rate": 3.928737535388214e-05,
+      "loss": 1.063,
+      "step": 101
+    },
+    {
+      "epoch": 0.27419354838709675,
+      "grad_norm": 0.43404703473814416,
+      "learning_rate": 3.9263571315692976e-05,
+      "loss": 1.0696,
+      "step": 102
+    },
+    {
+      "epoch": 0.2768817204301075,
+      "grad_norm": 0.4396716028324381,
+      "learning_rate": 3.923938366698129e-05,
+      "loss": 1.0317,
+      "step": 103
+    },
+    {
+      "epoch": 0.27956989247311825,
+      "grad_norm": 0.6860340156482403,
+      "learning_rate": 3.921481288941459e-05,
+      "loss": 1.0611,
+      "step": 104
+    },
+    {
+      "epoch": 0.28225806451612906,
+      "grad_norm": 0.39601683185098385,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 1.0294,
+      "step": 105
+    },
+    {
+      "epoch": 0.2849462365591398,
+      "grad_norm": 0.39641986440862376,
+      "learning_rate": 3.9164523912524224e-05,
+      "loss": 1.0663,
+      "step": 106
+    },
+    {
+      "epoch": 0.28763440860215056,
+      "grad_norm": 0.3898209322812333,
+      "learning_rate": 3.913880671464418e-05,
+      "loss": 1.0671,
+      "step": 107
+    },
+    {
+      "epoch": 0.2903225806451613,
+      "grad_norm": 0.408678962590762,
+      "learning_rate": 3.911270839077644e-05,
+      "loss": 1.0224,
+      "step": 108
+    },
+    {
+      "epoch": 0.29301075268817206,
+      "grad_norm": 0.4681397312637908,
+      "learning_rate": 3.908622946063728e-05,
+      "loss": 1.091,
+      "step": 109
+    },
+    {
+      "epoch": 0.2956989247311828,
+      "grad_norm": 0.47955178042664964,
+      "learning_rate": 3.9059370451522295e-05,
+      "loss": 1.0961,
+      "step": 110
+    },
+    {
+      "epoch": 0.29838709677419356,
+      "grad_norm": 0.4229760577312693,
+      "learning_rate": 3.903213189829589e-05,
+      "loss": 1.0386,
+      "step": 111
+    },
+    {
+      "epoch": 0.3010752688172043,
+      "grad_norm": 0.39011319960684926,
+      "learning_rate": 3.900451434338062e-05,
+      "loss": 1.067,
+      "step": 112
+    },
+    {
+      "epoch": 0.30376344086021506,
+      "grad_norm": 0.39672904488910227,
+      "learning_rate": 3.8976518336746396e-05,
+      "loss": 1.0424,
+      "step": 113
+    },
+    {
+      "epoch": 0.3064516129032258,
+      "grad_norm": 0.49393594827425025,
+      "learning_rate": 3.894814443589954e-05,
+      "loss": 1.0695,
+      "step": 114
+    },
+    {
+      "epoch": 0.30913978494623656,
+      "grad_norm": 0.38254416729289076,
+      "learning_rate": 3.8919393205871676e-05,
+      "loss": 1.0801,
+      "step": 115
+    },
+    {
+      "epoch": 0.3118279569892473,
+      "grad_norm": 0.4456422459103533,
+      "learning_rate": 3.889026521920847e-05,
+      "loss": 1.0934,
+      "step": 116
+    },
+    {
+      "epoch": 0.31451612903225806,
+      "grad_norm": 0.39398196216047476,
+      "learning_rate": 3.886076105595825e-05,
+      "loss": 1.1011,
+      "step": 117
+    },
+    {
+      "epoch": 0.3172043010752688,
+      "grad_norm": 0.3949327527665007,
+      "learning_rate": 3.883088130366042e-05,
+      "loss": 1.018,
+      "step": 118
+    },
+    {
+      "epoch": 0.31989247311827956,
+      "grad_norm": 0.39254792724729387,
+      "learning_rate": 3.88006265573338e-05,
+      "loss": 1.0607,
+      "step": 119
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 0.5007199853312655,
+      "learning_rate": 3.876999741946478e-05,
+      "loss": 1.0609,
+      "step": 120
+    },
+    {
+      "epoch": 0.32526881720430106,
+      "grad_norm": 0.4619751408736227,
+      "learning_rate": 3.873899449999524e-05,
+      "loss": 1.0955,
+      "step": 121
+    },
+    {
+      "epoch": 0.3279569892473118,
+      "grad_norm": 0.48219172224114765,
+      "learning_rate": 3.870761841631051e-05,
+      "loss": 1.063,
+      "step": 122
+    },
+    {
+      "epoch": 0.33064516129032256,
+      "grad_norm": 0.4054037874416271,
+      "learning_rate": 3.867586979322703e-05,
+      "loss": 1.0907,
+      "step": 123
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.43161457507331874,
+      "learning_rate": 3.8643749262979896e-05,
+      "loss": 1.0666,
+      "step": 124
+    },
+    {
+      "epoch": 0.33602150537634407,
+      "grad_norm": 0.36751029685084174,
+      "learning_rate": 3.861125746521028e-05,
+      "loss": 1.0557,
+      "step": 125
+    },
+    {
+      "epoch": 0.3387096774193548,
+      "grad_norm": 0.46690938120869707,
+      "learning_rate": 3.8578395046952686e-05,
+      "loss": 1.1023,
+      "step": 126
+    },
+    {
+      "epoch": 0.34139784946236557,
+      "grad_norm": 0.3988094995343537,
+      "learning_rate": 3.85451626626221e-05,
+      "loss": 1.0717,
+      "step": 127
+    },
+    {
+      "epoch": 0.34408602150537637,
+      "grad_norm": 0.48432619617982536,
+      "learning_rate": 3.85115609740009e-05,
+      "loss": 1.0271,
+      "step": 128
+    },
+    {
+      "epoch": 0.3467741935483871,
+      "grad_norm": 0.5127948499632843,
+      "learning_rate": 3.8477590650225735e-05,
+      "loss": 1.0575,
+      "step": 129
+    },
+    {
+      "epoch": 0.34946236559139787,
+      "grad_norm": 0.4132091412639387,
+      "learning_rate": 3.8443252367774164e-05,
+      "loss": 1.0355,
+      "step": 130
+    },
+    {
+      "epoch": 0.3521505376344086,
+      "grad_norm": 0.4439631972175399,
+      "learning_rate": 3.8408546810451176e-05,
+      "loss": 1.0541,
+      "step": 131
+    },
+    {
+      "epoch": 0.3548387096774194,
+      "grad_norm": 0.3956247259769062,
+      "learning_rate": 3.837347466937562e-05,
+      "loss": 1.0672,
+      "step": 132
+    },
+    {
+      "epoch": 0.3575268817204301,
+      "grad_norm": 0.44952249373265674,
+      "learning_rate": 3.8338036642966396e-05,
+      "loss": 1.0444,
+      "step": 133
+    },
+    {
+      "epoch": 0.3602150537634409,
+      "grad_norm": 0.4449484078947791,
+      "learning_rate": 3.830223343692857e-05,
+      "loss": 1.0514,
+      "step": 134
+    },
+    {
+      "epoch": 0.3629032258064516,
+      "grad_norm": 0.3905509358873801,
+      "learning_rate": 3.826606576423931e-05,
+      "loss": 1.0394,
+      "step": 135
+    },
+    {
+      "epoch": 0.3655913978494624,
+      "grad_norm": 0.4183744146790331,
+      "learning_rate": 3.8229534345133695e-05,
+      "loss": 1.0212,
+      "step": 136
+    },
+    {
+      "epoch": 0.3682795698924731,
+      "grad_norm": 0.46086732418604737,
+      "learning_rate": 3.819263990709037e-05,
+      "loss": 0.994,
+      "step": 137
+    },
+    {
+      "epoch": 0.3709677419354839,
+      "grad_norm": 0.4468564375555911,
+      "learning_rate": 3.8155383184817064e-05,
+      "loss": 1.0279,
+      "step": 138
+    },
+    {
+      "epoch": 0.3736559139784946,
+      "grad_norm": 0.3966511312736679,
+      "learning_rate": 3.8117764920235945e-05,
+      "loss": 0.9992,
+      "step": 139
+    },
+    {
+      "epoch": 0.3763440860215054,
+      "grad_norm": 0.46461846433833476,
+      "learning_rate": 3.807978586246887e-05,
+      "loss": 1.088,
+      "step": 140
+    },
+    {
+      "epoch": 0.3790322580645161,
+      "grad_norm": 0.4254641795470929,
+      "learning_rate": 3.804144676782243e-05,
+      "loss": 1.0764,
+      "step": 141
+    },
+    {
+      "epoch": 0.3817204301075269,
+      "grad_norm": 0.42137203485219293,
+      "learning_rate": 3.800274839977293e-05,
+      "loss": 1.0422,
+      "step": 142
+    },
+    {
+      "epoch": 0.3844086021505376,
+      "grad_norm": 0.4172681789743796,
+      "learning_rate": 3.796369152895117e-05,
+      "loss": 1.0453,
+      "step": 143
+    },
+    {
+      "epoch": 0.3870967741935484,
+      "grad_norm": 0.4531431509751161,
+      "learning_rate": 3.792427693312707e-05,
+      "loss": 1.0389,
+      "step": 144
+    },
+    {
+      "epoch": 0.3897849462365591,
+      "grad_norm": 0.3782466419505299,
+      "learning_rate": 3.788450539719423e-05,
+      "loss": 1.025,
+      "step": 145
+    },
+    {
+      "epoch": 0.3924731182795699,
+      "grad_norm": 0.4655605897605627,
+      "learning_rate": 3.7844377713154264e-05,
+      "loss": 1.064,
+      "step": 146
+    },
+    {
+      "epoch": 0.3951612903225806,
+      "grad_norm": 0.4384836890227208,
+      "learning_rate": 3.780389468010106e-05,
+      "loss": 1.0397,
+      "step": 147
+    },
+    {
+      "epoch": 0.3978494623655914,
+      "grad_norm": 0.4844715439450037,
+      "learning_rate": 3.776305710420482e-05,
+      "loss": 1.1193,
+      "step": 148
+    },
+    {
+      "epoch": 0.40053763440860213,
+      "grad_norm": 0.41760675460607827,
+      "learning_rate": 3.7721865798696056e-05,
+      "loss": 1.0124,
+      "step": 149
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 0.7337537478769387,
+      "learning_rate": 3.7680321583849365e-05,
+      "loss": 1.0508,
+      "step": 150
+    },
+    {
+      "epoch": 0.40591397849462363,
+      "grad_norm": 0.44725816367920673,
+      "learning_rate": 3.76384252869671e-05,
+      "loss": 1.0434,
+      "step": 151
+    },
+    {
+      "epoch": 0.40860215053763443,
+      "grad_norm": 0.40870612635720194,
+      "learning_rate": 3.759617774236292e-05,
+      "loss": 1.068,
+      "step": 152
+    },
+    {
+      "epoch": 0.4112903225806452,
+      "grad_norm": 0.4534649483932217,
+      "learning_rate": 3.755357979134511e-05,
+      "loss": 1.0614,
+      "step": 153
+    },
+    {
+      "epoch": 0.41397849462365593,
+      "grad_norm": 0.41986572053185917,
+      "learning_rate": 3.751063228219993e-05,
+      "loss": 1.0391,
+      "step": 154
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.3717380879536067,
+      "learning_rate": 3.7467336070174604e-05,
+      "loss": 1.0378,
+      "step": 155
+    },
+    {
+      "epoch": 0.41935483870967744,
+      "grad_norm": 0.41848537015206944,
+      "learning_rate": 3.742369201746038e-05,
+      "loss": 1.0439,
+      "step": 156
+    },
+    {
+      "epoch": 0.4220430107526882,
+      "grad_norm": 0.43443932018052933,
+      "learning_rate": 3.737970099317535e-05,
+      "loss": 1.0197,
+      "step": 157
+    },
+    {
+      "epoch": 0.42473118279569894,
+      "grad_norm": 0.421554546653683,
+      "learning_rate": 3.7335363873347056e-05,
+      "loss": 1.0487,
+      "step": 158
+    },
+    {
+      "epoch": 0.4274193548387097,
+      "grad_norm": 0.8430023271255561,
+      "learning_rate": 3.729068154089519e-05,
+      "loss": 1.0333,
+      "step": 159
+    },
+    {
+      "epoch": 0.43010752688172044,
+      "grad_norm": 0.4363044724173691,
+      "learning_rate": 3.724565488561387e-05,
+      "loss": 1.0213,
+      "step": 160
+    },
+    {
+      "epoch": 0.4327956989247312,
+      "grad_norm": 0.5335682969510431,
+      "learning_rate": 3.720028480415401e-05,
+      "loss": 1.0205,
+      "step": 161
+    },
+    {
+      "epoch": 0.43548387096774194,
+      "grad_norm": 0.4056834135687678,
+      "learning_rate": 3.7154572200005446e-05,
+      "loss": 1.0311,
+      "step": 162
+    },
+    {
+      "epoch": 0.4381720430107527,
+      "grad_norm": 0.5322107401886871,
+      "learning_rate": 3.710851798347891e-05,
+      "loss": 1.0601,
+      "step": 163
+    },
+    {
+      "epoch": 0.44086021505376344,
+      "grad_norm": 0.4138677278304246,
+      "learning_rate": 3.7062123071687944e-05,
+      "loss": 1.0361,
+      "step": 164
+    },
+    {
+      "epoch": 0.4435483870967742,
+      "grad_norm": 0.4775100325512625,
+      "learning_rate": 3.701538838853062e-05,
+      "loss": 1.0194,
+      "step": 165
+    },
+    {
+      "epoch": 0.44623655913978494,
+      "grad_norm": 0.40839482534046995,
+      "learning_rate": 3.696831486467114e-05,
+      "loss": 1.0463,
+      "step": 166
+    },
+    {
+      "epoch": 0.4489247311827957,
+      "grad_norm": 0.3963093446633738,
+      "learning_rate": 3.6920903437521305e-05,
+      "loss": 1.0238,
+      "step": 167
+    },
+    {
+      "epoch": 0.45161290322580644,
+      "grad_norm": 0.4344752184390704,
+      "learning_rate": 3.6873155051221846e-05,
+      "loss": 1.0472,
+      "step": 168
+    },
+    {
+      "epoch": 0.4543010752688172,
+      "grad_norm": 0.4167014186949368,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 1.0599,
+      "step": 169
+    },
+    {
+      "epoch": 0.45698924731182794,
+      "grad_norm": 0.43904590007956124,
+      "learning_rate": 3.677665121126871e-05,
+      "loss": 1.0559,
+      "step": 170
+    },
+    {
+      "epoch": 0.4596774193548387,
+      "grad_norm": 0.372185063148541,
+      "learning_rate": 3.6727897679371276e-05,
+      "loss": 1.0012,
+      "step": 171
+    },
+    {
+      "epoch": 0.46236559139784944,
+      "grad_norm": 0.43086731351488916,
+      "learning_rate": 3.667881103179844e-05,
+      "loss": 1.0133,
+      "step": 172
+    },
+    {
+      "epoch": 0.4650537634408602,
+      "grad_norm": 0.5796354347464544,
+      "learning_rate": 3.662939224605091e-05,
+      "loss": 1.0517,
+      "step": 173
+    },
+    {
+      "epoch": 0.46774193548387094,
+      "grad_norm": 0.4587453684541154,
+      "learning_rate": 3.657964230624351e-05,
+      "loss": 1.0164,
+      "step": 174
+    },
+    {
+      "epoch": 0.47043010752688175,
+      "grad_norm": 0.5102852182866393,
+      "learning_rate": 3.6529562203085595e-05,
+      "loss": 1.052,
+      "step": 175
+    },
+    {
+      "epoch": 0.4731182795698925,
+      "grad_norm": 0.4469591346380821,
+      "learning_rate": 3.6479152933861336e-05,
+      "loss": 1.0905,
+      "step": 176
+    },
+    {
+      "epoch": 0.47580645161290325,
+      "grad_norm": 0.45277428352010624,
+      "learning_rate": 3.642841550240983e-05,
+      "loss": 1.0961,
+      "step": 177
+    },
+    {
+      "epoch": 0.478494623655914,
+      "grad_norm": 0.45588595960031525,
+      "learning_rate": 3.6377350919105136e-05,
+      "loss": 1.0178,
+      "step": 178
+    },
+    {
+      "epoch": 0.48118279569892475,
+      "grad_norm": 0.6147997034643559,
+      "learning_rate": 3.632596020083612e-05,
+      "loss": 1.0148,
+      "step": 179
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 0.3734326271789308,
+      "learning_rate": 3.627424437098625e-05,
+      "loss": 1.0006,
+      "step": 180
+    },
+    {
+      "epoch": 0.48655913978494625,
+      "grad_norm": 0.4564187594173089,
+      "learning_rate": 3.6222204459413186e-05,
+      "loss": 1.0635,
+      "step": 181
+    },
+    {
+      "epoch": 0.489247311827957,
+      "grad_norm": 0.42811733614493086,
+      "learning_rate": 3.6169841502428285e-05,
+      "loss": 1.0469,
+      "step": 182
+    },
+    {
+      "epoch": 0.49193548387096775,
+      "grad_norm": 0.4227875509642681,
+      "learning_rate": 3.611715654277596e-05,
+      "loss": 1.0446,
+      "step": 183
+    },
+    {
+      "epoch": 0.4946236559139785,
+      "grad_norm": 0.40548546169007965,
+      "learning_rate": 3.60641506296129e-05,
+      "loss": 1.0564,
+      "step": 184
+    },
+    {
+      "epoch": 0.49731182795698925,
+      "grad_norm": 0.4161116484325749,
+      "learning_rate": 3.601082481848721e-05,
+      "loss": 0.9917,
+      "step": 185
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.39180067540636987,
+      "learning_rate": 3.595718017131736e-05,
+      "loss": 1.0081,
+      "step": 186
+    },
+    {
+      "epoch": 0.5026881720430108,
+      "grad_norm": 0.5307122561583237,
+      "learning_rate": 3.5903217756371066e-05,
+      "loss": 0.9972,
+      "step": 187
+    },
+    {
+      "epoch": 0.5053763440860215,
+      "grad_norm": 0.4633315164676552,
+      "learning_rate": 3.5848938648243976e-05,
+      "loss": 1.0196,
+      "step": 188
+    },
+    {
+      "epoch": 0.5080645161290323,
+      "grad_norm": 0.43457272116367207,
+      "learning_rate": 3.579434392783832e-05,
+      "loss": 1.0429,
+      "step": 189
+    },
+    {
+      "epoch": 0.510752688172043,
+      "grad_norm": 0.42602042879132207,
+      "learning_rate": 3.5739434682341355e-05,
+      "loss": 1.0355,
+      "step": 190
+    },
+    {
+      "epoch": 0.5134408602150538,
+      "grad_norm": 0.37328410492227004,
+      "learning_rate": 3.568421200520371e-05,
+      "loss": 1.0158,
+      "step": 191
+    },
+    {
+      "epoch": 0.5161290322580645,
+      "grad_norm": 0.47901349260363574,
+      "learning_rate": 3.562867699611764e-05,
+      "loss": 1.006,
+      "step": 192
+    },
+    {
+      "epoch": 0.5188172043010753,
+      "grad_norm": 0.6800894155552869,
+      "learning_rate": 3.55728307609951e-05,
+      "loss": 1.0819,
+      "step": 193
+    },
+    {
+      "epoch": 0.521505376344086,
+      "grad_norm": 0.6815573295093794,
+      "learning_rate": 3.5516674411945747e-05,
+      "loss": 0.9767,
+      "step": 194
+    },
+    {
+      "epoch": 0.5241935483870968,
+      "grad_norm": 0.40923877696875666,
+      "learning_rate": 3.546020906725474e-05,
+      "loss": 1.0048,
+      "step": 195
+    },
+    {
+      "epoch": 0.5268817204301075,
+      "grad_norm": 0.39166638466881304,
+      "learning_rate": 3.540343585136056e-05,
+      "loss": 1.0115,
+      "step": 196
+    },
+    {
+      "epoch": 0.5295698924731183,
+      "grad_norm": 0.46039879078749524,
+      "learning_rate": 3.5346355894832515e-05,
+      "loss": 1.0274,
+      "step": 197
+    },
+    {
+      "epoch": 0.532258064516129,
+      "grad_norm": 0.435003701062386,
+      "learning_rate": 3.5288970334348324e-05,
+      "loss": 1.0262,
+      "step": 198
+    },
+    {
+      "epoch": 0.5349462365591398,
+      "grad_norm": 0.46422099557675184,
+      "learning_rate": 3.5231280312671426e-05,
+      "loss": 1.0406,
+      "step": 199
+    },
+    {
+      "epoch": 0.5376344086021505,
+      "grad_norm": 0.3946242892533647,
+      "learning_rate": 3.51732869786282e-05,
+      "loss": 1.0351,
+      "step": 200
+    },
+    {
+      "epoch": 0.5403225806451613,
+      "grad_norm": 0.4593963303455073,
+      "learning_rate": 3.511499148708517e-05,
+      "loss": 1.0161,
+      "step": 201
+    },
+    {
+      "epoch": 0.543010752688172,
+      "grad_norm": 0.43211273427185715,
+      "learning_rate": 3.505639499892591e-05,
+      "loss": 1.0339,
+      "step": 202
+    },
+    {
+      "epoch": 0.5456989247311828,
+      "grad_norm": 0.4638011311631454,
+      "learning_rate": 3.499749868102802e-05,
+      "loss": 1.0195,
+      "step": 203
+    },
+    {
+      "epoch": 0.5483870967741935,
+      "grad_norm": 0.4606785516075864,
+      "learning_rate": 3.4938303706239814e-05,
+      "loss": 1.0809,
+      "step": 204
+    },
+    {
+      "epoch": 0.5510752688172043,
+      "grad_norm": 0.4750835163830621,
+      "learning_rate": 3.487881125335699e-05,
+      "loss": 1.0104,
+      "step": 205
+    },
+    {
+      "epoch": 0.553763440860215,
+      "grad_norm": 0.48069623342657913,
+      "learning_rate": 3.4819022507099184e-05,
+      "loss": 1.0534,
+      "step": 206
+    },
+    {
+      "epoch": 0.5564516129032258,
+      "grad_norm": 0.4485052357605267,
+      "learning_rate": 3.475893865808633e-05,
+      "loss": 1.008,
+      "step": 207
+    },
+    {
+      "epoch": 0.5591397849462365,
+      "grad_norm": 0.45226568470539963,
+      "learning_rate": 3.4698560902815e-05,
+      "loss": 0.9859,
+      "step": 208
+    },
+    {
+      "epoch": 0.5618279569892473,
+      "grad_norm": 0.4556713744237398,
+      "learning_rate": 3.463789044363451e-05,
+      "loss": 1.0468,
+      "step": 209
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 0.40515419542450315,
+      "learning_rate": 3.4576928488723056e-05,
+      "loss": 1.0069,
+      "step": 210
+    },
+    {
+      "epoch": 0.5672043010752689,
+      "grad_norm": 0.407850239298829,
+      "learning_rate": 3.4515676252063595e-05,
+      "loss": 1.024,
+      "step": 211
+    },
+    {
+      "epoch": 0.5698924731182796,
+      "grad_norm": 0.4245125668059516,
+      "learning_rate": 3.445413495341971e-05,
+      "loss": 0.9842,
+      "step": 212
+    },
+    {
+      "epoch": 0.5725806451612904,
+      "grad_norm": 0.5282266357639802,
+      "learning_rate": 3.439230581831126e-05,
+      "loss": 1.0511,
+      "step": 213
+    },
+    {
+      "epoch": 0.5752688172043011,
+      "grad_norm": 0.46721556238008377,
+      "learning_rate": 3.433019007799007e-05,
+      "loss": 1.0722,
+      "step": 214
+    },
+    {
+      "epoch": 0.5779569892473119,
+      "grad_norm": 0.3998174935596331,
+      "learning_rate": 3.4267788969415315e-05,
+      "loss": 1.0417,
+      "step": 215
+    },
+    {
+      "epoch": 0.5806451612903226,
+      "grad_norm": 0.39836497217157424,
+      "learning_rate": 3.420510373522896e-05,
+      "loss": 0.9522,
+      "step": 216
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "grad_norm": 0.5604060165845736,
+      "learning_rate": 3.4142135623730954e-05,
+      "loss": 1.0406,
+      "step": 217
+    },
+    {
+      "epoch": 0.5860215053763441,
+      "grad_norm": 0.4626752931850209,
+      "learning_rate": 3.4078885888854436e-05,
+      "loss": 1.0403,
+      "step": 218
+    },
+    {
+      "epoch": 0.5887096774193549,
+      "grad_norm": 0.4119865874583256,
+      "learning_rate": 3.4015355790140715e-05,
+      "loss": 0.974,
+      "step": 219
+    },
+    {
+      "epoch": 0.5913978494623656,
+      "grad_norm": 0.41688760669607,
+      "learning_rate": 3.39515465927142e-05,
+      "loss": 1.0354,
+      "step": 220
+    },
+    {
+      "epoch": 0.5940860215053764,
+      "grad_norm": 0.47263736408876167,
+      "learning_rate": 3.388745956725722e-05,
+      "loss": 1.0438,
+      "step": 221
+    },
+    {
+      "epoch": 0.5967741935483871,
+      "grad_norm": 0.48712838990373963,
+      "learning_rate": 3.3823095989984697e-05,
+      "loss": 0.9847,
+      "step": 222
+    },
+    {
+      "epoch": 0.5994623655913979,
+      "grad_norm": 0.39317905049275836,
+      "learning_rate": 3.3758457142618754e-05,
+      "loss": 0.9806,
+      "step": 223
+    },
+    {
+      "epoch": 0.6021505376344086,
+      "grad_norm": 0.484001386994586,
+      "learning_rate": 3.369354431236319e-05,
+      "loss": 1.0003,
+      "step": 224
+    },
+    {
+      "epoch": 0.6048387096774194,
+      "grad_norm": 0.3896751020684252,
+      "learning_rate": 3.362835879187783e-05,
+      "loss": 0.9314,
+      "step": 225
+    },
+    {
+      "epoch": 0.6075268817204301,
+      "grad_norm": 0.402131340210077,
+      "learning_rate": 3.356290187925278e-05,
+      "loss": 0.957,
+      "step": 226
+    },
+    {
+      "epoch": 0.6102150537634409,
+      "grad_norm": 0.4442069284277535,
+      "learning_rate": 3.349717487798261e-05,
+      "loss": 1.0651,
+      "step": 227
+    },
+    {
+      "epoch": 0.6129032258064516,
+      "grad_norm": 0.4075067959077034,
+      "learning_rate": 3.3431179096940375e-05,
+      "loss": 1.0117,
+      "step": 228
+    },
+    {
+      "epoch": 0.6155913978494624,
+      "grad_norm": 0.4595977891340027,
+      "learning_rate": 3.3364915850351525e-05,
+      "loss": 1.0277,
+      "step": 229
+    },
+    {
+      "epoch": 0.6182795698924731,
+      "grad_norm": 0.41565240224286376,
+      "learning_rate": 3.3298386457767804e-05,
+      "loss": 0.9873,
+      "step": 230
+    },
+    {
+      "epoch": 0.6209677419354839,
+      "grad_norm": 0.400290934516727,
+      "learning_rate": 3.3231592244040885e-05,
+      "loss": 1.0503,
+      "step": 231
+    },
+    {
+      "epoch": 0.6236559139784946,
+      "grad_norm": 0.43593503744528256,
+      "learning_rate": 3.3164534539296056e-05,
+      "loss": 1.0256,
+      "step": 232
+    },
+    {
+      "epoch": 0.6263440860215054,
+      "grad_norm": 0.4297576409774745,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.9873,
+      "step": 233
+    },
+    {
+      "epoch": 0.6290322580645161,
+      "grad_norm": 0.5286155107560961,
+      "learning_rate": 3.302963400346272e-05,
+      "loss": 1.0526,
+      "step": 234
+    },
+    {
+      "epoch": 0.6317204301075269,
+      "grad_norm": 0.4080215430723157,
+      "learning_rate": 3.296179385875381e-05,
+      "loss": 0.993,
+      "step": 235
+    },
+    {
+      "epoch": 0.6344086021505376,
+      "grad_norm": 0.4666697414536282,
+      "learning_rate": 3.2893695595732705e-05,
+      "loss": 0.9855,
+      "step": 236
+    },
+    {
+      "epoch": 0.6370967741935484,
+      "grad_norm": 0.44576593027115785,
+      "learning_rate": 3.282534057049322e-05,
+      "loss": 0.994,
+      "step": 237
+    },
+    {
+      "epoch": 0.6397849462365591,
+      "grad_norm": 0.45875921319019286,
+      "learning_rate": 3.275673014424231e-05,
+      "loss": 1.0695,
+      "step": 238
+    },
+    {
+      "epoch": 0.6424731182795699,
+      "grad_norm": 0.4483391985101821,
+      "learning_rate": 3.268786568327291e-05,
+      "loss": 1.0413,
+      "step": 239
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 0.3823024947210084,
+      "learning_rate": 3.261874855893675e-05,
+      "loss": 1.0634,
+      "step": 240
+    },
+    {
+      "epoch": 0.6478494623655914,
+      "grad_norm": 0.42590418591004187,
+      "learning_rate": 3.254938014761704e-05,
+      "loss": 1.1039,
+      "step": 241
+    },
+    {
+      "epoch": 0.6505376344086021,
+      "grad_norm": 0.4436207874701427,
+      "learning_rate": 3.2479761830701075e-05,
+      "loss": 1.0797,
+      "step": 242
+    },
+    {
+      "epoch": 0.6532258064516129,
+      "grad_norm": 0.5436242022516592,
+      "learning_rate": 3.240989499455269e-05,
+      "loss": 0.998,
+      "step": 243
+    },
+    {
+      "epoch": 0.6559139784946236,
+      "grad_norm": 0.42461660808494955,
+      "learning_rate": 3.2339781030484715e-05,
+      "loss": 1.0014,
+      "step": 244
+    },
+    {
+      "epoch": 0.6586021505376344,
+      "grad_norm": 0.4147658974390641,
+      "learning_rate": 3.2269421334731196e-05,
+      "loss": 1.0047,
+      "step": 245
+    },
+    {
+      "epoch": 0.6612903225806451,
+      "grad_norm": 0.3702000902999608,
+      "learning_rate": 3.219881730841964e-05,
+      "loss": 1.0057,
+      "step": 246
+    },
+    {
+      "epoch": 0.6639784946236559,
+      "grad_norm": 0.37405944820555137,
+      "learning_rate": 3.212797035754311e-05,
+      "loss": 0.9881,
+      "step": 247
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.39789221907192235,
+      "learning_rate": 3.205688189293219e-05,
+      "loss": 1.002,
+      "step": 248
+    },
+    {
+      "epoch": 0.6693548387096774,
+      "grad_norm": 0.35269099760384387,
+      "learning_rate": 3.198555333022694e-05,
+      "loss": 1.0445,
+      "step": 249
+    },
+    {
+      "epoch": 0.6720430107526881,
+      "grad_norm": 0.39171670743365294,
+      "learning_rate": 3.191398608984867e-05,
+      "loss": 0.9873,
+      "step": 250
+    },
+    {
+      "epoch": 0.6747311827956989,
+      "grad_norm": 0.36377972714827284,
+      "learning_rate": 3.184218159697166e-05,
+      "loss": 0.9678,
+      "step": 251
+    },
+    {
+      "epoch": 0.6774193548387096,
+      "grad_norm": 0.4760701686418637,
+      "learning_rate": 3.177014128149479e-05,
+      "loss": 1.0475,
+      "step": 252
+    },
+    {
+      "epoch": 0.6801075268817204,
+      "grad_norm": 0.36306748600915323,
+      "learning_rate": 3.169786657801306e-05,
+      "loss": 0.9737,
+      "step": 253
+    },
+    {
+      "epoch": 0.6827956989247311,
+      "grad_norm": 0.36397370143939106,
+      "learning_rate": 3.162535892578903e-05,
+      "loss": 1.0009,
+      "step": 254
+    },
+    {
+      "epoch": 0.6854838709677419,
+      "grad_norm": 0.41923544253489314,
+      "learning_rate": 3.155261976872412e-05,
+      "loss": 0.9855,
+      "step": 255
+    },
+    {
+      "epoch": 0.6881720430107527,
+      "grad_norm": 0.4349008134787599,
+      "learning_rate": 3.147965055532991e-05,
+      "loss": 0.9843,
+      "step": 256
+    },
+    {
+      "epoch": 0.6908602150537635,
+      "grad_norm": 0.4403161475473632,
+      "learning_rate": 3.1406452738699284e-05,
+      "loss": 0.9932,
+      "step": 257
+    },
+    {
+      "epoch": 0.6935483870967742,
+      "grad_norm": 0.4088632034626185,
+      "learning_rate": 3.1333027776477454e-05,
+      "loss": 1.0175,
+      "step": 258
+    },
+    {
+      "epoch": 0.696236559139785,
+      "grad_norm": 0.4089626667866183,
+      "learning_rate": 3.125937713083296e-05,
+      "loss": 0.9957,
+      "step": 259
+    },
+    {
+      "epoch": 0.6989247311827957,
+      "grad_norm": 0.44005061948101687,
+      "learning_rate": 3.118550226842857e-05,
+      "loss": 0.9902,
+      "step": 260
+    },
+    {
+      "epoch": 0.7016129032258065,
+      "grad_norm": 1.1016022022748841,
+      "learning_rate": 3.111140466039205e-05,
+      "loss": 0.991,
+      "step": 261
+    },
+    {
+      "epoch": 0.7043010752688172,
+      "grad_norm": 0.39448956783294353,
+      "learning_rate": 3.103708578228686e-05,
+      "loss": 1.0041,
+      "step": 262
+    },
+    {
+      "epoch": 0.706989247311828,
+      "grad_norm": 0.41388488702273174,
+      "learning_rate": 3.0962547114082804e-05,
+      "loss": 0.9928,
+      "step": 263
+    },
+    {
+      "epoch": 0.7096774193548387,
+      "grad_norm": 0.4065224464102798,
+      "learning_rate": 3.088779014012652e-05,
+      "loss": 0.9859,
+      "step": 264
+    },
+    {
+      "epoch": 0.7123655913978495,
+      "grad_norm": 0.39952347811781436,
+      "learning_rate": 3.0812816349111956e-05,
+      "loss": 0.9613,
+      "step": 265
+    },
+    {
+      "epoch": 0.7150537634408602,
+      "grad_norm": 0.43554876713734897,
+      "learning_rate": 3.073762723405069e-05,
+      "loss": 1.0289,
+      "step": 266
+    },
+    {
+      "epoch": 0.717741935483871,
+      "grad_norm": 0.469813057633801,
+      "learning_rate": 3.066222429224221e-05,
+      "loss": 1.0438,
+      "step": 267
+    },
+    {
+      "epoch": 0.7204301075268817,
+      "grad_norm": 0.4353123605440106,
+      "learning_rate": 3.0586609025244144e-05,
+      "loss": 1.0017,
+      "step": 268
+    },
+    {
+      "epoch": 0.7231182795698925,
+      "grad_norm": 0.40010712539262144,
+      "learning_rate": 3.051078293884226e-05,
+      "loss": 1.0254,
+      "step": 269
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 0.41179768187019394,
+      "learning_rate": 3.0434747543020585e-05,
+      "loss": 1.0167,
+      "step": 270
+    },
+    {
+      "epoch": 0.728494623655914,
+      "grad_norm": 0.39261397155250993,
+      "learning_rate": 3.0358504351931265e-05,
+      "loss": 0.9987,
+      "step": 271
+    },
+    {
+      "epoch": 0.7311827956989247,
+      "grad_norm": 0.4037853365263608,
+      "learning_rate": 3.0282054883864434e-05,
+      "loss": 1.0016,
+      "step": 272
+    },
+    {
+      "epoch": 0.7338709677419355,
+      "grad_norm": 0.3920371074761728,
+      "learning_rate": 3.0205400661218e-05,
+      "loss": 0.9427,
+      "step": 273
+    },
+    {
+      "epoch": 0.7365591397849462,
+      "grad_norm": 0.4525036893342772,
+      "learning_rate": 3.0128543210467273e-05,
+      "loss": 1.0566,
+      "step": 274
+    },
+    {
+      "epoch": 0.739247311827957,
+      "grad_norm": 0.41264407607647574,
+      "learning_rate": 3.0051484062134632e-05,
+      "loss": 0.9899,
+      "step": 275
+    },
+    {
+      "epoch": 0.7419354838709677,
+      "grad_norm": 0.37437706613357397,
+      "learning_rate": 2.9974224750759017e-05,
+      "loss": 0.9817,
+      "step": 276
+    },
+    {
+      "epoch": 0.7446236559139785,
+      "grad_norm": 0.3844600838817203,
+      "learning_rate": 2.9896766814865355e-05,
+      "loss": 1.0263,
+      "step": 277
+    },
+    {
+      "epoch": 0.7473118279569892,
+      "grad_norm": 0.4310511049000039,
+      "learning_rate": 2.9819111796933948e-05,
+      "loss": 0.9781,
+      "step": 278
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.40281595760365946,
+      "learning_rate": 2.9741261243369746e-05,
+      "loss": 1.0273,
+      "step": 279
+    },
+    {
+      "epoch": 0.7526881720430108,
+      "grad_norm": 0.4498302856339957,
+      "learning_rate": 2.9663216704471547e-05,
+      "loss": 0.9886,
+      "step": 280
+    },
+    {
+      "epoch": 0.7553763440860215,
+      "grad_norm": 0.4350406167421517,
+      "learning_rate": 2.958497973440114e-05,
+      "loss": 1.0247,
+      "step": 281
+    },
+    {
+      "epoch": 0.7580645161290323,
+      "grad_norm": 0.46748351737565624,
+      "learning_rate": 2.9506551891152334e-05,
+      "loss": 1.0072,
+      "step": 282
+    },
+    {
+      "epoch": 0.760752688172043,
+      "grad_norm": 0.3998308958015181,
+      "learning_rate": 2.9427934736519962e-05,
+      "loss": 1.076,
+      "step": 283
+    },
+    {
+      "epoch": 0.7634408602150538,
+      "grad_norm": 0.42326867383664013,
+      "learning_rate": 2.9349129836068732e-05,
+      "loss": 0.9895,
+      "step": 284
+    },
+    {
+      "epoch": 0.7661290322580645,
+      "grad_norm": 0.3949205497118407,
+      "learning_rate": 2.9270138759102108e-05,
+      "loss": 1.027,
+      "step": 285
+    },
+    {
+      "epoch": 0.7688172043010753,
+      "grad_norm": 0.40826149975955933,
+      "learning_rate": 2.919096307863104e-05,
+      "loss": 1.0128,
+      "step": 286
+    },
+    {
+      "epoch": 0.771505376344086,
+      "grad_norm": 0.6045575439891937,
+      "learning_rate": 2.9111604371342593e-05,
+      "loss": 0.9806,
+      "step": 287
+    },
+    {
+      "epoch": 0.7741935483870968,
+      "grad_norm": 0.3906743864943639,
+      "learning_rate": 2.903206421756862e-05,
+      "loss": 1.0126,
+      "step": 288
+    },
+    {
+      "epoch": 0.7768817204301075,
+      "grad_norm": 0.37994713789537804,
+      "learning_rate": 2.8952344201254253e-05,
+      "loss": 0.9984,
+      "step": 289
+    },
+    {
+      "epoch": 0.7795698924731183,
+      "grad_norm": 0.4560671009564336,
+      "learning_rate": 2.8872445909926358e-05,
+      "loss": 0.9846,
+      "step": 290
+    },
+    {
+      "epoch": 0.782258064516129,
+      "grad_norm": 0.40231158085064994,
+      "learning_rate": 2.8792370934661948e-05,
+      "loss": 1.0403,
+      "step": 291
+    },
+    {
+      "epoch": 0.7849462365591398,
+      "grad_norm": 0.4776678536973747,
+      "learning_rate": 2.8712120870056455e-05,
+      "loss": 1.0327,
+      "step": 292
+    },
+    {
+      "epoch": 0.7876344086021505,
+      "grad_norm": 0.45302618010000684,
+      "learning_rate": 2.8631697314192012e-05,
+      "loss": 1.0126,
+      "step": 293
+    },
+    {
+      "epoch": 0.7903225806451613,
+      "grad_norm": 0.4332121059542856,
+      "learning_rate": 2.8551101868605644e-05,
+      "loss": 1.0475,
+      "step": 294
+    },
+    {
+      "epoch": 0.793010752688172,
+      "grad_norm": 0.4498441085262953,
+      "learning_rate": 2.8470336138257315e-05,
+      "loss": 1.0178,
+      "step": 295
+    },
+    {
+      "epoch": 0.7956989247311828,
+      "grad_norm": 0.39208633969875073,
+      "learning_rate": 2.8389401731498018e-05,
+      "loss": 1.0127,
+      "step": 296
+    },
+    {
+      "epoch": 0.7983870967741935,
+      "grad_norm": 0.4042053763726035,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.9732,
+      "step": 297
+    },
+    {
+      "epoch": 0.8010752688172043,
+      "grad_norm": 0.42842239164240437,
+      "learning_rate": 2.8227033338913318e-05,
+      "loss": 1.0152,
+      "step": 298
+    },
+    {
+      "epoch": 0.803763440860215,
+      "grad_norm": 0.3807866452863404,
+      "learning_rate": 2.814560258645638e-05,
+      "loss": 1.0189,
+      "step": 299
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 0.43852909963759557,
+      "learning_rate": 2.8064009624260994e-05,
+      "loss": 1.0084,
+      "step": 300
+    },
+    {
+      "epoch": 0.8091397849462365,
+      "grad_norm": 0.5122035327018767,
+      "learning_rate": 2.7982256077151482e-05,
+      "loss": 1.0098,
+      "step": 301
+    },
+    {
+      "epoch": 0.8118279569892473,
+      "grad_norm": 0.38079784946729706,
+      "learning_rate": 2.7900343573150003e-05,
+      "loss": 1.0097,
+      "step": 302
+    },
+    {
+      "epoch": 0.8145161290322581,
+      "grad_norm": 0.3583539130301541,
+      "learning_rate": 2.7818273743444132e-05,
+      "loss": 0.9964,
+      "step": 303
+    },
+    {
+      "epoch": 0.8172043010752689,
+      "grad_norm": 0.3813956107048218,
+      "learning_rate": 2.7736048222354414e-05,
+      "loss": 0.9761,
+      "step": 304
+    },
+    {
+      "epoch": 0.8198924731182796,
+      "grad_norm": 0.3901758217275271,
+      "learning_rate": 2.7653668647301797e-05,
+      "loss": 1.0117,
+      "step": 305
+    },
+    {
+      "epoch": 0.8225806451612904,
+      "grad_norm": 0.41237780052722667,
+      "learning_rate": 2.757113665877502e-05,
+      "loss": 0.9653,
+      "step": 306
+    },
+    {
+      "epoch": 0.8252688172043011,
+      "grad_norm": 0.457306901223017,
+      "learning_rate": 2.748845390029794e-05,
+      "loss": 1.0524,
+      "step": 307
+    },
+    {
+      "epoch": 0.8279569892473119,
+      "grad_norm": 0.3791723859065832,
+      "learning_rate": 2.740562201839684e-05,
+      "loss": 0.9861,
+      "step": 308
+    },
+    {
+      "epoch": 0.8306451612903226,
+      "grad_norm": 0.500338650948681,
+      "learning_rate": 2.7322642662567592e-05,
+      "loss": 0.9705,
+      "step": 309
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.4052884593861236,
+      "learning_rate": 2.7239517485242836e-05,
+      "loss": 0.9892,
+      "step": 310
+    },
+    {
+      "epoch": 0.8360215053763441,
+      "grad_norm": 0.3969000439893693,
+      "learning_rate": 2.715624814175907e-05,
+      "loss": 0.9883,
+      "step": 311
+    },
+    {
+      "epoch": 0.8387096774193549,
+      "grad_norm": 0.5254585071566374,
+      "learning_rate": 2.7072836290323698e-05,
+      "loss": 1.08,
+      "step": 312
+    },
+    {
+      "epoch": 0.8413978494623656,
+      "grad_norm": 0.5111475952965409,
+      "learning_rate": 2.698928359198197e-05,
+      "loss": 1.0526,
+      "step": 313
+    },
+    {
+      "epoch": 0.8440860215053764,
+      "grad_norm": 0.4717493748353866,
+      "learning_rate": 2.6905591710583957e-05,
+      "loss": 1.0137,
+      "step": 314
+    },
+    {
+      "epoch": 0.8467741935483871,
+      "grad_norm": 0.3838063749897804,
+      "learning_rate": 2.6821762312751368e-05,
+      "loss": 0.9901,
+      "step": 315
+    },
+    {
+      "epoch": 0.8494623655913979,
+      "grad_norm": 0.3456617314343378,
+      "learning_rate": 2.6737797067844403e-05,
+      "loss": 1.0034,
+      "step": 316
+    },
+    {
+      "epoch": 0.8521505376344086,
+      "grad_norm": 0.37971130684639953,
+      "learning_rate": 2.6653697647928485e-05,
+      "loss": 0.9552,
+      "step": 317
+    },
+    {
+      "epoch": 0.8548387096774194,
+      "grad_norm": 0.3820801267530888,
+      "learning_rate": 2.656946572774095e-05,
+      "loss": 0.9236,
+      "step": 318
+    },
+    {
+      "epoch": 0.8575268817204301,
+      "grad_norm": 0.4114917943590629,
+      "learning_rate": 2.648510298465775e-05,
+      "loss": 1.0,
+      "step": 319
+    },
+    {
+      "epoch": 0.8602150537634409,
+      "grad_norm": 0.4185665498381875,
+      "learning_rate": 2.6400611098659988e-05,
+      "loss": 1.0435,
+      "step": 320
+    },
+    {
+      "epoch": 0.8629032258064516,
+      "grad_norm": 0.36227121606774076,
+      "learning_rate": 2.6315991752300503e-05,
+      "loss": 0.9797,
+      "step": 321
+    },
+    {
+      "epoch": 0.8655913978494624,
+      "grad_norm": 0.40186567244596927,
+      "learning_rate": 2.623124663067034e-05,
+      "loss": 1.0071,
+      "step": 322
+    },
+    {
+      "epoch": 0.8682795698924731,
+      "grad_norm": 0.3833356371805648,
+      "learning_rate": 2.6146377421365225e-05,
+      "loss": 1.0159,
+      "step": 323
+    },
+    {
+      "epoch": 0.8709677419354839,
+      "grad_norm": 0.41469411381713683,
+      "learning_rate": 2.6061385814451913e-05,
+      "loss": 1.0277,
+      "step": 324
+    },
+    {
+      "epoch": 0.8736559139784946,
+      "grad_norm": 0.92622435409038,
+      "learning_rate": 2.5976273502434584e-05,
+      "loss": 1.0001,
+      "step": 325
+    },
+    {
+      "epoch": 0.8763440860215054,
+      "grad_norm": 0.4316506228630945,
+      "learning_rate": 2.5891042180221094e-05,
+      "loss": 1.0712,
+      "step": 326
+    },
+    {
+      "epoch": 0.8790322580645161,
+      "grad_norm": 0.42656057546508047,
+      "learning_rate": 2.580569354508925e-05,
+      "loss": 1.0074,
+      "step": 327
+    },
+    {
+      "epoch": 0.8817204301075269,
+      "grad_norm": 0.3789318712710433,
+      "learning_rate": 2.5720229296653006e-05,
+      "loss": 1.0355,
+      "step": 328
+    },
+    {
+      "epoch": 0.8844086021505376,
+      "grad_norm": 0.367154670317836,
+      "learning_rate": 2.5634651136828597e-05,
+      "loss": 1.0394,
+      "step": 329
+    },
+    {
+      "epoch": 0.8870967741935484,
+      "grad_norm": 0.4735001007157819,
+      "learning_rate": 2.554896076980069e-05,
+      "loss": 1.0552,
+      "step": 330
+    },
+    {
+      "epoch": 0.8897849462365591,
+      "grad_norm": 0.4390567460028508,
+      "learning_rate": 2.54631599019884e-05,
+      "loss": 1.0043,
+      "step": 331
+    },
+    {
+      "epoch": 0.8924731182795699,
+      "grad_norm": 0.3642787415401991,
+      "learning_rate": 2.5377250242011338e-05,
+      "loss": 0.9854,
+      "step": 332
+    },
+    {
+      "epoch": 0.8951612903225806,
+      "grad_norm": 0.4524235630593109,
+      "learning_rate": 2.5291233500655584e-05,
+      "loss": 1.0029,
+      "step": 333
+    },
+    {
+      "epoch": 0.8978494623655914,
+      "grad_norm": 0.4097887869063476,
+      "learning_rate": 2.52051113908396e-05,
+      "loss": 1.0122,
+      "step": 334
+    },
+    {
+      "epoch": 0.9005376344086021,
+      "grad_norm": 0.3852040955735104,
+      "learning_rate": 2.5118885627580155e-05,
+      "loss": 0.9779,
+      "step": 335
+    },
+    {
+      "epoch": 0.9032258064516129,
+      "grad_norm": 0.40481656602470306,
+      "learning_rate": 2.5032557927958116e-05,
+      "loss": 1.0125,
+      "step": 336
+    },
+    {
+      "epoch": 0.9059139784946236,
+      "grad_norm": 0.4118716752579493,
+      "learning_rate": 2.494613001108431e-05,
+      "loss": 1.0364,
+      "step": 337
+    },
+    {
+      "epoch": 0.9086021505376344,
+      "grad_norm": 0.4489453038959667,
+      "learning_rate": 2.485960359806528e-05,
+      "loss": 1.0436,
+      "step": 338
+    },
+    {
+      "epoch": 0.9112903225806451,
+      "grad_norm": 0.41112406404210244,
+      "learning_rate": 2.4772980411968975e-05,
+      "loss": 0.9545,
+      "step": 339
+    },
+    {
+      "epoch": 0.9139784946236559,
+      "grad_norm": 0.4856093390929945,
+      "learning_rate": 2.468626217779047e-05,
+      "loss": 0.9854,
+      "step": 340
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "grad_norm": 0.37523760134058665,
+      "learning_rate": 2.4599450622417615e-05,
+      "loss": 0.9699,
+      "step": 341
+    },
+    {
+      "epoch": 0.9193548387096774,
+      "grad_norm": 0.4064413347216363,
+      "learning_rate": 2.4512547474596624e-05,
+      "loss": 1.0083,
+      "step": 342
+    },
+    {
+      "epoch": 0.9220430107526881,
+      "grad_norm": 0.44550717714004195,
+      "learning_rate": 2.4425554464897675e-05,
+      "loss": 1.0175,
+      "step": 343
+    },
+    {
+      "epoch": 0.9247311827956989,
+      "grad_norm": 0.44076297740074416,
+      "learning_rate": 2.433847332568042e-05,
+      "loss": 0.9718,
+      "step": 344
+    },
+    {
+      "epoch": 0.9274193548387096,
+      "grad_norm": 0.4971040038925624,
+      "learning_rate": 2.4251305791059533e-05,
+      "loss": 1.0317,
+      "step": 345
+    },
+    {
+      "epoch": 0.9301075268817204,
+      "grad_norm": 0.35978037050758516,
+      "learning_rate": 2.416405359687012e-05,
+      "loss": 0.9693,
+      "step": 346
+    },
+    {
+      "epoch": 0.9327956989247311,
+      "grad_norm": 0.41817202738352904,
+      "learning_rate": 2.4076718480633178e-05,
+      "loss": 0.9764,
+      "step": 347
+    },
+    {
+      "epoch": 0.9354838709677419,
+      "grad_norm": 0.4130988765844788,
+      "learning_rate": 2.398930218152101e-05,
+      "loss": 0.9548,
+      "step": 348
+    },
+    {
+      "epoch": 0.9381720430107527,
+      "grad_norm": 0.47899471351234146,
+      "learning_rate": 2.390180644032257e-05,
+      "loss": 0.9965,
+      "step": 349
+    },
+    {
+      "epoch": 0.9408602150537635,
+      "grad_norm": 0.3639159912649112,
+      "learning_rate": 2.38142329994088e-05,
+      "loss": 0.945,
+      "step": 350
+    },
+    {
+      "epoch": 0.9435483870967742,
+      "grad_norm": 0.41552533932477614,
+      "learning_rate": 2.372658360269796e-05,
+      "loss": 0.976,
+      "step": 351
+    },
+    {
+      "epoch": 0.946236559139785,
+      "grad_norm": 0.4127471276078075,
+      "learning_rate": 2.363885999562084e-05,
+      "loss": 1.0493,
+      "step": 352
+    },
+    {
+      "epoch": 0.9489247311827957,
+      "grad_norm": 0.42874463629780296,
+      "learning_rate": 2.3551063925086072e-05,
+      "loss": 1.0003,
+      "step": 353
+    },
+    {
+      "epoch": 0.9516129032258065,
+      "grad_norm": 0.4542236208271591,
+      "learning_rate": 2.3463197139445284e-05,
+      "loss": 1.0189,
+      "step": 354
+    },
+    {
+      "epoch": 0.9543010752688172,
+      "grad_norm": 0.8840248169596676,
+      "learning_rate": 2.3375261388458318e-05,
+      "loss": 1.0006,
+      "step": 355
+    },
+    {
+      "epoch": 0.956989247311828,
+      "grad_norm": 0.47762507803159143,
+      "learning_rate": 2.3287258423258405e-05,
+      "loss": 1.0101,
+      "step": 356
+    },
+    {
+      "epoch": 0.9596774193548387,
+      "grad_norm": 0.42765004964798886,
+      "learning_rate": 2.3199189996317205e-05,
+      "loss": 0.9896,
+      "step": 357
+    },
+    {
+      "epoch": 0.9623655913978495,
+      "grad_norm": 0.4236101839000849,
+      "learning_rate": 2.3111057861410026e-05,
+      "loss": 0.9931,
+      "step": 358
+    },
+    {
+      "epoch": 0.9650537634408602,
+      "grad_norm": 0.38884571703952686,
+      "learning_rate": 2.3022863773580813e-05,
+      "loss": 0.9394,
+      "step": 359
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 0.5378824587688318,
+      "learning_rate": 2.2934609489107236e-05,
+      "loss": 0.9842,
+      "step": 360
+    },
+    {
+      "epoch": 0.9704301075268817,
+      "grad_norm": 0.39925462372416454,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 1.0026,
+      "step": 361
+    },
+    {
+      "epoch": 0.9731182795698925,
+      "grad_norm": 0.9592078982505338,
+      "learning_rate": 2.2757927361296376e-05,
+      "loss": 1.0332,
+      "step": 362
+    },
+    {
+      "epoch": 0.9758064516129032,
+      "grad_norm": 0.4396877320552629,
+      "learning_rate": 2.2669503036368124e-05,
+      "loss": 0.9971,
+      "step": 363
+    },
+    {
+      "epoch": 0.978494623655914,
+      "grad_norm": 0.38966539914800313,
+      "learning_rate": 2.2581025551543516e-05,
+      "loss": 0.9469,
+      "step": 364
+    },
+    {
+      "epoch": 0.9811827956989247,
+      "grad_norm": 0.4216276354211585,
+      "learning_rate": 2.249249666874372e-05,
+      "loss": 1.0322,
+      "step": 365
+    },
+    {
+      "epoch": 0.9838709677419355,
+      "grad_norm": 0.4351959975704115,
+      "learning_rate": 2.240391815091344e-05,
+      "loss": 0.962,
+      "step": 366
+    },
+    {
+      "epoch": 0.9865591397849462,
+      "grad_norm": 0.35811079366878923,
+      "learning_rate": 2.2315291761985803e-05,
+      "loss": 0.9937,
+      "step": 367
+    },
+    {
+      "epoch": 0.989247311827957,
+      "grad_norm": 0.3605918004740936,
+      "learning_rate": 2.222661926684722e-05,
+      "loss": 0.991,
+      "step": 368
+    },
+    {
+      "epoch": 0.9919354838709677,
+      "grad_norm": 0.4176512601533839,
+      "learning_rate": 2.2137902431302264e-05,
+      "loss": 1.0332,
+      "step": 369
+    },
+    {
+      "epoch": 0.9946236559139785,
+      "grad_norm": 0.42340462982190896,
+      "learning_rate": 2.2049143022038472e-05,
+      "loss": 0.9922,
+      "step": 370
+    },
+    {
+      "epoch": 0.9973118279569892,
+      "grad_norm": 0.420010163587815,
+      "learning_rate": 2.196034280659122e-05,
+      "loss": 1.0155,
+      "step": 371
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.41657151819377736,
+      "learning_rate": 2.1871503553308447e-05,
+      "loss": 0.9901,
+      "step": 372
+    },
+    {
+      "epoch": 1.0026881720430108,
+      "grad_norm": 0.37625522072539047,
+      "learning_rate": 2.178262703131552e-05,
+      "loss": 0.9968,
+      "step": 373
+    },
+    {
+      "epoch": 1.0053763440860215,
+      "grad_norm": 0.3372266500196924,
+      "learning_rate": 2.169371501047995e-05,
+      "loss": 0.9412,
+      "step": 374
+    },
+    {
+      "epoch": 1.0080645161290323,
+      "grad_norm": 0.4054609590993035,
+      "learning_rate": 2.160476926137616e-05,
+      "loss": 0.9854,
+      "step": 375
+    },
+    {
+      "epoch": 1.010752688172043,
+      "grad_norm": 0.4699260706715865,
+      "learning_rate": 2.1515791555250236e-05,
+      "loss": 0.9842,
+      "step": 376
+    },
+    {
+      "epoch": 1.0134408602150538,
+      "grad_norm": 0.4110199971228794,
+      "learning_rate": 2.1426783663984648e-05,
+      "loss": 0.9907,
+      "step": 377
+    },
+    {
+      "epoch": 1.0161290322580645,
+      "grad_norm": 0.37859507690512056,
+      "learning_rate": 2.133774736006297e-05,
+      "loss": 0.9802,
+      "step": 378
+    },
+    {
+      "epoch": 1.0188172043010753,
+      "grad_norm": 0.3899354406871148,
+      "learning_rate": 2.1248684416534586e-05,
+      "loss": 0.9572,
+      "step": 379
+    },
+    {
+      "epoch": 1.021505376344086,
+      "grad_norm": 0.440058465578572,
+      "learning_rate": 2.115959660697935e-05,
+      "loss": 0.9844,
+      "step": 380
+    },
+    {
+      "epoch": 1.0241935483870968,
+      "grad_norm": 0.3520254725000719,
+      "learning_rate": 2.1070485705472305e-05,
+      "loss": 0.9814,
+      "step": 381
+    },
+    {
+      "epoch": 1.0268817204301075,
+      "grad_norm": 0.3539722695025887,
+      "learning_rate": 2.0981353486548363e-05,
+      "loss": 0.9639,
+      "step": 382
+    },
+    {
+      "epoch": 1.0295698924731183,
+      "grad_norm": 0.4194473931929923,
+      "learning_rate": 2.0892201725166918e-05,
+      "loss": 0.9535,
+      "step": 383
+    },
+    {
+      "epoch": 1.032258064516129,
+      "grad_norm": 0.40610578232489386,
+      "learning_rate": 2.0803032196676542e-05,
+      "loss": 0.9503,
+      "step": 384
+    },
+    {
+      "epoch": 1.0349462365591398,
+      "grad_norm": 0.39740618952578477,
+      "learning_rate": 2.0713846676779613e-05,
+      "loss": 0.9649,
+      "step": 385
+    },
+    {
+      "epoch": 1.0376344086021505,
+      "grad_norm": 0.40362301965301367,
+      "learning_rate": 2.0624646941496957e-05,
+      "loss": 0.9439,
+      "step": 386
+    },
+    {
+      "epoch": 1.0403225806451613,
+      "grad_norm": 0.3941991828164143,
+      "learning_rate": 2.0535434767132495e-05,
+      "loss": 0.9714,
+      "step": 387
+    },
+    {
+      "epoch": 1.043010752688172,
+      "grad_norm": 0.43714978534391263,
+      "learning_rate": 2.0446211930237828e-05,
+      "loss": 0.9825,
+      "step": 388
+    },
+    {
+      "epoch": 1.0456989247311828,
+      "grad_norm": 0.4288060089603866,
+      "learning_rate": 2.0356980207576923e-05,
+      "loss": 0.955,
+      "step": 389
+    },
+    {
+      "epoch": 1.0483870967741935,
+      "grad_norm": 0.42501218262610596,
+      "learning_rate": 2.026774137609068e-05,
+      "loss": 0.9294,
+      "step": 390
+    },
+    {
+      "epoch": 1.0510752688172043,
+      "grad_norm": 0.3797408497083259,
+      "learning_rate": 2.017849721286155e-05,
+      "loss": 0.925,
+      "step": 391
+    },
+    {
+      "epoch": 1.053763440860215,
+      "grad_norm": 0.4192330864087085,
+      "learning_rate": 2.0089249495078186e-05,
+      "loss": 0.9942,
+      "step": 392
+    },
+    {
+      "epoch": 1.0564516129032258,
+      "grad_norm": 0.432069127816113,
+      "learning_rate": 2e-05,
+      "loss": 0.991,
+      "step": 393
+    },
+    {
+      "epoch": 1.0591397849462365,
+      "grad_norm": 0.4099267346630584,
+      "learning_rate": 1.991075050492182e-05,
+      "loss": 0.9895,
+      "step": 394
+    },
+    {
+      "epoch": 1.0618279569892473,
+      "grad_norm": 0.4495162016467118,
+      "learning_rate": 1.9821502787138457e-05,
+      "loss": 0.9581,
+      "step": 395
+    },
+    {
+      "epoch": 1.064516129032258,
+      "grad_norm": 0.4164111570075975,
+      "learning_rate": 1.973225862390933e-05,
+      "loss": 0.9372,
+      "step": 396
+    },
+    {
+      "epoch": 1.0672043010752688,
+      "grad_norm": 0.42182057950960955,
+      "learning_rate": 1.964301979242308e-05,
+      "loss": 0.8968,
+      "step": 397
+    },
+    {
+      "epoch": 1.0698924731182795,
+      "grad_norm": 0.40971128479229557,
+      "learning_rate": 1.955378806976218e-05,
+      "loss": 0.9579,
+      "step": 398
+    },
+    {
+      "epoch": 1.0725806451612903,
+      "grad_norm": 0.3788909632878751,
+      "learning_rate": 1.9464565232867512e-05,
+      "loss": 0.9528,
+      "step": 399
+    },
+    {
+      "epoch": 1.075268817204301,
+      "grad_norm": 0.43822514312132327,
+      "learning_rate": 1.9375353058503054e-05,
+      "loss": 0.9564,
+      "step": 400
+    },
+    {
+      "epoch": 1.0779569892473118,
+      "grad_norm": 0.3639778816066556,
+      "learning_rate": 1.9286153323220393e-05,
+      "loss": 0.9153,
+      "step": 401
+    },
+    {
+      "epoch": 1.0806451612903225,
+      "grad_norm": 0.43609982064320735,
+      "learning_rate": 1.9196967803323464e-05,
+      "loss": 0.8832,
+      "step": 402
+    },
+    {
+      "epoch": 1.0833333333333333,
+      "grad_norm": 0.39633091853296737,
+      "learning_rate": 1.9107798274833092e-05,
+      "loss": 0.9292,
+      "step": 403
+    },
+    {
+      "epoch": 1.086021505376344,
+      "grad_norm": 0.3522476588548908,
+      "learning_rate": 1.901864651345164e-05,
+      "loss": 0.942,
+      "step": 404
+    },
+    {
+      "epoch": 1.0887096774193548,
+      "grad_norm": 0.3273883068895211,
+      "learning_rate": 1.8929514294527698e-05,
+      "loss": 0.8661,
+      "step": 405
+    },
+    {
+      "epoch": 1.0913978494623655,
+      "grad_norm": 0.4476510585706847,
+      "learning_rate": 1.8840403393020663e-05,
+      "loss": 0.9595,
+      "step": 406
+    },
+    {
+      "epoch": 1.0940860215053763,
+      "grad_norm": 0.3901195297026457,
+      "learning_rate": 1.875131558346542e-05,
+      "loss": 0.9486,
+      "step": 407
+    },
+    {
+      "epoch": 1.096774193548387,
+      "grad_norm": 0.3485725672680906,
+      "learning_rate": 1.866225263993703e-05,
+      "loss": 0.8658,
+      "step": 408
+    },
+    {
+      "epoch": 1.0994623655913978,
+      "grad_norm": 0.41331554397279857,
+      "learning_rate": 1.8573216336015355e-05,
+      "loss": 0.9012,
+      "step": 409
+    },
+    {
+      "epoch": 1.1021505376344085,
+      "grad_norm": 0.4151212580924691,
+      "learning_rate": 1.848420844474977e-05,
+      "loss": 0.9663,
+      "step": 410
+    },
+    {
+      "epoch": 1.1048387096774193,
+      "grad_norm": 0.38617856497225356,
+      "learning_rate": 1.839523073862385e-05,
+      "loss": 0.9066,
+      "step": 411
+    },
+    {
+      "epoch": 1.10752688172043,
+      "grad_norm": 0.3813756613319048,
+      "learning_rate": 1.8306284989520055e-05,
+      "loss": 0.9128,
+      "step": 412
+    },
+    {
+      "epoch": 1.1102150537634408,
+      "grad_norm": 0.3756411514924831,
+      "learning_rate": 1.8217372968684483e-05,
+      "loss": 0.9689,
+      "step": 413
+    },
+    {
+      "epoch": 1.1129032258064515,
+      "grad_norm": 0.40020922129279124,
+      "learning_rate": 1.8128496446691563e-05,
+      "loss": 0.9321,
+      "step": 414
+    },
+    {
+      "epoch": 1.1155913978494623,
+      "grad_norm": 0.37642937721697783,
+      "learning_rate": 1.8039657193408788e-05,
+      "loss": 0.9186,
+      "step": 415
+    },
+    {
+      "epoch": 1.118279569892473,
+      "grad_norm": 0.42399625828357296,
+      "learning_rate": 1.795085697796153e-05,
+      "loss": 0.9169,
+      "step": 416
+    },
+    {
+      "epoch": 1.120967741935484,
+      "grad_norm": 0.4030377840102742,
+      "learning_rate": 1.786209756869775e-05,
+      "loss": 0.9179,
+      "step": 417
+    },
+    {
+      "epoch": 1.1236559139784945,
+      "grad_norm": 0.4079171223898751,
+      "learning_rate": 1.7773380733152786e-05,
+      "loss": 0.9482,
+      "step": 418
+    },
+    {
+      "epoch": 1.1263440860215055,
+      "grad_norm": 0.38158416365856496,
+      "learning_rate": 1.76847082380142e-05,
+      "loss": 0.9027,
+      "step": 419
+    },
+    {
+      "epoch": 1.129032258064516,
+      "grad_norm": 0.4124329106576186,
+      "learning_rate": 1.7596081849086562e-05,
+      "loss": 0.8622,
+      "step": 420
+    },
+    {
+      "epoch": 1.131720430107527,
+      "grad_norm": 0.33896062702559865,
+      "learning_rate": 1.7507503331256283e-05,
+      "loss": 0.9051,
+      "step": 421
+    },
+    {
+      "epoch": 1.1344086021505375,
+      "grad_norm": 0.4010835912883116,
+      "learning_rate": 1.741897444845649e-05,
+      "loss": 0.9089,
+      "step": 422
+    },
+    {
+      "epoch": 1.1370967741935485,
+      "grad_norm": 0.39948068332931136,
+      "learning_rate": 1.7330496963631883e-05,
+      "loss": 0.9289,
+      "step": 423
+    },
+    {
+      "epoch": 1.139784946236559,
+      "grad_norm": 0.41161766967963404,
+      "learning_rate": 1.7242072638703627e-05,
+      "loss": 0.8759,
+      "step": 424
+    },
+    {
+      "epoch": 1.14247311827957,
+      "grad_norm": 0.3528790684520857,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.933,
+      "step": 425
+    },
+    {
+      "epoch": 1.1451612903225807,
+      "grad_norm": 0.3520898679904786,
+      "learning_rate": 1.7065390510892767e-05,
+      "loss": 0.9317,
+      "step": 426
+    },
+    {
+      "epoch": 1.1478494623655915,
+      "grad_norm": 0.3879958084060956,
+      "learning_rate": 1.6977136226419187e-05,
+      "loss": 0.912,
+      "step": 427
+    },
+    {
+      "epoch": 1.1505376344086022,
+      "grad_norm": 0.38003669279752317,
+      "learning_rate": 1.6888942138589977e-05,
+      "loss": 0.8905,
+      "step": 428
+    },
+    {
+      "epoch": 1.153225806451613,
+      "grad_norm": 0.38751859131970406,
+      "learning_rate": 1.68008100036828e-05,
+      "loss": 0.9537,
+      "step": 429
+    },
+    {
+      "epoch": 1.1559139784946237,
+      "grad_norm": 0.35387610231825767,
+      "learning_rate": 1.67127415767416e-05,
+      "loss": 0.9069,
+      "step": 430
+    },
+    {
+      "epoch": 1.1586021505376345,
+      "grad_norm": 0.3791427138266714,
+      "learning_rate": 1.6624738611541685e-05,
+      "loss": 0.9218,
+      "step": 431
+    },
+    {
+      "epoch": 1.1612903225806452,
+      "grad_norm": 0.3802278199985049,
+      "learning_rate": 1.6536802860554723e-05,
+      "loss": 0.8637,
+      "step": 432
+    },
+    {
+      "epoch": 1.163978494623656,
+      "grad_norm": 0.4032654259366595,
+      "learning_rate": 1.6448936074913938e-05,
+      "loss": 0.8948,
+      "step": 433
+    },
+    {
+      "epoch": 1.1666666666666667,
+      "grad_norm": 0.5392331130434859,
+      "learning_rate": 1.6361140004379165e-05,
+      "loss": 0.9121,
+      "step": 434
+    },
+    {
+      "epoch": 1.1693548387096775,
+      "grad_norm": 0.4241698900501277,
+      "learning_rate": 1.6273416397302043e-05,
+      "loss": 0.9073,
+      "step": 435
+    },
+    {
+      "epoch": 1.1720430107526882,
+      "grad_norm": 0.4332038553957603,
+      "learning_rate": 1.6185767000591202e-05,
+      "loss": 0.9443,
+      "step": 436
+    },
+    {
+      "epoch": 1.174731182795699,
+      "grad_norm": 0.36935972988007515,
+      "learning_rate": 1.609819355967744e-05,
+      "loss": 0.9264,
+      "step": 437
+    },
+    {
+      "epoch": 1.1774193548387097,
+      "grad_norm": 0.3451086959697611,
+      "learning_rate": 1.6010697818478996e-05,
+      "loss": 0.8848,
+      "step": 438
+    },
+    {
+      "epoch": 1.1801075268817205,
+      "grad_norm": 0.4120432714099029,
+      "learning_rate": 1.5923281519366832e-05,
+      "loss": 0.901,
+      "step": 439
+    },
+    {
+      "epoch": 1.1827956989247312,
+      "grad_norm": 0.45212322860187754,
+      "learning_rate": 1.5835946403129886e-05,
+      "loss": 0.906,
+      "step": 440
+    },
+    {
+      "epoch": 1.185483870967742,
+      "grad_norm": 0.3604005645749173,
+      "learning_rate": 1.5748694208940467e-05,
+      "loss": 0.8777,
+      "step": 441
+    },
+    {
+      "epoch": 1.1881720430107527,
+      "grad_norm": 0.3604056072515275,
+      "learning_rate": 1.5661526674319582e-05,
+      "loss": 0.919,
+      "step": 442
+    },
+    {
+      "epoch": 1.1908602150537635,
+      "grad_norm": 0.34313703093643527,
+      "learning_rate": 1.557444553510233e-05,
+      "loss": 0.9329,
+      "step": 443
+    },
+    {
+      "epoch": 1.1935483870967742,
+      "grad_norm": 0.37375432993164387,
+      "learning_rate": 1.548745252540339e-05,
+      "loss": 0.9481,
+      "step": 444
+    },
+    {
+      "epoch": 1.196236559139785,
+      "grad_norm": 0.43080367488426863,
+      "learning_rate": 1.5400549377582392e-05,
+      "loss": 0.894,
+      "step": 445
+    },
+    {
+      "epoch": 1.1989247311827957,
+      "grad_norm": 0.3750508182193543,
+      "learning_rate": 1.5313737822209532e-05,
+      "loss": 0.922,
+      "step": 446
+    },
+    {
+      "epoch": 1.2016129032258065,
+      "grad_norm": 0.4276115589621554,
+      "learning_rate": 1.5227019588031035e-05,
+      "loss": 0.8569,
+      "step": 447
+    },
+    {
+      "epoch": 1.2043010752688172,
+      "grad_norm": 0.36909790172392043,
+      "learning_rate": 1.5140396401934725e-05,
+      "loss": 0.8656,
+      "step": 448
+    },
+    {
+      "epoch": 1.206989247311828,
+      "grad_norm": 0.44266858883203647,
+      "learning_rate": 1.5053869988915691e-05,
+      "loss": 0.9371,
+      "step": 449
+    },
+    {
+      "epoch": 1.2096774193548387,
+      "grad_norm": 0.3725440482769685,
+      "learning_rate": 1.4967442072041895e-05,
+      "loss": 0.9103,
+      "step": 450
+    },
+    {
+      "epoch": 1.2123655913978495,
+      "grad_norm": 0.36815385084670166,
+      "learning_rate": 1.4881114372419854e-05,
+      "loss": 0.8499,
+      "step": 451
+    },
+    {
+      "epoch": 1.2150537634408602,
+      "grad_norm": 0.3730907651609142,
+      "learning_rate": 1.47948886091604e-05,
+      "loss": 0.928,
+      "step": 452
+    },
+    {
+      "epoch": 1.217741935483871,
+      "grad_norm": 0.4105069671252027,
+      "learning_rate": 1.4708766499344424e-05,
+      "loss": 0.8816,
+      "step": 453
+    },
+    {
+      "epoch": 1.2204301075268817,
+      "grad_norm": 0.3606969658371548,
+      "learning_rate": 1.462274975798867e-05,
+      "loss": 0.8993,
+      "step": 454
+    },
+    {
+      "epoch": 1.2231182795698925,
+      "grad_norm": 0.3819716903785989,
+      "learning_rate": 1.4536840098011613e-05,
+      "loss": 0.9016,
+      "step": 455
+    },
+    {
+      "epoch": 1.2258064516129032,
+      "grad_norm": 0.3468890936243422,
+      "learning_rate": 1.4451039230199317e-05,
+      "loss": 0.8533,
+      "step": 456
+    },
+    {
+      "epoch": 1.228494623655914,
+      "grad_norm": 0.4219447671378407,
+      "learning_rate": 1.4365348863171406e-05,
+      "loss": 0.8838,
+      "step": 457
+    },
+    {
+      "epoch": 1.2311827956989247,
+      "grad_norm": 0.38132389434045155,
+      "learning_rate": 1.4279770703347008e-05,
+      "loss": 0.9027,
+      "step": 458
+    },
+    {
+      "epoch": 1.2338709677419355,
+      "grad_norm": 0.40246629352796237,
+      "learning_rate": 1.4194306454910757e-05,
+      "loss": 0.9246,
+      "step": 459
+    },
+    {
+      "epoch": 1.2365591397849462,
+      "grad_norm": 0.3909042227194992,
+      "learning_rate": 1.410895781977891e-05,
+      "loss": 0.8486,
+      "step": 460
+    },
+    {
+      "epoch": 1.239247311827957,
+      "grad_norm": 0.3674117695470089,
+      "learning_rate": 1.4023726497565422e-05,
+      "loss": 0.8977,
+      "step": 461
+    },
+    {
+      "epoch": 1.2419354838709677,
+      "grad_norm": 0.41824933124811753,
+      "learning_rate": 1.3938614185548094e-05,
+      "loss": 0.8888,
+      "step": 462
+    },
+    {
+      "epoch": 1.2446236559139785,
+      "grad_norm": 0.40078055707466587,
+      "learning_rate": 1.385362257863478e-05,
+      "loss": 0.8963,
+      "step": 463
+    },
+    {
+      "epoch": 1.2473118279569892,
+      "grad_norm": 0.3611909331897887,
+      "learning_rate": 1.3768753369329664e-05,
+      "loss": 0.9065,
+      "step": 464
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3889978301002571,
+      "learning_rate": 1.3684008247699505e-05,
+      "loss": 0.8655,
+      "step": 465
+    },
+    {
+      "epoch": 1.2526881720430108,
+      "grad_norm": 0.3967333759192253,
+      "learning_rate": 1.3599388901340019e-05,
+      "loss": 0.9289,
+      "step": 466
+    },
+    {
+      "epoch": 1.2553763440860215,
+      "grad_norm": 0.3930903953239825,
+      "learning_rate": 1.3514897015342257e-05,
+      "loss": 0.9004,
+      "step": 467
+    },
+    {
+      "epoch": 1.2580645161290323,
+      "grad_norm": 0.39782516966072734,
+      "learning_rate": 1.343053427225905e-05,
+      "loss": 0.9356,
+      "step": 468
+    },
+    {
+      "epoch": 1.260752688172043,
+      "grad_norm": 0.3755719466909933,
+      "learning_rate": 1.3346302352071525e-05,
+      "loss": 0.9308,
+      "step": 469
+    },
+    {
+      "epoch": 1.2634408602150538,
+      "grad_norm": 0.4574705672937729,
+      "learning_rate": 1.3262202932155602e-05,
+      "loss": 0.8518,
+      "step": 470
+    },
+    {
+      "epoch": 1.2661290322580645,
+      "grad_norm": 0.387011509571043,
+      "learning_rate": 1.3178237687248632e-05,
+      "loss": 0.9082,
+      "step": 471
+    },
+    {
+      "epoch": 1.2688172043010753,
+      "grad_norm": 0.4362071151177883,
+      "learning_rate": 1.3094408289416052e-05,
+      "loss": 0.8711,
+      "step": 472
+    },
+    {
+      "epoch": 1.271505376344086,
+      "grad_norm": 0.39574828340378626,
+      "learning_rate": 1.3010716408018037e-05,
+      "loss": 0.8927,
+      "step": 473
+    },
+    {
+      "epoch": 1.2741935483870968,
+      "grad_norm": 0.38582577248876887,
+      "learning_rate": 1.2927163709676305e-05,
+      "loss": 0.8953,
+      "step": 474
+    },
+    {
+      "epoch": 1.2768817204301075,
+      "grad_norm": 0.39993052025189596,
+      "learning_rate": 1.2843751858240938e-05,
+      "loss": 0.8545,
+      "step": 475
+    },
+    {
+      "epoch": 1.2795698924731183,
+      "grad_norm": 0.3758518397155174,
+      "learning_rate": 1.276048251475717e-05,
+      "loss": 0.8853,
+      "step": 476
+    },
+    {
+      "epoch": 1.282258064516129,
+      "grad_norm": 0.3673034097829057,
+      "learning_rate": 1.267735733743242e-05,
+      "loss": 0.8618,
+      "step": 477
+    },
+    {
+      "epoch": 1.2849462365591398,
+      "grad_norm": 0.3579444429196611,
+      "learning_rate": 1.2594377981603167e-05,
+      "loss": 0.9049,
+      "step": 478
+    },
+    {
+      "epoch": 1.2876344086021505,
+      "grad_norm": 0.3512023955004902,
+      "learning_rate": 1.251154609970206e-05,
+      "loss": 0.905,
+      "step": 479
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 0.3821816087256963,
+      "learning_rate": 1.2428863341224988e-05,
+      "loss": 0.8603,
+      "step": 480
+    },
+    {
+      "epoch": 1.293010752688172,
+      "grad_norm": 0.4133864396732673,
+      "learning_rate": 1.2346331352698206e-05,
+      "loss": 0.8936,
+      "step": 481
+    },
+    {
+      "epoch": 1.2956989247311828,
+      "grad_norm": 0.4291865828849791,
+      "learning_rate": 1.2263951777645588e-05,
+      "loss": 0.8951,
+      "step": 482
+    },
+    {
+      "epoch": 1.2983870967741935,
+      "grad_norm": 0.3960987966319793,
+      "learning_rate": 1.2181726256555877e-05,
+      "loss": 0.8656,
+      "step": 483
+    },
+    {
+      "epoch": 1.3010752688172043,
+      "grad_norm": 0.3666707325539653,
+      "learning_rate": 1.2099656426850004e-05,
+      "loss": 0.9047,
+      "step": 484
+    },
+    {
+      "epoch": 1.303763440860215,
+      "grad_norm": 0.36431675184531326,
+      "learning_rate": 1.2017743922848518e-05,
+      "loss": 0.8792,
+      "step": 485
+    },
+    {
+      "epoch": 1.3064516129032258,
+      "grad_norm": 0.4255014642528041,
+      "learning_rate": 1.1935990375739011e-05,
+      "loss": 0.8804,
+      "step": 486
+    },
+    {
+      "epoch": 1.3091397849462365,
+      "grad_norm": 0.381858451178781,
+      "learning_rate": 1.1854397413543626e-05,
+      "loss": 0.919,
+      "step": 487
+    },
+    {
+      "epoch": 1.3118279569892473,
+      "grad_norm": 0.35522559885326904,
+      "learning_rate": 1.177296666108669e-05,
+      "loss": 0.9347,
+      "step": 488
+    },
+    {
+      "epoch": 1.314516129032258,
+      "grad_norm": 0.37099789206174566,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.936,
+      "step": 489
+    },
+    {
+      "epoch": 1.3172043010752688,
+      "grad_norm": 0.36298491750372225,
+      "learning_rate": 1.1610598268501982e-05,
+      "loss": 0.8553,
+      "step": 490
+    },
+    {
+      "epoch": 1.3198924731182795,
+      "grad_norm": 0.3634911000699515,
+      "learning_rate": 1.1529663861742692e-05,
+      "loss": 0.9017,
+      "step": 491
+    },
+    {
+      "epoch": 1.3225806451612903,
+      "grad_norm": 0.4152917791390396,
+      "learning_rate": 1.1448898131394364e-05,
+      "loss": 0.8788,
+      "step": 492
+    },
+    {
+      "epoch": 1.325268817204301,
+      "grad_norm": 0.4244046126280729,
+      "learning_rate": 1.1368302685807984e-05,
+      "loss": 0.9105,
+      "step": 493
+    },
+    {
+      "epoch": 1.3279569892473118,
+      "grad_norm": 0.4305679319947305,
+      "learning_rate": 1.1287879129943558e-05,
+      "loss": 0.8712,
+      "step": 494
+    },
+    {
+      "epoch": 1.3306451612903225,
+      "grad_norm": 0.3815632433540905,
+      "learning_rate": 1.1207629065338063e-05,
+      "loss": 0.9274,
+      "step": 495
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.395491462517581,
+      "learning_rate": 1.1127554090073639e-05,
+      "loss": 0.8852,
+      "step": 496
+    },
+    {
+      "epoch": 1.336021505376344,
+      "grad_norm": 0.3412204028961129,
+      "learning_rate": 1.1047655798745752e-05,
+      "loss": 0.909,
+      "step": 497
+    },
+    {
+      "epoch": 1.3387096774193548,
+      "grad_norm": 0.42252187704467103,
+      "learning_rate": 1.0967935782431382e-05,
+      "loss": 0.9152,
+      "step": 498
+    },
+    {
+      "epoch": 1.3413978494623655,
+      "grad_norm": 0.3660726577260679,
+      "learning_rate": 1.0888395628657413e-05,
+      "loss": 0.9087,
+      "step": 499
+    },
+    {
+      "epoch": 1.3440860215053765,
+      "grad_norm": 0.41163724258371354,
+      "learning_rate": 1.0809036921368966e-05,
+      "loss": 0.8401,
+      "step": 500
+    },
+    {
+      "epoch": 1.346774193548387,
+      "grad_norm": 0.4662876589603711,
+      "learning_rate": 1.0729861240897892e-05,
+      "loss": 0.8499,
+      "step": 501
+    },
+    {
+      "epoch": 1.349462365591398,
+      "grad_norm": 0.38293219107966625,
+      "learning_rate": 1.0650870163931275e-05,
+      "loss": 0.8773,
+      "step": 502
+    },
+    {
+      "epoch": 1.3521505376344085,
+      "grad_norm": 0.4100231412619493,
+      "learning_rate": 1.0572065263480046e-05,
+      "loss": 0.874,
+      "step": 503
+    },
+    {
+      "epoch": 1.3548387096774195,
+      "grad_norm": 0.3609567532765614,
+      "learning_rate": 1.0493448108847669e-05,
+      "loss": 0.9064,
+      "step": 504
+    },
+    {
+      "epoch": 1.35752688172043,
+      "grad_norm": 0.3995575869896119,
+      "learning_rate": 1.0415020265598872e-05,
+      "loss": 0.8599,
+      "step": 505
+    },
+    {
+      "epoch": 1.360215053763441,
+      "grad_norm": 0.39891198490046775,
+      "learning_rate": 1.0336783295528454e-05,
+      "loss": 0.8696,
+      "step": 506
+    },
+    {
+      "epoch": 1.3629032258064515,
+      "grad_norm": 0.3869673460119376,
+      "learning_rate": 1.0258738756630255e-05,
+      "loss": 0.8882,
+      "step": 507
+    },
+    {
+      "epoch": 1.3655913978494625,
+      "grad_norm": 0.3785572445627051,
+      "learning_rate": 1.0180888203066059e-05,
+      "loss": 0.8653,
+      "step": 508
+    },
+    {
+      "epoch": 1.368279569892473,
+      "grad_norm": 0.4042858657113599,
+      "learning_rate": 1.0103233185134647e-05,
+      "loss": 0.8197,
+      "step": 509
+    },
+    {
+      "epoch": 1.370967741935484,
+      "grad_norm": 0.3930704370351199,
+      "learning_rate": 1.0025775249240993e-05,
+      "loss": 0.8542,
+      "step": 510
+    },
+    {
+      "epoch": 1.3736559139784945,
+      "grad_norm": 0.35542964413505407,
+      "learning_rate": 9.948515937865375e-06,
+      "loss": 0.8476,
+      "step": 511
+    },
+    {
+      "epoch": 1.3763440860215055,
+      "grad_norm": 0.42702317687809166,
+      "learning_rate": 9.871456789532736e-06,
+      "loss": 0.9089,
+      "step": 512
+    },
+    {
+      "epoch": 1.379032258064516,
+      "grad_norm": 0.3930984004263597,
+      "learning_rate": 9.794599338782011e-06,
+      "loss": 0.9033,
+      "step": 513
+    },
+    {
+      "epoch": 1.381720430107527,
+      "grad_norm": 0.36398813175120787,
+      "learning_rate": 9.717945116135568e-06,
+      "loss": 0.8921,
+      "step": 514
+    },
+    {
+      "epoch": 1.3844086021505375,
+      "grad_norm": 0.3716360571507653,
+      "learning_rate": 9.641495648068739e-06,
+      "loss": 0.8826,
+      "step": 515
+    },
+    {
+      "epoch": 1.3870967741935485,
+      "grad_norm": 0.5477375096224166,
+      "learning_rate": 9.56525245697942e-06,
+      "loss": 0.9088,
+      "step": 516
+    },
+    {
+      "epoch": 1.389784946236559,
+      "grad_norm": 0.35084597720275207,
+      "learning_rate": 9.489217061157744e-06,
+      "loss": 0.8777,
+      "step": 517
+    },
+    {
+      "epoch": 1.39247311827957,
+      "grad_norm": 0.41785033927597015,
+      "learning_rate": 9.413390974755864e-06,
+      "loss": 0.8934,
+      "step": 518
+    },
+    {
+      "epoch": 1.3951612903225805,
+      "grad_norm": 0.3942271377890524,
+      "learning_rate": 9.337775707757792e-06,
+      "loss": 0.8795,
+      "step": 519
+    },
+    {
+      "epoch": 1.3978494623655915,
+      "grad_norm": 0.43266747223335905,
+      "learning_rate": 9.262372765949319e-06,
+      "loss": 0.9344,
+      "step": 520
+    },
+    {
+      "epoch": 1.400537634408602,
+      "grad_norm": 0.3555286580841176,
+      "learning_rate": 9.187183650888056e-06,
+      "loss": 0.8561,
+      "step": 521
+    },
+    {
+      "epoch": 1.403225806451613,
+      "grad_norm": 0.4326861883163117,
+      "learning_rate": 9.112209859873479e-06,
+      "loss": 0.8748,
+      "step": 522
+    },
+    {
+      "epoch": 1.4059139784946235,
+      "grad_norm": 0.3911645714163196,
+      "learning_rate": 9.037452885917197e-06,
+      "loss": 0.8741,
+      "step": 523
+    },
+    {
+      "epoch": 1.4086021505376345,
+      "grad_norm": 0.3721004990469904,
+      "learning_rate": 8.962914217713148e-06,
+      "loss": 0.9123,
+      "step": 524
+    },
+    {
+      "epoch": 1.4112903225806452,
+      "grad_norm": 0.3504669386782379,
+      "learning_rate": 8.888595339607961e-06,
+      "loss": 0.9166,
+      "step": 525
+    },
+    {
+      "epoch": 1.413978494623656,
+      "grad_norm": 0.38727418818749926,
+      "learning_rate": 8.814497731571432e-06,
+      "loss": 0.8756,
+      "step": 526
+    },
+    {
+      "epoch": 1.4166666666666667,
+      "grad_norm": 0.3530651014121482,
+      "learning_rate": 8.74062286916705e-06,
+      "loss": 0.8953,
+      "step": 527
+    },
+    {
+      "epoch": 1.4193548387096775,
+      "grad_norm": 0.9061717130965162,
+      "learning_rate": 8.666972223522559e-06,
+      "loss": 0.898,
+      "step": 528
+    },
+    {
+      "epoch": 1.4220430107526882,
+      "grad_norm": 0.3892987110948611,
+      "learning_rate": 8.593547261300716e-06,
+      "loss": 0.8561,
+      "step": 529
+    },
+    {
+      "epoch": 1.424731182795699,
+      "grad_norm": 0.38394033117089926,
+      "learning_rate": 8.520349444670093e-06,
+      "loss": 0.8868,
+      "step": 530
+    },
+    {
+      "epoch": 1.4274193548387097,
+      "grad_norm": 0.4557984907315737,
+      "learning_rate": 8.447380231275889e-06,
+      "loss": 0.8617,
+      "step": 531
+    },
+    {
+      "epoch": 1.4301075268817205,
+      "grad_norm": 0.375452470598354,
+      "learning_rate": 8.374641074210979e-06,
+      "loss": 0.8734,
+      "step": 532
+    },
+    {
+      "epoch": 1.4327956989247312,
+      "grad_norm": 0.4144183699699819,
+      "learning_rate": 8.30213342198694e-06,
+      "loss": 0.8523,
+      "step": 533
+    },
+    {
+      "epoch": 1.435483870967742,
+      "grad_norm": 0.42069408475654074,
+      "learning_rate": 8.229858718505212e-06,
+      "loss": 0.8791,
+      "step": 534
+    },
+    {
+      "epoch": 1.4381720430107527,
+      "grad_norm": 0.48297854323803846,
+      "learning_rate": 8.157818403028343e-06,
+      "loss": 0.8565,
+      "step": 535
+    },
+    {
+      "epoch": 1.4408602150537635,
+      "grad_norm": 0.39205140724296417,
+      "learning_rate": 8.086013910151334e-06,
+      "loss": 0.8831,
+      "step": 536
+    },
+    {
+      "epoch": 1.4435483870967742,
+      "grad_norm": 0.4124811246961915,
+      "learning_rate": 8.014446669773061e-06,
+      "loss": 0.8508,
+      "step": 537
+    },
+    {
+      "epoch": 1.446236559139785,
+      "grad_norm": 0.3697888279992508,
+      "learning_rate": 7.943118107067813e-06,
+      "loss": 0.8964,
+      "step": 538
+    },
+    {
+      "epoch": 1.4489247311827957,
+      "grad_norm": 0.35775344448605834,
+      "learning_rate": 7.872029642456895e-06,
+      "loss": 0.8804,
+      "step": 539
+    },
+    {
+      "epoch": 1.4516129032258065,
+      "grad_norm": 0.4050701342223883,
+      "learning_rate": 7.801182691580362e-06,
+      "loss": 0.8798,
+      "step": 540
+    },
+    {
+      "epoch": 1.4543010752688172,
+      "grad_norm": 0.37810189432390856,
+      "learning_rate": 7.730578665268815e-06,
+      "loss": 0.9053,
+      "step": 541
+    },
+    {
+      "epoch": 1.456989247311828,
+      "grad_norm": 0.4016865992606119,
+      "learning_rate": 7.66021896951529e-06,
+      "loss": 0.8787,
+      "step": 542
+    },
+    {
+      "epoch": 1.4596774193548387,
+      "grad_norm": 0.3745942539375088,
+      "learning_rate": 7.590105005447317e-06,
+      "loss": 0.8693,
+      "step": 543
+    },
+    {
+      "epoch": 1.4623655913978495,
+      "grad_norm": 0.4773383301892084,
+      "learning_rate": 7.520238169298937e-06,
+      "loss": 0.8696,
+      "step": 544
+    },
+    {
+      "epoch": 1.4650537634408602,
+      "grad_norm": 0.3915177577878335,
+      "learning_rate": 7.450619852382959e-06,
+      "loss": 0.8985,
+      "step": 545
+    },
+    {
+      "epoch": 1.467741935483871,
+      "grad_norm": 0.4002539636329483,
+      "learning_rate": 7.381251441063255e-06,
+      "loss": 0.8545,
+      "step": 546
+    },
+    {
+      "epoch": 1.4704301075268817,
+      "grad_norm": 0.4556350475396439,
+      "learning_rate": 7.312134316727093e-06,
+      "loss": 0.8445,
+      "step": 547
+    },
+    {
+      "epoch": 1.4731182795698925,
+      "grad_norm": 0.4195728788592859,
+      "learning_rate": 7.243269855757693e-06,
+      "loss": 0.9239,
+      "step": 548
+    },
+    {
+      "epoch": 1.4758064516129032,
+      "grad_norm": 0.44054388742759737,
+      "learning_rate": 7.1746594295067826e-06,
+      "loss": 0.9248,
+      "step": 549
+    },
+    {
+      "epoch": 1.478494623655914,
+      "grad_norm": 0.42709109156924885,
+      "learning_rate": 7.106304404267304e-06,
+      "loss": 0.8424,
+      "step": 550
+    },
+    {
+      "epoch": 1.4811827956989247,
+      "grad_norm": 0.39224047779710847,
+      "learning_rate": 7.0382061412461935e-06,
+      "loss": 0.8638,
+      "step": 551
+    },
+    {
+      "epoch": 1.4838709677419355,
+      "grad_norm": 0.3480064313882409,
+      "learning_rate": 6.970365996537285e-06,
+      "loss": 0.866,
+      "step": 552
+    },
+    {
+      "epoch": 1.4865591397849462,
+      "grad_norm": 0.4470655366145503,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.8854,
+      "step": 553
+    },
+    {
+      "epoch": 1.489247311827957,
+      "grad_norm": 0.427599328724393,
+      "learning_rate": 6.8354654607039535e-06,
+      "loss": 0.8864,
+      "step": 554
+    },
+    {
+      "epoch": 1.4919354838709677,
+      "grad_norm": 0.41347796402657694,
+      "learning_rate": 6.768407755959119e-06,
+      "loss": 0.8928,
+      "step": 555
+    },
+    {
+      "epoch": 1.4946236559139785,
+      "grad_norm": 0.3826360606173244,
+      "learning_rate": 6.701613542232202e-06,
+      "loss": 0.9089,
+      "step": 556
+    },
+    {
+      "epoch": 1.4973118279569892,
+      "grad_norm": 0.36892665307361505,
+      "learning_rate": 6.635084149648481e-06,
+      "loss": 0.8484,
+      "step": 557
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.36406926057608424,
+      "learning_rate": 6.568820903059632e-06,
+      "loss": 0.8632,
+      "step": 558
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 744,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 186,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2315107631628288.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}