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": 5.936988161537248,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.783,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 5.996076167246914,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.7924,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.166521138840724,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8033,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.021533896204463,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.7946,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.691147978934082,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.7988,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.681864055148113,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.8138,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.3491945752682755,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.7808,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 3.852493886476657,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.7377,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.3207582341060595,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7222,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.0972184875231474,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7235,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 1.9770325137274496,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7104,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.8366631162337494,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.6997,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.145094716436648,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7068,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.24502850481025,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.7103,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 4.020104212292287,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.6963,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.0403728007790316,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.6632,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.8669562910796387,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.6592,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.1901878229354548,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6467,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.4305735757007951,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.6362,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.2322379492967015,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6269,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.4479884266056984,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.6137,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.3366185484992361,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.5959,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.0603326183872248,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.5828,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.8273855226231982,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.5708,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.915447569167515,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5738,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.8096676461401566,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.5634,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.7953495021196135,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.568,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7699254854337969,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.5534,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7131983814126424,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5516,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.5850606253928281,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5386,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7585547518003569,
+      "learning_rate": 1e-05,
+      "loss": 0.5486,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7112699866686483,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5428,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.5114624746221124,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5449,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.609318275922263,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5343,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6738689253858802,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5284,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5527146717883072,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.5335,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.5934934695361289,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5346,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5935738627845668,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.524,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5188693949297002,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.5243,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6188072138513282,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5012,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.45760339268498235,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.5125,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5219733997536536,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.5095,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.4305887538647981,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.5007,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.42919856516036387,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5105,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.4902110675943764,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.506,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.43175501357208507,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.5007,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.47582416866960964,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.4941,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.43062262424601444,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.5022,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.4497938861851796,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.4892,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.42532333672415207,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.4909,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.4781269066455577,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.4874,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.4144870487305526,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.4796,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.5222072383539924,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.4927,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.44233268883626997,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.4873,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.539532628623396,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.4887,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6982733239685284,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.4812,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.9372760507398742,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.4989,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.9895278023141167,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.5026,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.9405416817033284,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.4864,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5283898216913656,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.4821,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 1.018094278297557,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.485,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 1.031920897896391,
+      "learning_rate": 2e-05,
+      "loss": 0.4855,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.7799827351449788,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.4667,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 1.1823370221005893,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4804,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.9308394025911111,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.467,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.8809059498831628,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4855,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 1.1179599962058304,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.4798,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.7676173573995732,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.4649,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.9575266850828924,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.4841,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.8723679170079903,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.4664,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.9086507046138652,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4711,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.7590803689158189,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.4797,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.803276267236173,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.4597,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.9713071544711981,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.4675,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.8183791338669142,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.4655,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.8388452848384592,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.4642,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.6392852538974592,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.4567,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.1655692389295744,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.4669,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.1973403105327678,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.4811,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.1688342048498468,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.4426,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.7687704173112182,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.4682,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.158960102279007,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.4603,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 1.008832521297967,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.452,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.1226433173622037,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.4641,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7228632910678581,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.4622,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.8301555928343344,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.4506,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.9498061873167613,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.4586,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 1.162572053041749,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.4646,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.2286490895245892,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.4532,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.7880302183253654,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.4677,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.4642379244358408,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.4646,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.709048048425309,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.4644,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.2582534384973085,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4522,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.8118858760189737,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.4547,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.0346576968664132,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4472,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.9824921232340533,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.4539,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.1372945580958065,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.4564,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.918631792931493,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.4508,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.8777886380448664,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4493,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.2462299165307604,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.4698,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.2263722941340078,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4516,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.9747748029724346,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.4546,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.2353034141491985,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.4555,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.0028838011291752,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4524,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.7731679057911387,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.4535,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.1252063349025268,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.4697,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 2.17059939821357,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.4553,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.6825099301920672,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.4457,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.8711708718409588,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.4606,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.4305898329796916,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4487,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.9936561959042196,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.4646,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.3324506844393733,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.4501,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.9509735557162446,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4529,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.5440489599606881,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4571,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.7927271179521982,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.4514,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.5356622825284794,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4601,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.3284623381042588,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.453,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.1379805615416965,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4512,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.153760063540902,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.4491,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.1096154101369575,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4449,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.0651668173871465,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4505,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.1038182777375045,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.4474,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.3115751588097635,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.4441,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.0581367893514628,
+      "learning_rate": 4e-05,
+      "loss": 0.4719,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.5136734547236972,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.4592,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.9468507999113946,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4451,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.7772496927760864,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4502,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.4100410884870593,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.4463,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.286601153394142,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4372,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.372727398835439,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4423,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.0172065824480652,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4398,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.301017587375291,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.4289,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.94660405404414,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4362,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.323557459105927,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.4478,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.1494866955379843,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4468,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.0056476071196063,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4444,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.2422391476180064,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.4396,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.9797777970503166,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.4416,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 1.089187756412149,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4443,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.9433593409741275,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4418,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 1.0015058407540203,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4391,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 1.100630070003238,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.4435,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.8966003406293152,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4472,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 1.0197103541436865,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.4307,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 1.1647001610589114,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.4391,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.8009280175148623,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4238,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 1.0414360202228428,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.4285,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 1.0245009006590182,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4406,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.9133143486647525,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.4335,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 1.0014796720025223,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4426,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 1.1248863028606013,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4351,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.8194780853042829,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4365,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.7094002573672049,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4391,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.9703352697919309,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.432,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 1.2876731058323576,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4354,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.5628527569978964,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4439,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 1.1401499269830293,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4299,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 1.3473295939630374,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4311,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.7324542711506017,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.441,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 1.7001574771557177,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.4357,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.8227392469636844,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4382,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 2.081061179855467,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4288,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 1.5174894815144322,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.4402,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 1.9071397313433294,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4355,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 1.8345461099785445,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4591,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 1.2773063899051995,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4436,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 1.251815903431452,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.4504,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 1.3155504590828249,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4338,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.7852244467015165,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4254,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 1.2671871077020722,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4458,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.7578508768080944,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4289,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 1.1504823292447752,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4314,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.8840766112598055,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.426,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.8762480946777426,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4242,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.9544572547100111,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4375,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 1.0145786206782936,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4287,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.8059544053940024,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4295,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.8076947285211279,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.4362,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.939879038881327,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4341,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.8294980978354318,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.4375,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.7335140956114578,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4313,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.6832485323176046,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4422,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.6912990236534681,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4305,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.7042689457533394,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.439,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.7578806159440943,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4335,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6253933717896283,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4369,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.7336103024731754,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4223,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5657026167445078,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.4092,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.7505587991518974,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4278,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.801108006652764,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4258,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.6033430271462489,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4297,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.808881242433917,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4235,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.7136329206119416,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4322,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.6093810086777781,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.4234,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.8285114491475551,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4359,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.792346630248819,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4134,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.544564231285453,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4271,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.48832864484954697,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.4269,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.7156749415813134,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4249,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6269196746691706,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4283,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.7511863707715519,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4395,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.748881843822213,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4244,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.6892138542004334,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4185,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5969075170813463,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.4279,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.49770956067264505,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.4272,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.5659461233658062,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4438,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.655583168015274,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.4284,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5024230620777052,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4261,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.7328965385643191,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.4205,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.8487333187025502,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4277,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.9697907604848575,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4309,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 1.1804305950496854,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.4304,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.6555037820609021,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4152,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6172580169687911,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4088,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 1.099988761588085,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4148,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 1.164492443541924,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4264,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.67140800314156,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4248,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 1.048500572256445,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4233,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 1.0041710220010753,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.4146,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.8317783292070139,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.4142,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.9704293115568093,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4232,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.8775238797923833,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.413,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.870869354193287,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4078,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.9557936135416215,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.4116,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.649907291447915,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4228,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.8718700026698442,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.419,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.7125296481834203,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4156,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.637663789257814,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.417,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.7582819845661826,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.4338,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5300332134128337,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4307,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.5584184356666023,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4149,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.6907015941226872,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.4177,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.6574330852927434,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.4313,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.9072793892585868,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.4281,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.7721103970093407,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4182,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.7480191851782908,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4133,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.6613276311153712,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4249,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.4484366989306564,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.4193,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.6700836237599572,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4101,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.7221154129431118,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4255,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.6743345110157175,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4116,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.651440699120805,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.423,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.7035777565917678,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4183,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.7009171424007387,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.4148,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.759655358220741,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4161,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.6173321956254341,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.4049,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6112435752753478,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.3961,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.6086354065440746,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.405,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.5036040165597703,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.3936,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.638609052824837,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3851,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6302609454317573,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.3881,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.7630059684945395,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.3903,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.696126771142918,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3887,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5452222174442946,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.3858,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5751543325946498,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.3803,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5199184332661733,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.382,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.6328018780886879,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.3812,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.6378728792589099,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.387,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5560040826198792,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.3728,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5055901291483774,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.384,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5474412222172641,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3822,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.5254076044351238,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3899,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5549259441945853,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.3952,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.5507069768627864,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.3943,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.5744782388592682,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.3834,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.7486506256682142,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3852,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5611675920926823,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3878,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.5436226054318275,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.3768,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5407596174206603,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.3889,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.4834544713823306,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.3935,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5039669231656807,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.3871,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5107073625972738,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3794,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.6481137410508383,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3811,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.4776514327087673,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.3913,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.41761967583111775,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.377,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.47612371133221837,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3857,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.48084921935014335,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3747,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.4608210283347673,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3877,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.4302658236078595,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3875,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.4333474090784221,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.3761,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.4628935870094825,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3893,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5196075701156119,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3853,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5464614231747582,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3855,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5374412298934265,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3954,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5015531642235208,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3895,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5810714056527506,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3809,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.6518314405040319,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.3801,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.48900719417422744,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3864,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.48149415695368464,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.389,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.562370092151868,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3951,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5039779452535235,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.399,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5620371327479219,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3721,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5266331469546948,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3877,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5185076921634013,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.3816,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.4859292302509097,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.3727,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5929521901562258,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3864,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.5135712911623639,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3836,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.4615459443644337,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3807,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.6457685738495501,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3804,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5085601016275436,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.3866,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.4706984302407798,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.374,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.5465622063039667,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3917,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.6155085882205694,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3855,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.4360406660714867,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3876,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.6009837461048754,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3818,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.6353332881697136,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.3833,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5342186377399533,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.3809,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.6188395437609957,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3831,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.6011386653399526,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.3892,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.5107423503422116,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.3846,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6291937336344684,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3837,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.7757147823049431,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3773,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.5402693280022179,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3809,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.513259100886482,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.389,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5127371338026838,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.3873,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5644618269799667,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3809,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.5568136542945616,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.3782,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5938125324914165,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.3839,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.43384523354547433,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.3901,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.604631802816574,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3878,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.6711072459217812,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3822,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5385434696819692,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.3854,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.642484331314374,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.3854,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5651942493480039,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.3742,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.494057148425093,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.382,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.5357922648161482,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3847,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.7696391499991987,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3858,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.711849221613141,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.3799,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.5543278295493265,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.3839,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5605472793549285,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3863,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.4534736830150365,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.3836,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5466791721511104,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3813,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.4190808738989276,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3864,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.45683986749705147,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3809,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4625437809085494,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.3801,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.6631085011993493,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3771,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.6879711425857173,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.3906,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.528136825551485,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.382,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.3976193429203857,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3764,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.600174537509824,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.3821,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5090492434328723,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.373,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.44456621636897803,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.3879,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.5404385130282798,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.3857,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5434403408639089,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.3847,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.5195467918342456,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.393,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.5444560229915655,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3816,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.507473547758473,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3933,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.5518412939748215,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.3839,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5525348166034687,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.384,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.7380638076816304,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.3831,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.6451765982898222,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.3765,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.4485980376127429,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3826,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.7586483037998232,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.3797,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.7661325258558708,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.3758,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.6651355156573764,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3755,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4507919321419338,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3868,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.7610685675556002,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3837,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.6495113090079314,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.3691,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4066544834418489,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3741,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.6399893082862018,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3845,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.698404438327975,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.3766,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.7100398335838476,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.3767,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.6579340569262963,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.3804,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.5058468499856125,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.3811,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.648882378661219,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.378,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.7880116535299435,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3806,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.5604950495530769,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3775,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.6557543722259643,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3797,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.611762163466818,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3769,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.6152213508891445,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.3792,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.6682283576938187,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3825,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.41212433452134734,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.3743,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.564973656637065,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.3732,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5472944121764735,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.3763,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.5253279630252079,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3849,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.5514740870416396,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.3728,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4832625439202029,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3687,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.6293437144859263,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.3762,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4705383702195288,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.3751,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.38451011539677094,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.372,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5801058138813191,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.379,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4871478527752416,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3873,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.348288847525493,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.3767,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.6076616557640847,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.3794,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.501004159635973,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.384,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.5300247024249137,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.3685,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5412142179044815,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.3826,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.6372859815291444,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.3916,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.7345598898252729,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.3809,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.4542145056512825,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.3708,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.5321114632027713,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.3835,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.6113759687459118,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.4011,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.38806701257058485,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.3729,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.4250919626042128,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3774,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.43732069708295085,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.3891,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.5055407069927802,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.3871,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.4299693742670061,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.3953,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.47633155176698366,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.3789,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.38524623996700125,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3806,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.41815457770134423,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.381,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.40739150630247056,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.3698,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.4206930684671684,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3761,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.47754050054862796,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.3857,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.5892670626884936,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3713,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.46582099567489993,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.3818,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.4538336319932268,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3849,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.4394519134151504,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.3804,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.4737007201115938,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3772,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.4663602886506442,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3819,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.5078554352909548,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3862,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.4029192525113126,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.3791,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4370236927662815,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.3662,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.5484041198068439,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3787,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4755749094664208,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.3843,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.49338576981580107,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.3881,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5325609405088088,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.383,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4846342856214498,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3785,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.4192096293684824,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3749,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.5545494994597345,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3898,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.4420635814011854,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3754,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.4301045767002967,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.367,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.46431993326502813,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.3837,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.45710091748337545,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.383,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.3914326135240673,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.3645,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.37681302993706034,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.383,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4849925716604395,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.376,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.5236723511196318,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3776,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.4212143929662816,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.3794,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4250584645821261,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3769,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.3596557262700219,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.3746,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.4026266142101231,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3859,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.43875597453960063,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3702,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.48971219166564167,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.3749,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.6210229405272382,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3802,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.42716062489207257,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.3772,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.45729752668814727,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.3784,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4731058426936451,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3717,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.44556550705251446,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.3702,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5282419302473247,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.3792,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.45417579810252595,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.381,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.5664796136563356,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3773,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.5210808296425715,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.3855,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.5421976168382501,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.37,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.5793704768974685,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.374,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.3990333501794223,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.3826,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.608057099026591,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.3782,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.5098409712188201,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3793,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.4885058078812183,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.3746,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.4472575733143011,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3865,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.46225116215662243,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3745,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.5649978339134497,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.3761,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.5271957422964947,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.3735,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.535044943258813,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3694,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.5459258588823542,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.3721,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.3769883583894301,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.3781,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.5293577835572923,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.377,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.5915721253785509,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3858,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.43183893861318196,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.3866,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.5375259330055725,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3685,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.49486939018675624,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.3802,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.5269403151200289,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3763,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.6011104885489981,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.3715,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4640595367169262,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3746,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.607054626342973,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3727,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.441124300252764,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.3816,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.6178646131828064,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.3835,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.5208878741748723,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3729,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.4713087520238735,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3732,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.43757046744204703,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3701,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.5211115870154958,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3789,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.5571668516546625,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3635,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.5130383393575652,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.3698,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4141585866167339,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3696,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.512401229240596,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.3957,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.39977290035074864,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3754,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.3708674498946742,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3725,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4433615581425922,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.3778,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.41879906067694195,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.3739,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.49117142588153273,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3711,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.47299897449925143,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3778,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.4765495694559382,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.3774,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.40535325010485324,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3699,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.3955322623773633,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.364,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.42215697966128146,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3698,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3473961310321904,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.3815,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.4363643852316631,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.3776,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.39564397777665616,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.3685,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.40440564073506796,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3762,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.5200585362766521,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.3795,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.4792904747280985,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.3765,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.48784588305087145,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.3782,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4840896807572561,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.367,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.5862349415907299,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.3751,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.49652425704711733,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.3792,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5147969409748607,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3353,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.49577532210649294,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3314,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5070709931794025,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3377,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.5093764894520139,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3339,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.4201407656727427,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3266,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.426360299476346,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3312,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.44822047718815905,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3255,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.3835451652509995,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.323,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.40274792707162604,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3204,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.482296851771605,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3253,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.381030007515268,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3227,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.38587027227550263,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3276,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.4441702744123553,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3272,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.44010470145793573,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3274,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4010997650164395,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3281,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.45227788650273426,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3366,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.5062217876967754,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3277,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.35478445891869176,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3304,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4744305501743394,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3365,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.37116967529742756,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3259,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.39944237102088864,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.326,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.5026844843750141,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3331,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.3600987384922928,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3255,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.4601551689797101,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3305,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.3677292258182786,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3239,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.5003415328462757,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3248,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.42653529623441283,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3245,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.3824564755488836,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.325,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.45020933913486716,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3342,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.35540865669916333,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3279,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.4517072712108219,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3312,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.45180057442238863,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3338,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4068632347318826,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3255,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.48562733336049607,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.324,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.33290183354232306,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3174,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.4680904964011835,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3299,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.40465259212260624,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.333,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.33121177562649484,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3248,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.45948160287213535,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3338,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.35774912477929954,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3317,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.443109958574887,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3358,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.35475625403060396,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3333,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.41720705559679727,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3355,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.3450381058054706,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3221,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.34728083612746563,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3255,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.45918048681966916,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3245,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.36518182030834345,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3239,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.47053647362216616,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3275,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.39658337427052215,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3352,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.40216911852817744,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3313,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.41519564679416127,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.331,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3935484900563437,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3341,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.4158185822118441,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3259,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.3158908294156565,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3262,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.3503744556172808,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3329,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4138862370150515,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3318,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.35000121075222124,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3344,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.35600653062571225,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3283,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.43748586702578673,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3185,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.4067533251503402,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3326,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.4096500810413246,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3427,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.4421657207798782,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.329,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.37099366069803785,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3254,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.4998294293778229,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3296,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.424692061510652,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3266,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3962729428538522,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.332,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.44715187410451085,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3371,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.3951945469642002,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3327,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.34549257856090676,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3316,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.43044056202863407,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3277,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.3661915208852815,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3258,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.31431360141018705,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.322,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.31519146831495615,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3322,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.2999925387049272,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.3312,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3373910953346095,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.33,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.32395297254350025,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.329,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.36621441283575346,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3264,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.326717311144412,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3298,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3594774838519097,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3319,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4036711539296406,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3369,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.36095420220744584,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3364,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.3475786242518193,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3269,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3856188113893497,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3364,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3405033636827431,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3324,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.37558539480121933,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3332,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.36229020479114254,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3303,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.33576733038363077,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.339,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3874295589689305,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3336,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.30470123221891726,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3292,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3832058101507129,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3276,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3624746152401595,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3234,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.37632464498186136,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3295,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3095686251758561,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3276,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.39114881062879114,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3314,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.347804475413578,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3258,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3705221430223693,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3242,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.30232002836261335,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3269,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3931644783716169,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3378,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3797136233670582,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3243,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.4370922841120496,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3414,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.28092471910491146,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3331,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.4025431098596012,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3285,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.30560131579794203,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3267,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.36387812319507135,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3334,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.37795861877200376,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3347,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3147809615973309,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3272,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.44805517366784353,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3332,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.3659661218587533,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.336,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.37745061516703543,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.337,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3247948375683391,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3349,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.38705952744990313,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3393,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.28969750070325456,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3258,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3668481412677029,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3275,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.33737134665348356,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3375,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.34835616711738,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3334,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.35482797308618647,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.324,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.3551895426789627,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3363,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.4182336662856277,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3344,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.34815823418966746,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3347,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.3856481948600094,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.327,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.323123460171064,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3281,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.4035860639911447,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3261,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.32262477249638044,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3256,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.3843646423475993,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3248,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3492551674274208,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3393,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.3488231631495367,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.323,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3614915500810095,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.332,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.36115498836284543,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3321,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3158731360805781,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3308,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3405269606372706,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.329,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.34256007798188437,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3294,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.30825498669980317,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3296,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.3165485890375905,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3237,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.3308460348288788,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3347,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3527973372931797,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3324,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.2841770373365786,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3363,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.309644162805971,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3295,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3150983803040254,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3342,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.367481716828116,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3347,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3304711037441348,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3349,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.290105367090217,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3283,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.3436204123740427,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3277,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3248709345174395,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3336,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3596248230651265,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.318,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.3008679782360721,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3314,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3658081583270334,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3317,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3127263243097063,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3384,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.2915624202480451,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3288,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.2935149616809104,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3349,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.2856601872727709,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3279,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3053563264393908,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3363,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.33740317943120035,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3346,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.350255799311788,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3392,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.29565556845721186,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3318,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.29761050328660704,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.323,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.36649220546397576,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3261,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3317494067481808,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3289,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.33864224642848284,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3308,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.307606840359857,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3296,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.31357117111554467,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3281,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.2846489421182154,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3314,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3376730929658264,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3228,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.28329215169420957,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3353,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.34516416118263743,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3314,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.2977521411792869,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3285,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.2831367691852573,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3323,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3360018867682133,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3377,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.27902122714109723,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3338,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.316842332732184,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3313,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.3385460718648581,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3394,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.28414172933729337,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3324,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3932160683450083,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3274,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.27337349028175556,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.319,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.4605782440683997,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.328,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.26708639264612427,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3311,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.4006147661509594,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3301,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.2756464667918102,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3291,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.4326083875593862,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3344,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.2750077131234643,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3342,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.40314664415489326,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3336,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.30111205893200027,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3322,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.2975952535745934,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3182,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.29617398691570107,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3272,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.32621138978883185,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3334,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.2889265878401286,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3342,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.33380894453870624,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3249,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.2910128598083235,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3284,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.3167363732440571,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3301,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3193914682830037,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3344,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3161653145858243,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3343,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.30404969017206324,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3294,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3430482600460599,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3271,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.28325737844511334,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3253,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.316127885675174,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3321,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.2722931307165916,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3304,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.2949158792005035,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3249,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.2843183101907043,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3334,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.2872786083629862,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3232,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.2889314260307538,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3376,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.25246709465092754,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3314,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.2864998797781692,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3221,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.3085866279836148,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3429,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.32658436435295374,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3341,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.26796917595228903,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.33,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.36407345684595666,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3309,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.34009690347039345,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3281,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.2557603557529842,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3331,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.32807053782829754,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3367,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.2629977839060835,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3311,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.3738619660334726,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3354,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.29447340850842874,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3367,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.3630454457495915,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3301,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.323452823354321,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3243,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.3493122011073849,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3207,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.29265013075793644,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3273,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3101465713732119,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3249,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.29655035519320566,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3283,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.2869620688035963,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3344,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.2820284020016837,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3307,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.27531065492247103,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3372,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.30889411672950773,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3374,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3066231967870878,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3263,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.284143331999225,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3391,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.35330906919043825,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.328,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.27230485289510387,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3276,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.32387427727385504,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3352,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.2954745292467968,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3309,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3019314343328286,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3318,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.2996787047467097,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3364,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.2583701485995984,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3282,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.31551679560313073,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3276,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.26263044852975587,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3346,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.2961302501059733,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3279,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.2743908608527755,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3248,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.26128006610659776,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3305,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.2729038686929604,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3262,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.29763727238756305,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3294,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3140613581591309,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3351,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.3050016431942175,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3317,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.26607120453458627,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3222,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.35505384352930736,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3368,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.29894371914700146,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3374,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.29366730665627744,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3222,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2936575725828362,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3304,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.29486939245229915,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3182,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.2630031855876463,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3174,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2695163339729091,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3226,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.36971095758730843,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2934,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.30988291172706117,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2815,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.4552346065179451,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2901,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.36922753988217755,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2837,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.38415633784076475,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2908,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4617726285717917,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2864,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3552317578213674,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2873,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.40605334675689936,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2851,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.3611258144595335,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2797,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.3030376435287826,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2882,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.35176722924111803,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2873,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.2967007799629949,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2837,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3521915089302685,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2766,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3187768596755206,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2853,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3038013997539601,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2799,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.29734220930940564,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2802,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.2880776463154963,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2749,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.2768368354890299,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2806,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3152744899650864,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2851,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.24351024592480808,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2825,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.2881465773200234,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2891,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.27516026542813143,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2865,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.28097227659543533,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2903,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.27639518805408103,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2855,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.2508771193231363,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2738,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.24257806481124478,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2829,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.246890682978764,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2842,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.2602302361758383,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2874,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.24495481699772234,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2849,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.2554884081132603,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2826,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.2613067500815135,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2873,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.24039751604710624,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2788,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.25649795734369196,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2803,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.23967453166071917,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2825,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.23306331555598528,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2846,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.2716603485716035,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2867,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.24092382760684555,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2842,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.2714545489432854,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2898,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.2445316720288755,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2797,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.2723334419628404,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2797,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.2544430937274045,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2879,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.2303692283934811,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2843,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.27009994699375844,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2894,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.24852510053989096,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2817,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.24121736596183874,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2796,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.24163892640862866,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2846,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2533214581786637,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2856,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.24879472448450535,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2831,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.25648511798900275,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.28,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.251506102176315,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2805,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.2720641410773265,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2871,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.2518806840584813,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2851,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.26057261455527997,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2796,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.26867081607481913,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2844,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.2913580307442013,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.287,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.23920279650798768,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2869,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.26011191225428026,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2912,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.25878982975726067,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2846,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.28299908623935355,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2772,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.21955402889556344,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2774,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.25806514849631157,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2835,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.25415845659000486,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2761,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2610398118286776,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2857,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.21860322904292576,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2791,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.2506486543930195,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2815,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.22757358484805126,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2884,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2810623065192524,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2819,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.24415924717354753,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2844,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.2698422802625152,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.284,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.24406306487257748,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2804,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3011194914774119,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2807,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.24150889950665502,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2886,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.29607321264731823,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2838,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.2916741191917139,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2805,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.24835909474568243,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2757,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.2673354809977882,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2812,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.26408531939483376,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2794,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.2351852792770479,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2789,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.2761548986230799,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2891,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.24169562120448287,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2854,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.24506791410323076,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2815,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.2587889457942651,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2846,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.22834396474666338,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2768,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.23722785173482266,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2751,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.2195851705463082,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2704,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.23422646901063193,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.288,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.23932403650294232,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2863,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.24456460287441228,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2816,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.2612055429578305,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2857,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.23597491682026986,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.282,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.27116385559527884,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2787,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.21649951106596213,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2764,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.23583650735216444,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2822,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.261855223632437,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2844,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2388774638975177,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2924,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2712131969956795,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2804,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.225034311770614,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2843,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.23002292169547053,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2857,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.25022257264144615,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2797,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.244198811092421,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2847,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2376964540452606,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2766,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.21717476369433739,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2829,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.22102967484396246,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2728,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.24986368144161067,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2867,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.22311080441907488,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2853,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.23045030522732582,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.277,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.2212316449993143,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2903,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.20757111665295844,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2853,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.23499082410224295,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2791,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.20657450129572608,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2827,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.21107330828972673,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2802,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2437488012613818,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2914,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.22840052156879295,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2844,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.2215357309622726,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2826,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.2187659100927803,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2807,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.21029806035406495,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2844,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.21961230497857853,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2757,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.22127028372094731,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2888,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.21952986169209027,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2863,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.19898544502853382,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2836,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.2148342566974255,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2863,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.22262955103350981,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2833,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.19428498646112993,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2747,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.2085178097315452,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2957,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.21847904425554354,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2822,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.21208699162317773,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2868,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.20167062642606814,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2841,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.20497686588641142,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2826,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.20947127082184566,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2833,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.22169374941216427,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.28,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.2271950644930737,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2821,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.22873269404722293,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2797,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.21784538768615522,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.292,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2165473270840735,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2868,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.22483684512598015,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2793,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.23520159328946658,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2777,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.21874284831126387,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2882,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.20633624322356311,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2791,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.24490629502409156,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2828,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.2316495145217077,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2815,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.20395835027042242,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.294,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.24465595922449662,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2849,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.223011026739393,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2811,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2165289706645141,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2842,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.21523244635751726,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2809,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2111684463646235,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2868,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.22850868354943635,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2804,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.19161436560068948,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2823,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.22222157294327816,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2797,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2236420717663059,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2803,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.18809112247164314,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2817,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.2407845769049102,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2726,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.20821423413109416,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2725,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.21036925425001868,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2857,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.24350422807475292,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.286,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.20864097580909494,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2842,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.20693441181507033,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2917,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.23079859947782624,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2828,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.20863697346582552,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2758,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.20490899450596015,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.277,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.20524293647538022,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2799,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.20098151258047858,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2772,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.21294796673500643,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2844,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.2123703672135084,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2811,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.19490623616643687,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2849,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.22639423863954508,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2838,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.20343633242444192,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2824,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.20156194084646506,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2826,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.20528452042614437,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2857,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.20406822419500803,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2796,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.20663503204187628,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2876,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.20941662848694695,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2896,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.21176788559678966,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2844,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.20224834481834125,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2819,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.2034664248590997,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.288,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.1931544721220232,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2728,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.21560836718090096,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2877,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.2066698606524316,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2808,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.18955459754703566,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2716,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.1990011834079639,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2835,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2086072217605244,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2865,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.2124669544102974,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2923,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.20534626340544548,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2851,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.21319368126697466,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2855,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.21478710984526597,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2854,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.20287825544525528,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.287,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.21988651963327238,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.278,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.20749433736492076,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2749,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.1957919634929649,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2824,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.20837153264353547,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2853,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.20050022907982037,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.289,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2025866737703353,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2797,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.20686667169405346,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2919,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.21132118092465904,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2792,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.20019585317794397,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2876,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.19843342927028196,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2842,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.2339205154587219,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2817,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.20789566850601965,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2819,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.21890045782044476,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.278,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.22001024323156665,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2806,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.212485646476411,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2824,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.22407337712719055,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2856,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.21814315725229458,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2733,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.23291659734420472,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2795,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.22801308532571138,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2858,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.19962881864440943,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2861,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.20561930115611815,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2818,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2170057918620523,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2663,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.21876921375618208,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.285,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.20321087242825023,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2799,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.21463820574092216,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2758,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.21993105406850677,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2833,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.21171872673441613,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2893,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.17989161311676863,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2754,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.2266137974420987,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2849,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.18927655211989655,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2819,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.1923953830063835,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2822,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.19095372282782686,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2831,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.1983074182860253,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2817,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.1866600692991791,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2771,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.19112700945474273,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2802,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.20150553038203803,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2838,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.18831571280955142,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2745,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.18858880853558388,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2828,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.1918001465749445,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2877,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.19560275691348045,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2779,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.18535343165741203,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2758,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.1856960594120172,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2815,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.20018816904048162,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2864,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.1982881468540929,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2796,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.19467610193530846,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2834,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.19075781916012338,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2818,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.19773912777308025,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2772,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.18220514677124516,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2863,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.18394885807116293,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2792,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.18848005496287312,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2873,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.1921194324299326,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2779,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.1804413810362821,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2792,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.1923330081376227,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2848,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.19127968512539187,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2788,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.19275391846108744,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2904,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.19563639200520577,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2826,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.19080438157655388,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2805,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.19862112076328275,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2872,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.1892388223533194,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2811,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.1995109184858583,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2809,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.20190841896517173,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2844,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3151091491141735,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.2491,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.25236859265431105,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.2483,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.22119008598118373,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.2527,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.2704928737251955,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.2553,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.30128487194430903,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.2478,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.23972337593716894,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.2447,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.27168643300266254,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.2589,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.2612118628822639,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.2407,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.2344275921406667,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.2461,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.23563851096487967,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.2465,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.25769238530874383,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.2492,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.25740270807446347,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.2479,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.23260927061446943,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.2495,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2300145828044243,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.2413,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.2459309659360881,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.2498,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.24117085924835718,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.2471,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.21636714090273623,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.2432,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2203061049245313,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.2409,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2246181068532877,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.2443,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.2155757571489107,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2513,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2124900171881119,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.2456,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.20958436618489215,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.2502,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.21048639593901908,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.2408,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.20952779808025998,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.2479,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.19551026311928066,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.2424,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.19964402775490428,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.2571,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.20305213406560316,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.2508,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.20071005309338943,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2483,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.19846842978703214,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.2528,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.19462410294598384,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.2496,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.2011278102264566,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.2459,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.20461474155361342,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.2446,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.2169948682375836,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.2457,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.191456885341802,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.2497,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.18714333237117345,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.2486,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.18277411894403778,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.2413,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.20492943293332858,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.2473,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.1833446978090672,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.2473,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.1927758616052744,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.2403,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.19190182394029684,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.2466,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.18907715659921168,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.2477,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.18922737625268884,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2445,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.2037862420265145,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.2461,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.19059499089472717,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.2509,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.19380536602359594,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.2511,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.19638316943811437,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.2497,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.1822075790350845,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.2516,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.17802548324057677,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2402,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.18903189644835908,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.2469,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.1877768783419464,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.2491,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.17894821912134184,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.2435,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.18672431932574649,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.2527,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.17764403366797918,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2417,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.17611756244252325,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.245,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.17877743961601625,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.2471,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.1870473879037109,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.2443,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.1786526138108315,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.244,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.1817108959786566,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.2467,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.18272577807988244,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.2484,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.1727895297099349,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.2461,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.18567663525428665,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.2514,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.1838059934586139,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.2471,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.1869112320001349,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2446,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.17083359423178837,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.2519,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.17699642948756777,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.2465,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.188789267576077,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.2494,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.1903541333784071,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2499,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.18211742304873627,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2449,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.1817652280541179,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.255,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.1795894266762573,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2527,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.1878476756583302,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.25,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.18153017653719394,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.2518,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.18783120247727603,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.2516,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.1853369228722584,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.2378,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.17538496633932674,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.2455,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.176834219156733,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.2484,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.18456574491055433,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.2494,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.18100826999148845,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.2479,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.19151423236252266,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.255,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.17751871345388506,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.249,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.17681771407899338,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2439,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.17943462321244677,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.2485,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.17459417303906488,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.2469,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.17706846024489367,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.2511,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.17390655086814222,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.2416,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.16670793509983478,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.2383,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.16770408768441955,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.2461,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.18080301441030192,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.2462,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.17348780846334616,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.259,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.1772516302981245,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.2561,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.16657971882992711,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.2446,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.17453397993582864,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.2472,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.16746909167826785,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.2511,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.17545091341912525,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.2471,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.1809496169826255,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.2441,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.1749486639208794,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.2397,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.17800380486726475,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.2508,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.17355487164781797,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.2493,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.176256992257977,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.2503,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.1746518050622904,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.2445,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.1799375350255879,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.2496,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.1775652913018096,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.2451,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.1792721810767745,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.2449,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.18615682229508412,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.249,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.1768290328947494,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.2531,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.17874967039716347,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.2517,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.17907210647304195,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2519,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.17946210316122413,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.2474,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.17715006849842593,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.2476,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.18091476753319846,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.2515,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.17122516971121296,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.2453,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.17183733721941694,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.2584,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.17740351861421794,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.2475,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.17497693983099,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.2479,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.17208300729141274,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.2458,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.16944430845967257,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.2535,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.16854501001612324,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.2455,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.17567851630248046,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.2458,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.1725844674371029,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.2485,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.1705795567800365,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.2443,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.17273424179761918,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.2449,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.17445295971048563,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.2464,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.17746818194050912,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.256,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.1713703821350256,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.2515,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.1725871155401383,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.2465,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.16791168180046612,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.2441,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.1733322183393252,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.2457,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.17294256359035512,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2487,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.17805174342397728,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.2472,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.16681677897498212,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2431,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.16744241646792615,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.25,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.17312086550489136,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.2467,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.17058310572793253,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.2514,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.1705055417789276,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.2454,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.1721475432030363,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.241,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.16573797129821927,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.245,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.16900921313632777,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.2484,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.17444430477355932,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.2513,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.16932444328656138,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2492,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.17096793012894346,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.2501,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.1677323997322114,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.2415,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.1668321431547507,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.2409,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.16643457781852183,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.2508,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.16914269201475274,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.2458,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.16515995639494452,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2395,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.1679006012943442,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.247,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.16814384739947758,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.2519,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.16496998700720064,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.2517,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.16271257075546303,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.2455,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.16642664912016883,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.248,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.17134740169512577,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.2519,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.16570763233481853,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.2447,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.17056014802520972,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.2473,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.1739353375639988,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.2535,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.1692843985620213,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.2419,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.16334414222832633,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.2554,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.16258971070354117,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.2378,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.16973517348718944,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.2434,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.16564312336161824,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.2426,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.17378697027351805,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.2558,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.17565026738010495,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.2475,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.1729025356748644,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.2554,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.17067340364604258,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.248,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.16795591444944355,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2453,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.16493779388912952,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.2521,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.16605595511141888,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.2484,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.17416321846249044,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.2479,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.16469434312081963,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.2429,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.16462364752719238,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2426,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.17009206002373584,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.2438,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.16897292321952467,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.2533,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.16860248713077636,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.2477,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.16311229926836884,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.2375,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.16995610588430501,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.2463,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.16645980873112085,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.2498,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.16143923670617477,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.2468,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.1752856110343769,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.2445,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.168060387780208,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.2469,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.16652330059068582,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.2402,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.16496787018682885,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.2432,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.18613637503198457,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.2537,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.16483448905332476,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.2444,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.16200436858701184,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.2442,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.1699016818371283,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.2478,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.16366112612691994,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.2545,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.16537537365059984,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2461,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.15874638376146008,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2482,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.16314018057428015,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.2397,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.16790488305884524,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.2476,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.16858403896723154,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.2557,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.16467651441717848,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.2421,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.16021720216283525,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.2377,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.1720954601690301,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2466,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.1684004440692331,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.2503,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.17117551505707548,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2525,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.1650888714809417,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.2485,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.16752492125842697,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.2476,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.17097495399373297,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.2498,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.17028556032934522,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.2534,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.1707685135859241,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.2508,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.1617040159851817,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.2472,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.16910696426920313,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2458,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.1636852754148427,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.25,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.16587559085688897,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.2414,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.16588111931071187,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.2389,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.16727001071211178,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.248,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.1695227636998959,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.2501,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.16600373644493754,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.2457,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.15745433480077092,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2371,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.16655563331373505,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.2544,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.1677122642223484,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.2504,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.16370050463459196,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.2496,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.1692814918337184,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.2458,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.1704217140491643,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.2476,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.15645932817755442,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.244,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.1682400728346171,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.2525,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.1632173136326492,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.2477,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.164824309501138,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.2484,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.16447746814034706,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.2462,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.16271449652457304,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.2419,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.16456264783179803,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2427,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.16388947771961543,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2443,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.16137756899417352,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.2412,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.16412127214179237,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.244,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.1669225248183675,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.257,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.16810262840542972,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2494,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.1692112775391159,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.252,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.1677415633828858,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.2521,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.1679580571345036,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.2482,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.1688353149265896,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.2556,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.16645910001907369,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.2467,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.1652279512877237,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.2461,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.16683217492916938,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.2518,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.17117390580916717,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.2508,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.15953132607248988,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.2395,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.16265424620459812,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2474,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.16116151341595586,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.2417,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.17055164996238095,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2509,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.17133299513440767,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.2459,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.16834587013549335,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.2502,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.16427312514853165,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.2439,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.1625085784705975,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.2418,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.16127984627440436,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.2436,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.1621244675710595,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.247,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.16703027653507052,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.2513,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.16637650452033112,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.2423,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.17029567786115393,
+      "learning_rate": 0.0,
+      "loss": 0.2476,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 5.015512062164468e+18,
+      "train_loss": 0.34359321679904875,
+      "train_runtime": 58237.925,
+      "train_samples_per_second": 2.713,
+      "train_steps_per_second": 0.021
+    }
+  ],
+  "logging_steps": 1.0,
+  "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": 5.015512062164468e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}