diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 6.488517056616143,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.9838,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.491842763115525,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0117,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.560851702322604,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0097,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.4040384768274645,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0095,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.007185103718162,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.9848,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.754346175048982,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.966,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 5.042208682009327,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.9597,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.476732534864257,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.924,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.4038352780034065,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.8781,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.0785367251026665,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.842,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.06071865313496,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.8714,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.0697206432862405,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.8821,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.5829051473878,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.8981,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.292134759452397,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.8357,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.9914307248639065,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.8145,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.1545049749600516,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8292,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.7276183268592082,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.487458415786927,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.7929,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 2.030630292303825,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.725,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.9711327989466227,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.7507,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.7538210352243542,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.7609,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.7356961235322637,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.7335,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.5569984001038504,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7209,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.7781641761600944,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.6959,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.5999740993451816,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7234,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.2849329695030374,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7203,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.1590223783641709,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7068,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 1.2589802704084065,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7156,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 1.0185213402473399,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.695,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.9241035385614386,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.6902,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.921541962766516,
+      "learning_rate": 1e-05,
+      "loss": 0.6848,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.9108352302643685,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.6792,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.7839492523117404,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.6921,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7993903866277967,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.6932,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.7383786065576277,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.6943,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.714522776421187,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.6588,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6474082797811723,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.6772,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.7186868631928922,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.6856,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.6253475968452356,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.6432,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6747535466039078,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.6866,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.7097627043373957,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.6831,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6548601967809136,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.6323,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.7815314289134181,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.6691,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.5759982489730228,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.6515,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.6225290664919806,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.6303,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.6808643886321082,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6397,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5387877500957177,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6215,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5888604245478755,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6345,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.6067393048116112,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6709,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.6481280145928161,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6416,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6200783819777416,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6442,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.63368095490806,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6358,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.6420169938687877,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6701,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5903199156864191,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.6415,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6510898527277015,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6353,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6215575330384858,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.621,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.600303130763037,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6362,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.5966561206754749,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6561,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.5816009956565077,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.603,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5977834201212888,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.657,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.5879702398055634,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6326,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.5460694387997012,
+      "learning_rate": 2e-05,
+      "loss": 0.631,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.587710206419495,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6398,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6302446628560033,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6256,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.6867271921795559,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6266,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.6753313894690963,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6239,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.6145830602097488,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6109,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6833898839510378,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6244,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.5863526793747839,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6295,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.5737939734979786,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6294,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.5890363365925398,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6106,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.5400793663016636,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6024,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.5595463578542522,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6433,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.6897084952770773,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6413,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.5967443091458805,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.617,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.7483801605572824,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6157,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.6794980848070205,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6244,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.725119828800973,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6319,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.7834594785559509,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6099,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.81415490682408,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6422,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.6270020413924754,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6416,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.6968727438059128,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6292,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.6853469909517652,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6033,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.8028628535173731,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6323,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.8308449243007074,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6277,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.8059638321056457,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6007,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.7390850728185728,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.5958,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.6108418340437672,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.622,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.8631765058915774,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6272,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.937332395567973,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6413,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.6433384917082896,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.647,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.8625776265247889,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6321,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.8916249160401396,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6381,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.8432857449333482,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6345,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.8851889390776275,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6386,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.9584017296833093,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.613,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.8758045275948639,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6159,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.8408944088280765,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6193,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.9202244955210259,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6168,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.656454131896896,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6115,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.718064859890484,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6015,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.6036342198073066,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6152,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.5739625909501989,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6142,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.6630024370325052,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6181,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.6855214543746326,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6287,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.5964962788337694,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.5928,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.8017772161794435,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6288,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.8392618989906229,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6179,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.6328110997072297,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6118,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.6562031540109046,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6212,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.6403660544158826,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6175,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.8177411965614324,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.665,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.804237982354954,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.5861,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.6348860994566816,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6156,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.610501321173102,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6273,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.5696167289638654,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.5815,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.6117577497447687,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6347,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.6239548100948096,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6184,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.5781432869011526,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6251,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.6583957158912485,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.612,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.5836487309450371,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6238,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.6711502636030767,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6102,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.6516727034517207,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.609,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.7511216864402834,
+      "learning_rate": 4e-05,
+      "loss": 0.6039,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.7381066380094419,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6045,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.7461493748886927,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6438,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.6920191167068377,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6089,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.8741259559346722,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6052,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.7712716618841896,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6296,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.7729012923023032,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6107,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.6414816304911543,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6002,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.7274528111577646,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.5993,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.7623006604863163,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6178,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.6809254967628274,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6293,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.635283651093413,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.6082,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.728980608116728,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6266,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.6688343065126685,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6331,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.773846680784205,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6177,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.7068341469819536,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6433,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.7129088634127193,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6157,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.7931241245499933,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.615,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.8649786804588175,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.6257,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.84555135384226,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6181,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.6013160661260054,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.5908,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.9009131300127472,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.6238,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.6109086447819924,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6025,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.6939595677743696,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6259,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.6595340871634277,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.607,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6957552963240163,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6236,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.8263577997972275,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6264,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.7196494199748849,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.5862,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6128853267951033,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6257,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.8365218318434001,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6083,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.5901651343261879,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6473,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.7048069434578391,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.5884,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6733922356886289,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6053,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.7019992938752496,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.5935,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.6543765271262101,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.5736,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.769071696978237,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.5964,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.7385469133937939,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6222,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.7470313926634368,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6085,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.5913675036820166,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6069,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.8025093337351001,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6103,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6564664213479472,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.5973,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.7801087446325724,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6052,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6187950161847625,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.5922,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.7348413768787011,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6092,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5801205657055276,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6112,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.7732315404136701,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6015,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6878212689770185,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.5957,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6180516321968165,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.5962,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5560498400503798,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.5913,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.8130421604744894,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6106,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.4902103746293368,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6041,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.8084909316374113,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6063,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.789388611240299,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5937380533557202,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.6027,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.8228908118750582,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6203,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.6084902708088246,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6126,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6408638383319485,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.602,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.7041550271074072,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.5984,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5190993224774086,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6149,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.6195284345738444,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.603,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.4995220932517508,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.5862,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.4980721416707236,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.582,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5724544528457033,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6193,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5033384982406077,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.6014,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5251472119582573,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.5976,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.4901115365896256,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.5937,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.47766784320365435,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.5901,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.4990648049598026,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.5985,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.48075310005062755,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.586,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5171988351118791,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6242,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5260576209728298,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.5726,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5129689609007791,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.5926,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6350577915390109,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.5985,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.4615428539445055,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.5794,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6862075438253917,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.59,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.507041260060036,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.5902,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6856841298206175,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.5989,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5669036730119824,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.5976,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5831180034253164,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6028,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.49685082777889567,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6009,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5610672038702395,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6016,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5107305199489949,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.5935,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.5133861594547461,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6027,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.526542121602908,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.5851,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5328399329193324,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6242,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.6380686926439901,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6076,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.49727171641895435,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.5803,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5759841858246805,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6098,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.547829685581286,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6061,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5362028661895956,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6114,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.5127803447912183,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6103,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.6271582568130905,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6156,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.6049155161751846,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.5878,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5994798142530453,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.5998,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5657669478569832,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.5889,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.5230503584780634,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.594,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.52600569366568,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.6105,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.49650853796338473,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.578,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5536723782511638,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6042,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.49366668377552764,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.5824,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5518477520007163,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.5827,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.5735088187711106,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5902,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.6112953572604547,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.5709,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5691805986073377,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.5977,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.7286920721285883,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6074,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.49059805992748395,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.5993,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.601228197779407,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.5727,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.6329475383170163,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.6055,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5165449307414129,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.585,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.6297962532470386,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6057,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.529510469635868,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.5735,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.6681154238237161,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.5984,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.6921484453965062,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5916,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.4726146114022753,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.5804,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.6692998864361963,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5702,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.4317897302158741,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5649,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.5837947819820484,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.598,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5167716952840079,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.5969,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.47534533627356174,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.5754,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.5210524996698946,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.6204,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5177712908964579,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6308,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 1.1259545727368052,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.6334,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5863884183570057,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.6098,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5470129601281639,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.5994,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.6207622430247736,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.4991,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.4482559950027568,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.4807,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.5637537543889152,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.4966,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5878317866299134,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5134,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.5519946753754869,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.4809,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.6061856793383723,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.4778,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5630179208342209,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.4823,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5950571897542761,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.4809,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.8406621130294117,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.4966,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.5842771547402082,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.4846,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.8106679046451019,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.4727,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5787351070040733,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.4815,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 4.571526740891978,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.4823,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 1.055201558021638,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.4945,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.7787485858246274,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.4982,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.6969820458424935,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.4788,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.7792361731865706,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.4855,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.5614282972108187,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.4622,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.7843739099602726,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5076,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.547211401243452,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.4745,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.6111271860307037,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.4851,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.750134931435773,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.4701,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5072948747971479,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.4728,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.6936478782577271,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.4786,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5829518453138356,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.475,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.6927619763534547,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.4756,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5669900928980867,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.4673,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.6691815040514285,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.4741,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.5490325440433688,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.4621,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.5581198348633424,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.4853,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5116312514244309,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.4603,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.4690162242910926,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.4682,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.5115504594157863,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.4632,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5369067399994051,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5072,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.4902386218961615,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.4973,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.527017615455475,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.484,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.4922038062822375,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.4897,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5780114749817676,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.4865,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.47928189033208846,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.4849,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5717485092414771,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.4779,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5584129032365848,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.4771,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.49388388859142524,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.4528,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.567864557214818,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.4895,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5116920005770661,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.4882,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5315972187284718,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.4908,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5005123829930448,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.4878,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.6121878103354472,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.4845,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.5189415718977324,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.4671,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.45473864632774785,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.4764,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.5983862049172116,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.4813,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5309732932384092,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.4795,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5271146897249691,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.453,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5340464751484275,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5043,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.5179304809484698,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5086,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.5419445829696093,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.4707,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.4481411986442045,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.493,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.48801066782378977,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.4848,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.4727132079209016,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.4823,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.43740663407875224,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.4799,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.4395280236964227,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.4874,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.46472337643937583,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.4821,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4868657508774205,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.4823,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.4753757354811639,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.4679,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6142460517797782,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.4883,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.4485729130865517,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.4787,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.6010560579483393,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.4733,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.5085898441303737,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.4751,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.6614704560639814,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.4881,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5698837697137465,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.4752,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.536250224085041,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.503,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5682829309761901,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.465,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.4981191868155575,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.472,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.51905484537068,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.4567,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.6218565676600042,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.4768,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5063691955009686,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.4859,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.5461646529355767,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.4679,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5072657711416516,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5046,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.43475746182744146,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.4739,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.46251617960383185,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.4818,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.4793879234260547,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.4689,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.47854456633329273,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.4717,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.46084213379131406,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.4874,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.4669928876313869,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.4673,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.48057402415305756,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.4978,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.4724545642480333,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.4804,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.5632018103635066,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.4961,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5376792675884199,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.4679,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.5163250815884753,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.4731,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.48381416277325284,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.487,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.48758469652524655,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.4934,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.46349345561101346,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.4853,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.48354151724767336,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.4828,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.5076675824985196,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.4595,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.4318277034896379,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.4678,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.527383801354853,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.4683,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.4823262024393848,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.4691,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.46622234962555803,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.4875,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.473033899217531,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.4924,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.4532972564409828,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.4718,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.4590512308592308,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.4913,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.49125958658741414,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.493,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.4529107749530721,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.488,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.500930912030396,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4758,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5232162976910298,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5026,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.46168906757435163,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.5477341431803332,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.4722,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4673055753146816,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4816,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.4730717994534724,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.4838,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.49954410895986867,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.483,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.48786966912440144,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.4584,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.601032846469937,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5117,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.44268137114666484,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.4827,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.5958346333826091,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.4945,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.5124544561394077,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.4804,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.5454121035288885,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.4836,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.6270737302062719,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.4766,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.5293073532241447,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.4937,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.5293952534855034,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.4439,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5197247959964276,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.4705,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.47343436817710627,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.4764,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.4957418091230616,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.4588,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.4220257924429162,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.4788,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.5681909624779194,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.4787,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.45036130365196003,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.493,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.5091449975233299,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.4761,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.41726423820487035,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.4914,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.4912302449780703,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.463,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.4656398907688437,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.4888,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4203180196632588,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.4813,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4490963597127167,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.4795,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.43968936807772596,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5016,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.46222207964828493,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.4899,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.4653321883792481,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.4741,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.44170271482705264,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.4843,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4959409598202799,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5091,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.5205281618408834,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.4986,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4482079011256607,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.4882,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.5037869232493665,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.4946,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4593724008945327,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.4731,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.4304092092835098,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.478,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.4765156413411035,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.4805,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.48456792993703846,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.4781,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.46507083120468967,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.4724,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.4507084750215642,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.4648,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.5191601103575004,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.4792,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4597342553449746,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.4832,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.49282672522690274,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.4908,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.40769625977130575,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.4881,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.49692875336280223,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5202,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.43468244388815275,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.4951,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4980111877498451,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.4757,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.5268404406603876,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.4856,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4440775307195554,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.457,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.545192427762877,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.49,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.41276878575326714,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.4691,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.5336512258856534,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.4737,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4373145849388031,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.4856,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.5021002649070809,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.4862,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.49253550005156793,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.4712,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.48303439825480327,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.4703,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.4809491872033028,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.4765,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.38767048621461775,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.4492,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.5343736402410104,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5201,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.3947619749772024,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5017,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4706899131786748,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.471,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.4380021846219721,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.471,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4920342078321052,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.485,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.4992721226744725,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5018,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.4656628858263658,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.4987,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.49907768058769564,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.4645,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.4330828755506886,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.4916,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.5300148932383634,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5043,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.4984296320342563,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5085,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.45528301776015695,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.4765,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.42612465115090303,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.4755,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4904875238585446,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.4814,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.42221714867148086,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.4908,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.5218182925158537,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.4824,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5228700619114814,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.4603,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.4406882422809787,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.4791,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.48833841573861536,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.4729,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.44929145190015646,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.4564,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4205512803667638,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.4905,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.4707231918519382,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.4844,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.40933714538164323,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.4842,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.5055161734249473,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.4802,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.6515462131943561,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.4698,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.5159789364673705,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.4973,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.4630477537673088,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.4725,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.5356283031620273,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.4833,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4765835748606461,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.4768,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5206952453688716,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.462,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.5382001922317455,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.4713,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.43212838314751745,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.475,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.5408421499155016,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.4924,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.49805349006108224,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.4665,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.46845664943219745,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.4596,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.5343112059550036,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.4996,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.403893333114678,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.4946,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.47873304709947,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.4747,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.4383926473643357,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.4916,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.39879134712796904,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.4603,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4410898148532026,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5041,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4213176638526812,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5031,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.41879066763430084,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.4653,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.47442936269938496,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.4889,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.5095068160302981,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.478,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.5124052584444823,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.4513,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.5028972356519625,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5001,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.41941358415171864,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.4792,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.4299113708353298,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.4835,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.44546815550034813,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.4585,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4348061260301917,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.4865,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.44799049218441844,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.471,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.42751896984629745,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.4839,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4442379165613648,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.4889,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.4336393775720779,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.4793,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.42667065273785637,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.4851,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.43994348736650557,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.48,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.426183758308043,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.48,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.3800764331667417,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.4828,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.4159265310368966,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.4715,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.4538598912757569,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.4822,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.45730520832155436,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.4771,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.4514428069437989,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.467,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.43125645236951626,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.4652,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.4739846144905348,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.496,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.3790221467636159,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.4839,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.8708239588005133,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.4714,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4005457286330933,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.4979,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.4416128717133207,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.4922,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.43876106242069257,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.4882,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.42494579309155706,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.4701,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.382617150525804,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.4507,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.4377551455670754,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.4723,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.4280349433422426,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.4926,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.4076605804189469,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.4718,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.4185065258704524,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.4783,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.4214752363242562,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.4899,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.46574459746397784,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.4581,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.3949364124959398,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.4732,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.46302027558151976,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.4886,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.4022435335210247,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.4744,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.4025787867402323,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.4731,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4435685331722375,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.4746,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.4014953062011836,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.472,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4603705761698549,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.4775,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.6237979543610555,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.36,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.4244200017171846,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3544,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 1.2183089290680131,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3533,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.6350364321373382,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.351,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.6356067659977936,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3617,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5165663433616658,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3316,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.504875667570676,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3253,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.5894196784144857,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.326,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.5497833382071338,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3457,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.657985850251248,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3329,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.5513847391440315,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3374,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.5653102981072471,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3432,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.4608710450513492,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3354,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.5638617823635887,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3594,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.5561727693625654,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3486,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.4496398649234376,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.356,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4854096371327895,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3576,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.45517373263533045,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3359,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4704655688527164,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3547,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.45750552050437854,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3299,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.43408784066718126,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3746,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.49991338531947344,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3388,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.4512790821675922,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3564,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.46705423834689297,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3286,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.4870406570313934,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3422,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.4313766949780084,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3636,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.48652869110516783,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3245,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.481192758497379,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3389,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.42035587002848396,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3223,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.5219426944197832,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3292,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.40102531489247284,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3293,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.5040738309546104,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3404,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.40488057823311385,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3433,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4933918953092265,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3417,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.39987260155976934,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3398,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.4594424071196785,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3566,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.4499032128212761,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3765,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.40821350791866035,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3602,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.5168795198815282,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3677,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.4207413309622277,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3372,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.4433208212015895,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.338,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.40728095038283413,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3357,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.44258528657512974,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3394,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.44717069432433953,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3252,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.43117047096157396,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3456,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.437132358320938,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3363,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.4132034180617907,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3258,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.4475004988999546,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3291,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.42024121163969735,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3374,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.37249644230881457,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.345,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.4346180201339378,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3304,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3886052930082508,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3534,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.4253541044685603,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.344,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.3915487849032251,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3389,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.4392090368879758,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3437,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.40074100192962725,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3481,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.414716278801322,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3426,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.4063837179987421,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3501,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.4493856528426106,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3401,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.4321055903561466,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3419,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.40020981929981964,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.34,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.42314653471010527,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3531,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.40192009141172513,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3516,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.4232161887596023,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3299,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.40815284181252803,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3666,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.40746902225507164,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3581,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.36572480667178836,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3332,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.43052173138919464,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3685,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.39212965395238136,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3442,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.43560111427114967,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3738,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.38151367307664724,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3539,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.43123262680230023,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3784,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.40283463516410634,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3463,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3849753718182534,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.3531,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.4001464646529994,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3578,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.4132900018105115,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.343,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3829223827835311,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3558,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.4116631214294977,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3355,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.4183075896396679,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3318,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.385795684003839,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3253,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.42239056040163864,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.321,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.38328198202568436,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3426,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.45411592047656496,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3442,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.40832058756896905,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3801,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.4014707602594903,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3453,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.39708126119821524,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3378,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.43685135156722177,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3548,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.399738399312973,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3591,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.42426403655674355,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3408,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.38336884042630104,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3312,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.39481810888027263,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3464,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.37547726393632097,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3563,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3926287263483619,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3473,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.40172057845218434,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3473,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3677466328389825,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3545,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3822668837903286,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3441,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.4467199817818504,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3424,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3785719461257869,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3447,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.4017689061548256,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3621,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3811889592090436,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3602,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.38087040352365714,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3392,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.36420136994633834,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3659,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3909665458602439,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3448,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.39919275835003837,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3364,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.39670908831838514,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3486,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3589751653276972,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3484,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.3815660948039021,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3398,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.4092340920209578,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3396,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3800511987140216,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3531,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.40191071169644565,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3514,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3964615679480839,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3555,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.38273672311708185,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3361,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.4221864195185271,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3535,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.40717945219273166,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3408,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.4369653670444074,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3664,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.41990489730339864,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3548,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.3982274783949462,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3313,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.44626865829990775,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3582,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.403249380124603,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3524,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.4099605819286177,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.354,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.44344372544051874,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3592,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.3717463201634941,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3472,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.4050909420795883,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3346,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.4597568415005695,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3599,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3427936099904917,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3402,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.6865167378519901,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.3604,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.4237883245282982,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3652,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.41481736955731646,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3339,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.4183628195531185,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3599,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.40924774149946935,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3479,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.39135118357270804,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3418,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.4573473501479158,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3444,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.4019297878608232,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3699,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.4125963110322634,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3369,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.4178085241574783,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3726,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.47074939269643196,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3373,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.39276731165940326,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3464,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3985988736638046,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3425,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.4102939663088928,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3541,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3661046641234527,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3337,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3896618580756155,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3252,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.37499028707590865,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.337,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.37751640836198525,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3572,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3826495428199723,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.338,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.37150174041946493,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3617,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.39589827052159704,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3525,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.424655963068267,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3532,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.36861519525370423,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3409,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.4088965215123674,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3347,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3840819400726976,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3433,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.37783287354431344,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.35,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.4281997399101576,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3464,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3898363474188964,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3548,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.39434017361307067,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.369,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.43848269258151684,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3404,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3837803097202175,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3634,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.45777355615690185,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3506,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.399020251001507,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3396,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.37899029847436694,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3537,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.43750309194895065,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3406,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3573288448319921,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3627,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.41639670059804923,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3355,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.40059718400934635,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3435,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.40615043784136884,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3388,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.35845326880339656,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3474,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.38425174085066566,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3461,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.38077149867165244,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3449,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.37240622442994425,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3446,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.35272642968165835,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3534,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.3739448699004146,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3473,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.39890978004288696,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3392,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3849603185518759,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3399,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.38150729185800675,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3228,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.37537465271090925,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3273,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.379883863256238,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3512,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.36029626873553827,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3431,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3602847075645701,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3591,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.38758881946645807,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3571,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.380553305079719,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3485,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.38502236661843087,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3403,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.36268372656289904,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3722,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.38376276626202943,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3591,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.35918367421798436,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3645,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3539129838426197,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3431,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.37648260013230345,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3467,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.3589909161364253,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3532,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.3563845269036438,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3537,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.3659280630186045,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3614,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3569236334649886,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3476,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3409076408183191,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3503,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3654349410147738,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3409,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3721093539419411,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3276,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.37204774476256686,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3466,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.36725701768599045,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3184,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.3944947982889772,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3332,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.39366191935898376,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3242,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3619087690117044,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3325,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.35126756160095113,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3595,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.3599645764825714,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3201,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3405943983062492,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3416,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.3672427594094656,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3427,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.3818455556802112,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3588,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.35878747620327506,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3455,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.4111429440671408,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3283,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.3619140380591461,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3511,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3808088536811456,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3564,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.36466858614310593,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3212,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3591022441876105,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3368,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.38110355448509053,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3286,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.35472379077876404,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3463,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.35994184006868096,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3446,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.35276462991092067,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3548,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.3726111074704383,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3453,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.36664272729907166,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3343,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.3675258920186994,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3425,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.37077880207865355,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3581,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.4080551474026723,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3489,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.37123570914328274,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3365,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.36871036024141257,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3475,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.37800158686837226,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3428,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.34733297965394627,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3311,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.35109867802411565,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3576,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3752402098499291,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3342,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.3617419910764231,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3338,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.3474267533623038,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3378,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.33372858829126734,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.346,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.3461371201868121,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3286,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.37536206120838694,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3341,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3692789189620133,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3448,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.37796319598108175,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3292,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.3700291411804605,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3362,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.379751513691956,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3373,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.38017461259744034,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3459,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.38447945203133793,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3361,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.36306621828688235,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3605,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.3575409897600508,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3368,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.3892412258983185,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3211,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.40420484718041855,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3517,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.3679311637999481,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3274,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.3760931811328419,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3295,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.34478851632347174,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3404,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.34791166760264947,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3395,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3628542114413687,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3512,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.3687562047959714,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3394,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3390688525329683,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3644,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.35631136424811455,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3158,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3634439720096346,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3295,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5127892460189657,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2742,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3600242927035702,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2565,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.4651256548777765,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.27,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.5796804501090224,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2623,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.41935536513181604,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2343,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4524707397067875,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.253,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.4996874009683226,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2311,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3815720035626037,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2407,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.4139183846859278,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2511,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.4731904937649414,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2452,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.40224670340443475,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2145,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3571560406532187,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2324,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.41870194515525083,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2349,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.39321852924842116,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2533,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.36800271085076386,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2432,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.41133841660883286,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2171,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3817751509223139,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.232,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.34253919655939297,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2406,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.37197610634338846,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2255,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3544023990926499,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2413,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.3442014961821397,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2354,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.3605591113225392,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2566,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.3198281814143874,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2326,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3759548555663298,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2624,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3123770015606557,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2341,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3359971832493139,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2429,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.33986665640657987,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2326,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.33443453133397055,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2347,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.34731818432362277,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2212,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3089126534548594,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2379,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.33083171493174235,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2571,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3159341746075754,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2479,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.306396361561657,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2567,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.32664326128363186,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2499,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.30663577171969103,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2388,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.316880910624754,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2412,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3200299899792315,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2243,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.32080629784839215,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2389,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.31748410412759315,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.239,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.30409933489890406,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2524,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.3156851144293311,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2503,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.3165086435160779,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2296,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.3100314315340557,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2542,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.3220973778602081,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2464,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.3088807686054856,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2584,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.31037797013606544,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2439,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.32075750070627723,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2441,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.31526754178928057,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.234,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.314545057098824,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2016,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.32088702748589437,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2476,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.3203655880796936,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.227,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.3379855547731413,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2437,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.33909521399565284,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2504,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.3053842944784662,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2447,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.343209756594469,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2361,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.3330405029654128,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2315,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3178051873907614,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2278,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.33509478531272885,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2311,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.333754210391831,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2383,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.31272160902572016,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2219,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.31441444062042273,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2334,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.3125637238951599,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.234,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.3174375882568965,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2383,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.32156337666782736,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2272,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.31594960637075165,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2404,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.32142000358353784,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2325,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.32428713930976344,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2346,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.33574754756383074,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2387,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.3182014689169926,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2244,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.33002150219254633,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2572,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3175192857858185,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2354,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.3382773402711016,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2268,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.3235773305457003,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2352,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.36918286934253947,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2429,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.32772694085199405,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.237,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3432202818913799,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2402,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.35638003355894304,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2321,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.3317134032316959,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2306,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.3301026168696287,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2175,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.34866619288677664,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2439,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.3488272455440678,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2439,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.3106488351859939,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2476,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.3302321980655739,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.221,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.31456780940347134,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2292,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.31633653160878994,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2425,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.29948532635691233,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2364,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.31330017628743306,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2369,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.36174960591091654,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2431,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.32228123489656985,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2171,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.3260893883554254,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2249,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.33821056557456697,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2148,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.31813333805633676,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2595,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3311564620321812,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2544,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.32413571198443136,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2227,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.3174586353863803,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2271,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.29939006258597806,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2512,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.31782975390842083,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2611,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.32900244673011686,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2517,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.31414880811761126,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2618,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.315685710050947,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2446,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.3299120581731276,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.245,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.320582980985963,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2389,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3151601998811778,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2555,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.3405560629482894,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2288,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.300778461683436,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2665,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.3094837361439057,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2436,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.3359180927369983,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2259,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.3136117403994995,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2442,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.29502122372845696,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2338,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3160668468915925,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2367,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.3121429110664388,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2392,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.32941624772281775,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.231,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.3132574043142219,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2278,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.32111061680407504,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2514,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.32694317832391184,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2509,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.3238021926359802,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2439,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.30804857596949553,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2424,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.3475937340859129,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2357,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.311731011204041,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2104,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.3127359911427087,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2498,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.31594849839774164,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2471,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3187853130851829,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2464,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.31519728649366496,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2055,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.3002206267709899,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2371,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.30915349652015467,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2549,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.31585791117789086,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2484,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.3030032414652774,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.24,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.31221954629984183,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2397,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.30208348425857645,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.211,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.29990233732720656,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2299,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.308884621492457,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2279,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.31819145669672505,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2432,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2912952163970857,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2516,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.3394563766130612,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2449,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.31410380374899366,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.19,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.32366918243983644,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.254,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.3252765665996917,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.229,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.32864545177483434,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2436,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.3052299066817629,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2338,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.3172967090706767,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2579,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.3209244570550279,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2129,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.3185372361722814,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2252,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.3021864586602019,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2278,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.31313692407413324,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2115,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.3116289419248438,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2116,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.3215490968316066,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2301,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.30560595917488376,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2508,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.3063786472512692,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2241,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.33173380074490677,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2439,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.3211705537472339,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2247,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.3226918322183599,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2276,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.31595975273876215,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2588,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.32198741925705593,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2306,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.30457137252247307,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2404,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.30773085403597483,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2459,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.30582103262928007,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2312,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.30821793443815826,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2269,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.3101992802856724,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2482,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.3235745517674042,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2265,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.31642775230087117,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2316,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.3168792493028996,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2388,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.3275542145411769,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2567,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.30785289646312447,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2484,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.3117623465353288,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2499,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.30322931410138476,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2647,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.3016011946374802,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2623,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.30403577042866914,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2458,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.2862798184895751,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2433,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.3103453909322648,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2313,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.32766053525350797,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2188,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.32769415736685437,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2501,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.32480069853816557,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2416,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.30380942763067653,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.221,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.30643871398285094,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2534,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.3055940679745494,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2377,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.2928648729180799,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2317,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.2981765626850706,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2335,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.32020270850687227,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.243,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.2897334192560804,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2199,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.3165887331221415,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2224,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.30547824749187563,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2209,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.306131210048712,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2205,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.3093433083110049,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2421,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.2818649805360757,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2614,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2953177614128598,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2195,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.3021898002141087,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2448,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.3068055122456213,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2083,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.29693141741051554,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2204,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.2952501391009479,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2268,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.29999061281615774,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2293,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.28969782527585347,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2229,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.3001144249448089,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2426,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.3086900669153358,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2154,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.3028243254366506,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2504,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.3057646724836491,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2504,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.3042869793532624,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2276,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.31359691830853065,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2256,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.2915687349559122,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2362,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2879411148081762,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2209,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.3038531253492277,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2445,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.30226339013195164,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2592,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.2923087013065385,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2457,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.28764657463838306,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2437,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.3265012967062694,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2482,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.3115736621587423,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2416,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.29990147769621134,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2448,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.29777507791635904,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2367,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2899909072420407,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2424,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.29989071779679377,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.244,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.30082656557031756,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2378,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.30984793661105736,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2447,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.3046206902617553,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2196,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 3.0611350131680175,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2504,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.6188190514952008,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2247,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.30077024545935277,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2476,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.31184122431903444,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2406,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.31608794754940045,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2092,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.31276979915768366,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2295,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.30145824273741406,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2169,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2947257939558641,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2395,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.31027845898539574,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2258,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.30818955047906776,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2455,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.32352221581732354,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2369,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2922836742533046,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2168,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.3529068851895582,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2055,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.3111458139333043,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2454,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.3174372718691167,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2174,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.28800731680321856,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2416,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.3103769645929696,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2331,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.2868209918843912,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2464,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.3176434780593694,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2358,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.29826758598183684,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2574,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.31004511902081267,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2347,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.29706823121676945,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.22,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.29572264526449255,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2465,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.31260152299075816,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2482,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.32268197833377377,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2481,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.30538932714438055,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2216,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.28671300225853263,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2352,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.32217342760557677,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.199,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2860953620828632,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2586,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.31843463552817114,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2275,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2955630865154896,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.228,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.28775472017161846,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2404,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2855030957757979,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.244,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.30278421905141234,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2515,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.3109313713574354,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2222,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.4249348943586145,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.2029,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.3388662680316896,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1832,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.2774563059720763,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1752,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.2572414627387981,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1703,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3879902241315148,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1898,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.40317446646389565,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1739,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3747389196516336,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1861,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.33116196799389874,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1724,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.30821962249569324,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1943,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.334427889383557,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1944,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.30416170661888575,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1634,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.323065509253925,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1783,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.26707353223872754,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1821,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2508648963362181,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1798,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.311240467738595,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1799,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.28291390178869047,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1787,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2947701262508921,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1958,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.27849492403570997,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1918,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.260023990115122,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1578,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.27180129301158723,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1973,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.27396750970196243,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1779,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.2711927775197743,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1667,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.2726879193620165,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1729,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2624673255616413,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1798,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2604551271628027,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1867,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2528048313099701,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.171,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.25403563282761243,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1814,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.24871985767098317,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.1844,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.24093300881134372,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.178,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.2598968958549289,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1789,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.2507323272540277,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1749,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2533295374151932,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1702,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.242876559504609,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1789,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.24345121496125802,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1719,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.2575480950780191,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1808,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.248485637188185,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1657,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.25011708563020507,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1908,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.23688779564134313,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1572,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2442291487869935,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1749,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.2588285490962054,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1937,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.25558991077004,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1722,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.25556235128574667,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1713,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.24837981136711254,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1493,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.24614217973400251,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.19,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2425730995878895,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1842,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.2471787917882262,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1632,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2368964591685685,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1851,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2499532095286193,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2027,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.24278027968995403,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.2006,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2296973936641116,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1984,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.24935534745244323,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1517,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.23802071268159494,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1942,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.2517197023298541,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1768,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.24204583147614084,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1933,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.2594489229629389,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1591,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.24576861328500424,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1831,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2476495791699108,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1695,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.24389999793523776,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.2062,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.22918881009850314,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1785,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.2398596116748201,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1733,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.23438067722116188,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.176,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.230103672335117,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1709,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.23155559428164993,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2025,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2460880421177084,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.189,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.2397705515506026,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1799,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.24528785972113415,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1807,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.24603483097392195,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1627,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.23412226655382964,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2046,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2602629900085939,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.174,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.2537853690881413,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.1949,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.24543837033845053,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1865,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.2350767614335368,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1813,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.2396363437688502,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1632,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.23862342604197717,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1848,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.239697413195575,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1717,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.24850933314521897,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1795,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.23529485075580858,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1959,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.2473698765107055,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1494,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.23962034321991116,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1851,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2411791772048851,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1869,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2514491311726064,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1739,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.25037846544686415,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1957,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.24002012677289775,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1782,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.23713493242528963,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1567,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.24326596471618728,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1904,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.24323096426737634,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1503,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.24511894114461805,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1726,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2379145326679749,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1959,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.2434009247941428,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1859,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.24854408610030776,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1802,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2460214380791878,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1895,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.2558114423816459,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1845,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.2618566805497942,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1964,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.2372363734737944,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1649,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.24131149113522293,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.175,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.23701487092435905,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.156,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.23406470091772985,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1592,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.2460972222173523,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1733,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.2509297148576884,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1585,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.24266436618065215,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1983,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.25018507799989026,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1892,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.23744593214350074,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1872,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.24878002753849082,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1838,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.23780058597050344,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1809,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.24047905950183357,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1942,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.23940825374677488,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1677,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.24578467679297933,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2067,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.2520590238129417,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1837,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.2547273531103631,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1888,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.23911692101307633,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1776,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.24409392143053832,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1625,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.239050307139418,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1814,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2603961840155368,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1603,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.2387121641723726,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1783,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.2501173581087889,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1794,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.23770692374700053,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1722,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2257725088312985,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.172,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.24741739399588636,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1796,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.2331797190420079,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1973,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.2433704010735851,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1912,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.25036865230259076,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.176,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.2483608767543686,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.166,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.23501416623477245,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1743,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.25039602767716007,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1938,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.23365348886805007,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1624,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.24540197494249277,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.2083,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.2387707467525637,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1925,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2410849996155922,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1797,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.24433896201136013,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1868,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2512865049519605,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1785,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.24330358245149572,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1805,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.24569707933982832,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1762,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.23158745516647541,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1605,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.25080581427793786,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1835,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.24407240938522426,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1885,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.23796079418882524,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2051,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.23820183211572019,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1691,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.24220976677581163,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1919,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2487898745522739,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1895,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.24630960453470802,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1602,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.24493053228967349,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1642,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.2536046217719789,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.192,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.244591313315103,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1689,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.23908318826524858,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1784,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.23962124512928298,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1775,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2500933142266786,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1873,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2420586615940292,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1964,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.24486936249773358,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1711,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.24111374153424897,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1657,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.2403126822588927,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1847,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.24747804557665898,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1878,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.2458534073458229,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1656,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.2511056730522594,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1761,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.23250859989898412,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1822,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.2536584379361638,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1931,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.23213388401528745,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1845,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.24544920407378928,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.2036,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.23203402775996884,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1937,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.24755151845288306,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1767,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2436370458939428,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1708,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.22381449729173558,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1487,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.24723326544665286,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.154,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.24627779937474983,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1516,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.250161946750372,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1635,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.23534163891800763,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1838,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.23119174050559493,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.172,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.24024681999294728,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1789,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2351546125471259,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1668,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.25313685061356633,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1723,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.23302640750669412,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1696,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.24744686300781962,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1757,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.24168419832630947,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1788,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.23929750406441982,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1717,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.23573797489685064,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1863,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.24132457781462616,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1701,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.2520899035403124,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.177,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.2577329118249276,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1743,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.24986780491165284,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1934,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2451022683282714,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1809,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.23755737994088816,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1814,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.24242272758959543,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1645,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.23857354946941736,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1682,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.23741311672453028,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1653,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.23630741397132934,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1682,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.23823332610926695,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1571,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.23939152365886962,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1921,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.24256897975130273,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2001,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.24152719500802192,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1658,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.24278471938868096,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1632,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.24708119924081284,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1585,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.2290069019931959,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1783,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.22551375306552557,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1733,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.7582761258329792,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2033,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.24154071420367507,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1926,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.24688673322395424,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1869,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.24068514768709118,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1642,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.23999122013576807,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.165,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.24587443423395317,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1846,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.24246637357545398,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1703,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.23370264535473934,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1822,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.23527211168982148,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1663,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.24245199041457127,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1719,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.25198578106045166,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1711,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.2523353626719471,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.186,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.24491114295909866,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.168,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.24211891227780455,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1777,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.24500775918068726,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1959,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2378093500203398,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1774,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.24735783165547612,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2084,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.23735148161020125,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1471,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.23472186403151332,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.2045,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.24262344953434523,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1806,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.24233647096375266,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1811,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.23794039944046455,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1927,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.2292608120375297,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1717,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.24592846435383156,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.2076,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.23349906555614872,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1797,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.24290606832596584,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.2014,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.24690392434916084,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1866,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.22922333291642882,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1843,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.24113406840308782,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1653,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.23641715025239537,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1708,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.24471728939409787,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1924,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.24076973902127713,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1818,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.24651257296242976,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1875,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.2514254259502178,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1741,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.2348092820961346,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1921,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.24887508483796023,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1912,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.24357423536939724,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1601,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.2660717461271157,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.162,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.25004531681785924,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1803,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.23406051840021794,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1644,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.2472537399801839,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1722,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.2265833001566974,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1795,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.23007598575574023,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1609,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.23431765547430947,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1647,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2383450554615408,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1695,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.24022139722047284,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1545,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.23313233930965716,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1874,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.23407801974009435,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.179,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.24406136417428206,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1669,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.24831531972418808,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1641,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2435632052652087,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1716,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.22577954125013802,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1698,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.23408901099837393,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1638,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.23319031981858537,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1786,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.23372405712267605,
+      "learning_rate": 0.0,
+      "loss": 0.1709,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 2.215016501910438e+18,
+      "train_loss": 0.37683970581906046,
+      "train_runtime": 23348.177,
+      "train_samples_per_second": 6.767,
+      "train_steps_per_second": 0.053
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.215016501910438e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}