diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 6.592596129242112,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0394,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.506976366151957,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0372,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.527842999585323,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0507,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.42531415641361,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0386,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.064388910647478,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0259,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.85279207430575,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0186,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.591809614464636,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.9952,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.3435811164728415,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9909,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.8962399125672578,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9503,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.605404259361814,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9527,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.529814018032564,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9388,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.916232113449286,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9135,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.290394809053606,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9225,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.056758462355701,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9306,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.7259191303395305,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.909,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.8066872822013025,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.868,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.4858085914307093,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8674,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.065561752078362,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8491,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.5324483322424303,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8212,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.2020251796866415,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8211,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.069253265942003,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.7964,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.1744139569444436,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.7962,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.0659616729246753,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7902,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9398742805507136,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7668,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.8775610133355134,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7614,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.8523436300969153,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7636,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.8082225400321855,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.743,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7895423677242899,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7596,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7938693012448301,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7266,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.7443858624964176,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.743,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7302887703122535,
+      "learning_rate": 1e-05,
+      "loss": 0.7303,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7167369701732684,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7172,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6799909789787891,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7323,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7368406064837565,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7346,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6538872307825125,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7207,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.6210475807803189,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.716,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.632937898748455,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7251,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6288348345610689,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7031,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5800468401767186,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.6937,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.5706009472176787,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.6928,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6651991112065846,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7064,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5447489514258648,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.6992,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.6615770435684957,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.6843,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.6209541280096642,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.6887,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.4711470064071446,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.6808,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5376462825437299,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6882,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5724251471490316,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6601,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5701862168263178,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6637,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5444931141129594,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6715,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.716216188614575,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6841,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6367422904834984,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6852,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5213203693918277,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6687,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.5882025162765521,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6764,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.6319328663564162,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.6647,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.700231629913372,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6741,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6624891992032664,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6528,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.7329293040593617,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.686,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.8410488680681081,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6707,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.6740233098449517,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.6473,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5617768776728258,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6562,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.6327551530663185,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6444,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.7048604599830605,
+      "learning_rate": 2e-05,
+      "loss": 0.6596,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.7078829170557731,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6599,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.5693521858605396,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6479,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.6098273842675217,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6597,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.6902534519922352,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6458,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.7510932711859679,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6487,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 1.0548684770196421,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6387,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 1.0933141293582218,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6415,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.6353969672266749,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6575,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.7580341181786413,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6417,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.9424658450358209,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6332,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 1.0387729219410378,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6361,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.9165553471010629,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6357,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.9462548804298766,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6323,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.05960213749378,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6455,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.5281700590099836,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6224,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.7003044662258384,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6506,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.7017937765791307,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6372,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.7460518813359621,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6416,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 1.9275050030683885,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6431,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.1651171166382053,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.631,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 2.0989267600030455,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6437,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.8100718209583189,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6265,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.3318179408631003,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6366,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.2940955978029594,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6228,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.50352512346348,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6314,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 1.026516220830851,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6454,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.6637830584470399,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6305,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 1.381287112281501,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.641,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.4334939234724224,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6276,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.3832418507092428,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6285,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.897333945049004,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6497,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.239905934163425,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6168,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.9220385973502622,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.613,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.156302398966173,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.625,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.203490044998432,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6361,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.0689495340402513,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6174,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.1259312315906123,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.633,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.1928668673047242,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6472,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.9924108550626478,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6218,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.0878098003237038,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6324,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.1897978131539033,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6253,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.0034069875315113,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6138,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.0200953344270325,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6275,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.2965871400951074,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6194,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.4826475177915457,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6342,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.6203544722480057,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6155,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.4671804685687997,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6269,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.0165503152648814,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6359,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.2560249639112702,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6234,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.2386173942306375,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6036,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.0000390745915282,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6261,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.7522499510533015,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6239,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.5333503464522513,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6242,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.0698499586959718,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6142,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.9190426240150593,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6209,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.9655395568438445,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.599,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.5981885167729812,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6125,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.8575868364315055,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6106,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 2.088463050533532,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6218,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.3250796364524067,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.625,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.9711919158840414,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6376,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 2.1550490070741213,
+      "learning_rate": 4e-05,
+      "loss": 0.6344,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.051044675560032,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6252,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.4997687351385554,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6197,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.166954811559951,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.622,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.1746782955486104,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6085,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.153506239361951,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6299,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.8563008485775347,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6173,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.0867358535291751,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6143,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.9035686884042523,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.615,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.9933962434986701,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6273,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.8046352151979452,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6151,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.9686892468276675,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.6037,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.873691345359781,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6211,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.9729417945888627,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6138,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.8827360911598304,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.5987,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.9111746675608422,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6192,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 1.1318096676405636,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6228,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.9840056935980274,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.5921,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 1.2663781434828514,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.6145,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.7049998831432098,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6221,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 1.4889534138073188,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6293,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.8438745428500058,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.6147,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 1.281514433116222,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6101,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 1.0863245367184637,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.613,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.8089418714351836,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6084,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.7512825241759263,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6011,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.7509437159218592,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.5983,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.6309398688733749,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6141,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7079432335707451,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6008,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.6915257375578997,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6114,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.745230461111207,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6074,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.821971586683642,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6212,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.8643372455046772,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6099,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.7607334470836382,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6099,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.8700143678313443,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.5937,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.8861256821475484,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6001,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.6464384470716845,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.5996,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.49368789000471747,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.5946,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.7050746195382298,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.595,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.9148647499727327,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.598,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 1.0225895859477532,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6039,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.71967312363913,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.5943,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6354524707460004,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6024,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.762470238906043,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6027,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.8125550738794727,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6003,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.7879580830172848,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6051,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.8745236911213279,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.5956,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6316892437730401,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.5961,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.4824722228816982,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.5933,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.6136010102898714,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.5915,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.7521936333642422,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.5992,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5934902002518723,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.5838,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.4896635333463287,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.5923,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5953957560774809,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.5974,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.597232141340009,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.5995,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.6360998522601392,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.588,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6758359596604998,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.5801,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.7224457987354092,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.5984,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.8546783098740593,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6075,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5702436753677476,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.5918,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.4462189780709065,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.6082,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.6086616457535555,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.5925,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6639553287710187,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.604,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5106914173206332,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.5882,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.40428708984344114,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.5889,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5190015673538136,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6055,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.5861526851069244,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.5959,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.45700545154555056,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.5892,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5393467978653784,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.5736,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5460180005979386,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6023,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.682864797740952,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.5984,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5726394587093792,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6087,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.5266321195914921,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.5926,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.5671129269715227,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6064,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6529261318342878,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.5749,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.48882772374461597,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.592,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.42063935327563995,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.5983,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5412734721790223,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.5943,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5325621015732706,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.5803,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5183798712169456,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.5923,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.41863571257660487,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.5838,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5383210860912557,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.5838,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.49590524406538095,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.5839,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.581125012922291,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.5945,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.6508609350441835,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.601,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.5844870069300858,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.587,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.45559675303588504,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.589,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5923845721203203,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.5937,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.8384942721459583,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.5902,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.7292337605392146,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.5791,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.53121864836113,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.5681,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.8343238306870454,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.588,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.707910325117411,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.5917,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5029501251222129,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.5783,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.685979782995521,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.582,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.5558148486911247,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.5958,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.592268444135982,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.5899,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5289461471987181,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.5807,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5781796164731441,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.5924,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.5525920612134495,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.5734,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5282492694263565,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.5914,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.6080142809882331,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5721,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.48173906732890476,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.5945,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.4733250492754633,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.5761,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.5242962876609238,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.5921,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.4672426667397661,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.591,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5120455198106444,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.5879,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.5496982898246724,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.5823,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.48809450885789774,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.5898,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.5746473415838541,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.5731,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.5139658751916213,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.5948,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.5425375945336598,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.5761,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.596390970874216,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5977,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.49287461274418454,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.5631,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.563544393531918,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.594,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.53270375233901,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5962,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.5626343570372748,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.5874,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.6302223131514441,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.5861,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.6363548552236378,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.5981,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.6067824670002451,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.5891,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5942220383217163,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.5694,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.6085519294465225,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.5806,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5951367633718576,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.5779,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.45713789737110155,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.5952,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5607552693151918,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5422,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.6804519169319554,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5468,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.8310967300128222,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5436,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.7961026255662674,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5466,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.5346333771572079,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5297,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.7626990740428511,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5293,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5044887300522654,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5472,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5563614942061913,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5204,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.7175676446301038,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5477,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.4694912843018092,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5395,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.6099291837805586,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5349,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5217931558612395,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5342,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.41122946839127644,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5396,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5230753747853975,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5367,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.40045706634267136,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5402,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.4599005868799554,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5287,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6198055143613357,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.551,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.7914395476708819,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5526,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.6467089351202852,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5342,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.4103453146664098,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5389,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.554014818586112,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.537,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5425088007653494,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5336,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.49606018242294886,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5506,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5914457029103453,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5458,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5796859548669658,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5317,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.4077447052007872,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5316,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.48923915914564775,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.534,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.4719607970883062,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5294,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.38908917065634097,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5363,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.5301020845454625,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.522,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5135685563602376,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5476,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5056201846892486,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5404,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.5044982300128447,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.542,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.44254905675038864,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5343,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.42811691495345344,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5289,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.39043628410895576,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.516,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.4255610633116343,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5344,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5305555100867168,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5248,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5160581534051737,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.532,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.44000756657391055,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5478,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.4591433612596256,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5308,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.48121361970862897,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5382,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.41577250793265275,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5353,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.4354804949958693,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.53,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5145112943194424,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5443,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.49721907557528777,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5338,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.36751769158158126,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5293,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.49117448096211525,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5402,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5213677366043558,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5214,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.48983870134828555,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5388,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.4285994158331689,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5314,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.4639677123836367,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5258,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.4898451949748322,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5375,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.5589627602163791,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5444,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.4873125323867275,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5475,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.49038355521950294,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5347,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.523597222530327,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5393,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.49180390033734905,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5384,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.46692585820715105,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5344,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.4864811977449901,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5169,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5163401606463577,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5347,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.5735741391313983,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5266,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.5247299583606063,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5247,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6280338341086364,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5381,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.6290751180357627,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.529,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.46411813852626943,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5395,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.6178659757658371,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5408,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.6682674778347385,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5399,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5011464594894001,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5324,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.6194311608163797,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.526,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5500103378753705,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5305,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.6610981879820875,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5345,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.3989418548051119,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5388,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.49132740104778244,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5461,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5745375944697455,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5257,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.6041211739699839,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5507,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.4142735474232387,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5338,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5419623189097731,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5472,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.5855454118670395,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5429,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.49920850489059393,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.542,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.46707810910319586,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5326,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.5401645732033996,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5357,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.501669027121719,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5258,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.5491744298674858,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5387,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.45990432592920455,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5314,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.5192001788346428,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5405,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.49614596308694386,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5269,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4223313131554214,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5472,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.49037387745063504,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5298,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.44945730095361214,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.54,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.5860760677957111,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5362,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.4460606858466598,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5488,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.46273748918666424,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5403,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5343597350791324,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5516,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.44569657092469267,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5371,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.4947673625166357,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5303,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5454179922269152,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.54,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.4841354050293889,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.537,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.48194339089758415,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5476,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.3725089595662055,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5363,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.5210467871900962,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5348,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.46843993029749426,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5306,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.4548195475260701,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5283,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5282891649626286,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.534,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.4285118147810508,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5313,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.3967008568860967,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5242,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.46899818202565,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5178,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.39691329709986084,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.549,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.3786294873709974,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5286,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4317265512434733,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.535,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.42947218763133804,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5301,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.5083357476505861,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5374,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.4070157701681747,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5335,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.41848576797547415,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5427,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.40149027156891987,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5272,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.5649554754898293,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5366,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4465222661174202,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.525,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.5001964137027698,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5257,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.4619697106995791,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5342,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.3997639933369118,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.5251,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.43384730389260395,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5292,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.420967891652761,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5338,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.42378208503828607,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.516,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.5272421333738763,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5248,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.5418085398311312,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5393,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.41771221059635427,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5275,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5265327612155242,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5368,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.5213410201742668,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.543,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4234608021022804,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5293,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.5434229131171179,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5396,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.452096847238539,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5265,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.42034704176211257,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5259,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.6228489139205159,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.541,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.4746729133734871,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5167,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.47742429795502994,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5259,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.5759453157632813,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5263,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4427125755202634,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5226,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.5144067712545041,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5223,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 1.6219697837043974,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.554,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.61212197006645,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5267,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.6835826422404668,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5252,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.4555891005926363,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5354,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.569322718130674,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5423,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.487167248219097,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.546,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.575202969969505,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5219,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.5756143701976743,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5346,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.4688036488769756,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5327,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.6098184036324183,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5391,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.5324087501520648,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5426,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.40230414111797475,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5319,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.42367436819803544,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5304,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.47379356759607605,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5169,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4224384414214043,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5401,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.42069987979227397,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5356,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.3902179125796158,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5304,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.38395217262169623,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5316,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.44497084205976667,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5291,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.393903642498209,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5281,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.3676020593334558,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5294,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.42144619344485684,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5375,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.3050511943204512,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5356,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.40107788405642336,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5279,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.3567894186945346,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5303,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.3804929212471849,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5338,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.3708580642544647,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5254,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.3477586032607913,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5318,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.3374807570368879,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5372,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.37970860849709387,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5271,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.33133841961141824,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.532,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.36928051361242037,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5341,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.399145398345468,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5262,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.4019842651926719,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5313,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.3154143789079844,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5209,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.42069171863613425,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5359,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.34157015496788035,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.518,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.421643522681158,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5422,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.30496782758211866,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5461,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.38788534871571734,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5422,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.3714443010027537,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5266,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.35232378581081053,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5213,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.4113194341964936,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5267,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4042991165156725,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5447,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4117358007700567,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.536,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.3456611149009826,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5382,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.4105905324140896,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5353,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.40128456497547954,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5286,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.4005258753668288,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.509,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.46356084998017766,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5251,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.7033166229891099,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5556,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.3890152782737612,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5301,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.39488451324556295,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5392,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.36886194524890525,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5308,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.5820264077023618,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5322,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.36118528963976015,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5234,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.38076973566804145,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5248,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.4426046867917231,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5305,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3573910277805849,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5229,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.3710892548884385,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.5269,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.3295331037793307,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5214,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.3531969889338483,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5388,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.39881042986054477,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.522,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.3775426770228196,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5309,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4516240547440662,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5328,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.3569974984647304,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.53,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4346253951073267,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5378,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.396104091322656,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5179,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.38079666947307134,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5219,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.3790974880449714,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5331,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.3544263147596008,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5338,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.35483114436391294,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5375,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.37178332231617855,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5168,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.3666356913898718,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5272,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4121343789450127,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5393,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.39395024625692876,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5255,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.47827790545943655,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.5221,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.39608690797973584,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5387,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.49100774637669514,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5326,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.39431871857663975,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5209,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.4378571446683425,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.531,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.4110929243708989,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5078,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.3520857839116882,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5279,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.3831158324322985,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5211,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.4506952241994417,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5176,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.37048214150426245,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5351,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.41938973058854273,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.527,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.45642240294982434,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5196,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.42806986349283355,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5311,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.9796890508233712,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5211,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.47898187487124133,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5373,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4552641270311821,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5297,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.40852659535305785,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5248,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.4691427646191743,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5228,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.42206968380109167,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5273,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.47260334523892705,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5367,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.47562320470961267,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5333,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.42146934996492763,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.516,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.4164706660529278,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.532,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.43056259256968044,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5285,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.41216734801527094,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5198,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.41482779197507025,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5325,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.35355962942614566,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5154,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.42760537469825183,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5215,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.37586541049047323,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5336,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 1.7106816173616926,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5377,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4200488489795108,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5348,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.45554954053473645,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5358,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.39082684919839034,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.5176,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5153187872269751,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.4681,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.36493356227628054,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.4711,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5890361072387363,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.4576,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.5042851113029464,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4681,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.4388823357254649,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4686,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5106502001978391,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4593,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.763742051357144,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.4735,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4924716041206782,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.466,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.4728433500347379,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.4707,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.7363913071935382,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4666,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.47578374967615933,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.4616,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.4107724533327689,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4662,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.46279984655908585,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.4732,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.3800783063242569,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4685,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.485276706279541,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4769,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.42602206291745426,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4729,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.7599743683775304,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4694,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.47035251237462367,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4687,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.3750060921637407,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4793,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.4080647828931155,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.45,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.4328162833772501,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4656,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.46840209404318756,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.4661,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.3961680558837243,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.4674,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.38879165472962546,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.4662,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.3614287240401774,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4651,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.4093174599471791,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4665,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.3229160037367485,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.4684,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.39940089388642597,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4723,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.3453927322048681,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4678,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4532411429845806,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4893,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.8280062247016479,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.5013,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.30764535998717074,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.4734,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4522792069524533,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4744,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.3446736241721524,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4601,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.38779593499690057,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.4584,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.36306128227596773,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.461,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.4580600990118887,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.4641,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.38550317657391997,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.4746,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.5722589953158204,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4703,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.34882140331312494,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.4609,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.44477375115246465,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4691,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.4779666260355985,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4611,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.42145875814206113,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.4781,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.49669394425090235,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.4654,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.37718529054237154,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.4681,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.4466592922915662,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.4772,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.4607419536010462,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.4607,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.31892982557319344,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.4679,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 2.086141429454093,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4974,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.35802448415072224,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.4687,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.4400308470167157,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.4694,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3087081675846323,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.4751,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.4315384951376978,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4637,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.32994811148754094,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.4585,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.3402114821198145,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.4634,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.35892756080053595,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.4664,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.34091154774617005,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4657,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 3.848673432789534,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.4974,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.4685790555240565,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.4628,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.37951787008964954,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.4644,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.5746313004423547,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4625,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.4671725568418823,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.4778,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.35448175494604667,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4732,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.4701872440721875,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.4644,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.37543820848319576,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.461,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.5032452933573156,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.4751,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.36719878651573995,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4648,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.4643190973716062,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4609,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.4223844379133487,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4809,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 4.406660700584511,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.5053,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.6251395461206533,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4696,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.39933408514625685,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4705,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.5575822084851448,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.4735,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.4006722044599538,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.4667,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.5050314544148644,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.4714,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.38828526432869753,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.4628,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.497977571018419,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4687,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.41969004178550157,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.4748,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.4864753662106977,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.4732,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.3322832440795783,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.4691,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.700237350302134,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.4809,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.32613766883017,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.4606,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.40633919230691035,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.467,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.36590908898629976,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.4654,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.39948833070431844,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.4739,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.36498657229054,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.4656,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3631013758297528,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.4756,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.33130345975630643,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4611,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 2.6912304373293026,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4729,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.5853032407526865,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4741,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.4554902768134084,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.4759,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.4022513201305509,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.4763,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.39810417287623806,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.481,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.38272579052921035,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4649,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.39829893894835905,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.4749,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.387145529753367,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.4653,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.37621426358548105,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.4994,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.39221578098972254,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.4738,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.36405154061630784,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.4772,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 1.897983289865519,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.4729,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3821677670396239,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.4658,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.38458554708100845,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4689,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.4171684030533659,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.4808,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.38875763006183806,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.4616,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.44129855214627395,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.4712,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.381075895789507,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4698,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.4473305409008085,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.4792,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.3795682632355748,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.4699,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.5009364427902563,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4723,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3568128990190176,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.4769,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.6410637431877302,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.4714,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.3840993036285302,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.466,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.6344134145005079,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4646,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.34887428225962247,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.477,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.6323699380516465,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.475,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.32781074167184204,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.4651,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.4699903732397377,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.4765,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.300790221028441,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4742,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.342703368925382,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.4689,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.34371249515312663,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.4632,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3236418791555408,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4715,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.3362297118032693,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4689,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.33880016505165306,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.4705,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.34423995292544485,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.47,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.34300043177334105,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4778,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.38189881652296276,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.4686,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.2951219677347068,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.4758,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.31494577333613244,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4729,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.2931978800854603,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.4708,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.34421593800567524,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.4646,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.2965092932323547,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4652,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.34356440876084765,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.4653,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.2947594278921515,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.4581,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.3185785160196857,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4717,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.325669803035618,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4699,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.2879092667526612,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4626,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.32459001757436884,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.4695,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.2784422863423834,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.4611,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3234604834792893,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4766,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.28334765114943566,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.4696,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.5695614302123929,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.4774,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.30285149055517735,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.4583,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.2883002105171314,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.4576,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.36645537454430077,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.4726,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.26162022647065275,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.4722,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3611431521636031,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4676,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.2711192223429061,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.4603,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.36310273572074026,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.4627,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.2867924618142804,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4719,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3064231259798632,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.453,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3154216304607383,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4615,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.2628795420089613,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4586,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3595413274070339,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.4607,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.24771875780575559,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.4581,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.27229183317950967,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4691,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.27500650678195554,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.4683,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.2733264994240145,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4599,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3361992713631868,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.4561,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.292102609944501,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.4668,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.26688033791185417,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4562,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.2972940781127822,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4824,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3121641152237659,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.466,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.2811580997615316,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.4569,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.31111744768757615,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4726,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.30025482404195186,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4693,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.2518380142237068,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.457,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.2832793668647009,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.4713,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.28622891888557656,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.4594,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.27008223248537144,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4702,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.28104066105421804,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.4649,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.9252739577480253,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.4709,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.2809551783954774,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4611,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.2911325816530719,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.4664,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.27879859081896835,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4674,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.29697170940194606,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4573,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.29172054470102626,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.4649,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.25927227348762,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4656,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 3.1519585104261525,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.6502,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3959039652313687,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.4798,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.25779492941592147,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.4697,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.34455623115577544,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.4564,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.41496808600255547,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.4566,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.31268452161874866,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.47,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3730529942897945,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4779,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.30041478759952017,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.4751,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.2966632998807582,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.4638,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.3103566980501334,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4719,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.28931904327968627,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4659,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3127867307298735,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.4701,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.29114227041202007,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.4596,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.2998392633724874,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.4676,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.27139629968359574,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.4676,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.299833402958747,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.4692,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.28024768821841084,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.4713,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.9151982525114871,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4758,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.280533482699468,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4559,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.2913008104140479,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.4621,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.28660254722003364,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.48,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.2935006960226871,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4552,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.30658331499948266,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.4533,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.2730075779400773,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.4718,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.28966325085488964,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4765,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.27654254603391676,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4825,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.2754346726075553,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.4564,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.28263271951630087,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.4755,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.29632962958876624,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4601,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.26585440049215525,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4662,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.2902017394021577,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.4587,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 1.3653273556912249,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.4813,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.2731291997221204,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.4556,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.267168752882776,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4607,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.259492954610671,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4606,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.24724915229549746,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4706,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.30238867093167554,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4654,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.2813229573492844,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.4642,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.26872327338347163,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.4665,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.2519683882081943,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.4581,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.27151360868587515,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.4734,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.2430514889145036,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.4752,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.2579365217123858,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.4691,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.25666238349266757,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.4542,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.34519141741191517,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4665,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.30201234507660013,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4831,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.27969601685844697,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.4626,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.29083915096298296,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.461,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.291003062923558,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.4736,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.30682107307318096,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.4743,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.2571772355149004,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.4756,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3279476962911183,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4651,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.5532750064664336,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.4683,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.2817435396245607,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4676,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.2840770832548764,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4735,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.2715431529268532,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.4743,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.2658444613147528,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.4678,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.31247854058270114,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.4606,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.27291761159326394,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.4731,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.2705841929961933,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.4596,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.25914487948202786,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.4684,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.27890272071312644,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.4617,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.25811068548463045,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.4589,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.25504163716944056,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4668,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.26770410028018926,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.4737,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.2518915594533249,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4647,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2703466210295371,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.4741,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.2929319394930963,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4703,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.279370741955299,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4625,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2659295942250537,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.459,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.41834066737060477,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.4115,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3175221732182355,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.4094,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.46503880600899955,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.4159,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.37680567702984463,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.4048,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.3615071569736759,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.412,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.37564801929700764,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.4065,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3591828713262447,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.4055,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3401483074895844,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.4111,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.3620673792741283,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.4123,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.5343526001146316,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.4197,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3847794640627987,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.4089,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3536359552158747,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.3998,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.35226616530858784,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.4096,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.36426685941787906,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.3929,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3176683529489528,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.3916,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.2831158492156244,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.3947,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.318667475418531,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.4015,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.33580898659054315,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.4072,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.30677649566242593,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.4024,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.36107469777366785,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.4155,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.33294675721769523,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.4096,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.3207452140755846,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.4041,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.35398072749427134,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.416,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3017481304900901,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.4088,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.34390544343949303,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.4133,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.30157174000262466,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.4052,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3424025246065729,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.3943,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.2907194891355342,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.409,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3608293905883106,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.4126,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.2974573785661174,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.4103,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.28473119598042534,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.4104,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.27244488933560734,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.3923,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.2948169828495671,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.4041,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.26479105416209997,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.4013,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.29373610176231485,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.3995,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.25659177676284245,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.4066,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 1.247987237666219,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.4196,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.26106917486566467,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.408,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.28067100646924614,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.3956,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.2707161687300642,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.403,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.2569675818574651,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.3984,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.2552370975124539,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.4192,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.2729226663378076,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.4127,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.25161439756056797,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.4138,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.26404192905722396,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.4131,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.293078995276795,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.4073,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2605991848247656,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.4054,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.25807026092025287,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.4049,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.2805025300915431,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.4185,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.28666007672422256,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.4075,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.24768782296422726,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.402,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.2885546417866596,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.4157,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.2718873833795982,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.408,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.26840194907386894,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.4081,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.2658550038426362,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.406,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.24134393871707452,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.4091,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3080599580729905,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.4098,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.2407073954206501,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.4016,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.26872885768523225,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.4058,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.29195148920935443,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.3997,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.252635438614317,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.4104,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2688593860583338,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.4049,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2613931430095154,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.4137,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.2501452891001241,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.4022,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.271033576645497,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.4061,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.27510299904375457,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.4075,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.24700520319401706,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.4017,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.30876071879412387,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.4116,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.2589850435901565,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.4161,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.3253376386223437,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.4132,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.2609716880162293,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.4045,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.28795418693733504,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.4168,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.29026619343250376,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.418,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.2581665773284318,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.4166,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.308083885486012,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.4071,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.29370916655227186,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.418,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.24953857406031543,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.4204,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.31438247940877,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.4155,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.24117046034713244,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.4047,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.264129330943157,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.3975,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2863029258921466,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.4088,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.23331860808589394,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.4078,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.2958823552671633,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.4069,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.24917307845105638,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.4109,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.30063671405698594,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.412,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.26863948369607116,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.4175,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.26401907234329713,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.4058,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.2584930091624232,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.4069,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.24048455096853116,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.4113,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.2761755814800438,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.404,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2733492458578424,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.4166,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.2700109947237333,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.4168,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.24642037251324633,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.4038,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.2603992199111931,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.4151,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.24873193297225546,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.4089,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2360400959807868,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.4102,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.2502821335246811,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.4104,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.24874596498245238,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.4031,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.24557689406865355,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.413,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.2509887447818246,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.4042,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.24916671844182828,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.4058,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.24621999087708643,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.409,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.257872597501811,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.4104,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.23426541938344467,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.4087,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.2502263938049658,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.3959,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.2562870242356867,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.4065,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.23789418745563157,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.4147,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.24415293574262173,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.4153,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.26507752842689875,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.4029,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.24914856604418495,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.4013,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.23058324151744689,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.4044,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2271595868941948,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.4122,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.25877044683606626,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.4086,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.24579965531483164,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.4128,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.28902780453576626,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.4085,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.2545929557864567,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.397,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 3.762921942393005,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.4562,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.26918352760264247,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.4091,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2524229942596609,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.3924,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.2620839567389171,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.4039,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.249689706631901,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.4094,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.26581929818380967,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.4037,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.24547179391730428,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.4063,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.2612183637836692,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.4048,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.24975729298298469,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.4148,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.24916689406859302,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.4036,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.2549513417935424,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.3952,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.24248320141669066,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.4087,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.24160416967889958,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.4057,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.24649490613533298,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.4194,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.2615234674246874,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.4067,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.23275419615708076,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.4123,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.27206700504691417,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.4072,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.27476221343405205,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.403,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.2751161611637868,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.4029,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.2458823546218567,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.4151,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.26333377484788706,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.4162,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2756898267401348,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.4107,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.22286370520901225,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.4139,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.23224136682590815,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.4176,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.24697093119344163,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.411,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.3352325433409668,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.4024,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.23253698889981742,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.407,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.23287751228743359,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.4031,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.23604219444279298,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.4075,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2414303312294198,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.4188,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.34334228324921345,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.4067,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2161500123339974,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.4048,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 1.4771739517122178,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.4104,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.815146604579632,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.4282,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.24218968769482815,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.4013,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.24469221762080923,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.4068,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2263945838940423,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.4174,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.2611209463396334,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.4101,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.24400336627609673,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.406,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2558559174737731,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.4097,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.2764713167639999,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.4151,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 1.5971436789388445,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.4144,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.30278475538059446,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.402,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2628443479879542,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.4113,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.24465646470563393,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.4078,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.3145397386043414,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.4207,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 1.4010007444251757,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.4258,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.24893817437267576,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.4158,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.24925504435997492,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.4057,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.2439928831643244,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.4147,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.2564174630348654,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.4153,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.2521477395388077,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.4116,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.24043055656336834,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.4003,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.23728627045244743,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.4063,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2469099079609321,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.405,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.24994377854227579,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.4166,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 3.435443742195246,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.4102,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2690474232668554,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.4179,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.2424799034967081,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.4092,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.23396920280417374,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.407,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.252718521927306,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.4013,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.3565109566578244,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.4051,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.26419098318921547,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.4095,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.2364999469101792,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.4082,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.23590915770353504,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.4128,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.24739010843560452,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.4084,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.2382614935809034,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.3978,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.23368475602091932,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.4061,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.22363091743184751,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.4088,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.24034197176546168,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.402,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.230323908650305,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.3971,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.22654436428458175,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.405,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.6941629803663271,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.4238,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.22467389635043314,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.4164,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.24020031535579067,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.3997,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.22524442095308,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.4126,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.24437807555799787,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.4096,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2350235275478412,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.3978,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2321328503152467,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.4152,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.24865411055543465,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.4097,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.24647438270248637,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.407,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.23494090458009523,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.4146,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.24783249564854873,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.4145,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.23259008014273494,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.4053,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.24546623195117812,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.434,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.22207576438304202,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.401,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.23375273984824277,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.4028,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.2576510761017135,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.4065,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.23803626730523478,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.4097,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.22833866215565402,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.4084,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.21428633399157063,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.3996,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.23623949895083596,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.4036,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.24693837622152487,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.4213,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.21976214703389432,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.4079,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2283780701056759,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.4079,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.23863581667094558,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.4068,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.2256878438893102,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.399,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.23108200590905975,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.4206,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.22662361911629705,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.4145,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.2178239103815553,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.4066,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.22652261389028322,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.393,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.2199896076494278,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.4009,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.22701303834607653,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.4061,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.22869198192882115,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.3997,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.21606602110734122,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.4203,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.20999836288025098,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.4046,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.224171766636016,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.4027,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.22727799432091417,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.4015,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2304829636676509,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.4103,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.21678335254636982,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.3887,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.224134880175259,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.42,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.21701145186382734,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.4085,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.22000481429615099,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.4083,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.2263028918125072,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.4019,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.23463607188102434,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.4058,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.2255117804707956,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.4084,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.21835654654262435,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.3992,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2259530399641989,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.3965,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.22666167087709832,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.4067,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.215926210632483,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.3953,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.2123668284321307,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.4022,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.20706394540134043,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.4118,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.21250336925370658,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.4126,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.2066099608872915,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.4069,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2160402531494787,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.4002,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.22601863278495452,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.4076,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.21895855905361294,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.409,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.5459715991208831,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.4196,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.20825850768576498,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.4104,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.2012435991690814,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.4124,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2139459500390923,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.407,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.36567866969851637,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.3609,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.28710455256874506,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.3681,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.24306080621297968,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.3667,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.32169646430668153,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.3688,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3167474264604573,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.3601,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.27705849598212523,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.3611,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.2574541611729595,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.366,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.3070275425854589,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.3701,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.2948464574274173,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.3603,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.25765500782788453,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.366,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.24709158557220595,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.3678,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2572243467919691,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.3556,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.26806162917403153,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.3764,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2902463523516385,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.3644,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.24240552408120408,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.3728,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2580548890600478,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.37,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2373448419288514,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.3515,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.24233682913767615,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.3666,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.260862990494156,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.3508,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.23289705251316345,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.366,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.22636984028327758,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.3734,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.25202459358242507,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.3675,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.23656149278553332,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.3665,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2360913579754303,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.3726,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2405535677996727,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.3607,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.22304794179986287,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.3568,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.21983673375670598,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.3577,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.23252441578306338,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.3746,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.23644265868535938,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.3588,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.22328508784719087,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.3705,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.2274869232171359,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.3747,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.23346862524466216,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.3608,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.23079320088572738,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.3713,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2237354954098298,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.3704,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.22079218972249823,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.3661,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.21638090909095817,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.36,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.22149375930303622,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.3668,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.23651566703543592,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.3694,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.6011526745589089,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.3616,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.22314147971508502,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.3734,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.2073017729509436,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.3634,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.2356971625472572,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.3639,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.2233661926999842,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.3705,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.21852083670084632,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.3821,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.21455446871543954,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.3643,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.2203709379116882,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.3664,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.22408079151446833,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.3657,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2294571167457172,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.3769,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.2269716568398208,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.3551,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.22596797467670537,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.3666,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.22313038272108604,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.3516,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.22065698548304158,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.3618,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.2365846775216945,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.3734,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.223323635605792,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.3729,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.21735161384384424,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.3715,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.21859880180269056,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.3715,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.21120622655695803,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.3651,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.22210509442609236,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.3569,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.22762991785866873,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.3569,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.2201192078550884,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.3684,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.2190483679596628,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.3609,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.22433815722136713,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.3582,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.21382954761872325,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.3631,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.21711753051953528,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.362,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.21976761577001386,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.3578,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2182961543457152,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.3688,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.2157368185717507,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.3659,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2203673031196247,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.3539,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.8598564780057826,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.3821,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.21587289618270153,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.3647,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.21370283737294737,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.3685,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.22441729476434272,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.3595,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.22767585118645867,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.3632,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.21758205756993818,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.3716,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.2245063433905315,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.3608,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.21366351887700574,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.3656,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2109364705030163,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.3667,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.21222327884687767,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.3564,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.22258082419396757,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.3732,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.21230616254008436,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.3629,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2248390378610199,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.3669,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.22751696709236105,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.3648,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.21150122345489042,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.3698,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.20480974603338356,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.369,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.21250584001587522,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.3682,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.221065919941582,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.3611,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.20883643695133336,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.3605,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2186454097964238,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.3593,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.21093880768480092,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.3587,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2158128128344383,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.363,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.21726710379123745,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.351,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.23267376275087526,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.3616,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.20799500007204952,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.3728,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.21631035891902353,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.3651,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.2142852995821113,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.3657,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.21003866082625755,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.373,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.221705661232631,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.3653,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.20659903000347288,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.3716,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.2127982678940133,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.367,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.21293392739809133,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.3667,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.21177331179108622,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.3663,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.22453375033024028,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.3702,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.2132362139171614,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.3582,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.21547095072864805,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.3609,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.22566850622534734,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.3572,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.2168113957432023,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.3705,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.20236981016567182,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.3749,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.21413564148101294,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.3677,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.2013116706599572,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.3674,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.20834488575488294,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.3685,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.22605581796657315,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.3602,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.21148247284204352,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.3646,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.21014842890690658,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.366,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.201085386538109,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.3573,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.2028102671843852,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.3714,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.20578314793152477,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.3667,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2230622376485439,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.3606,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.21493701500119677,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.3628,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.21811712325094545,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.3606,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.21702109811466017,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.3624,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.20993724073005451,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.365,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.20906516452136603,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.3528,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.21094390124312318,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.3649,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.20758953651167403,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.3722,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.21179050564803648,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.3723,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.20825054303440277,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.3674,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.2200566493495881,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.364,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2358601286450109,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.3779,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.21308788083069238,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.3697,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2076672158443412,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.3687,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.20725541622424073,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.3563,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.22090631622158904,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.3637,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.21857611154734985,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.3739,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.22068448720132613,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.3617,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.20268425385916744,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.365,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.23024812205137252,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.364,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.20623551199071471,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.3759,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.21909687971200387,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.3735,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2888854957149479,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.3621,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.20659146490576108,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.3705,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.21135341302975905,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.3687,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.20579386695656068,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.3626,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.21269315185515777,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.3679,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2155730537660372,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.3659,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.21071325088857534,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.3626,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.21248978845402347,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.3686,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.21184211151183482,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.381,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.212218765407063,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.3736,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.19930121442144677,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.3675,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.22148442180240763,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.3546,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.21260382779564982,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.3791,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.20615530468822565,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.3624,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.20817409893242342,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.3723,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.208239367412326,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.3682,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.20793045298357218,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.3617,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.208286194298556,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.372,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.20774285741358148,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.3686,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.243061485769701,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.3657,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.21037490271724368,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.3725,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2165853308912644,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.3766,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.20417700522790258,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.3822,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.21392090052384488,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.3605,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.21198305007163343,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.3659,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.2106247105594778,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.3569,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.8331094277136462,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.3683,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.20450555016589408,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.3601,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.21013013181862464,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.3672,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2065694712923297,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.3694,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.2049275335602078,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.371,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.20291769059839154,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.3637,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.19639550303438585,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.3679,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.22247556946388145,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.3747,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.21952269969127045,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.3486,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.20651935763596976,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.3619,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.20524180151409252,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.3617,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.20989958392089497,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.3751,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.21356330041830904,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.3699,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.20861698053117977,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.3643,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2135945584587292,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.3617,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2058819889276827,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.3623,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2080953874551225,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.369,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.21117999925102673,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.3658,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.20626741642907084,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.3684,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.2052875371188728,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.3654,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.203047333879563,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.3687,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.20691827299789972,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.3749,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.20805338683223487,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.3691,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.21663575693491827,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.3603,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.20656350230716947,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.3704,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.19747834563774008,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.3599,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.2092793423330312,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.362,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.21190511049114363,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.3673,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.21066245541224846,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.3652,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2070250710838792,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.3673,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.20641391230247513,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.3619,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.211818746407011,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.3659,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.21680928156803045,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.3638,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.21333513039036212,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.3603,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.20769010736462037,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.3703,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.5543317153733524,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.3663,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.2061932406993292,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.366,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.21786985554935526,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.3613,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.19661056295396517,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.3728,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.20952500985599232,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.3669,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2942866008214293,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.3721,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2034751965846657,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.3602,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.20852170721402807,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.3706,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.22370524303844394,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.3565,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.21430840862633638,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.3599,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.20083892476611728,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.3795,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.2075510731007578,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.3544,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.2049665240639406,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.3698,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.2082371791074692,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.3683,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.20440446292872616,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.3682,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.2041079477877983,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.3656,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.21405534029194634,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.3632,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.20144072217930944,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.3644,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.20408846551090187,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.3658,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.20924554864351322,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.3713,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.21785449763618353,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.3647,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.21042616838215858,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.3676,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.20091951102301928,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.3701,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.20396312464283523,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.3597,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.20140336819526475,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.3661,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.20157995944213672,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.3779,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.19914294535181296,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.3734,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.2163909603311944,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.3634,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.2114769295656952,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.3635,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.20167979991165055,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.3652,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.20455336528406945,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.3661,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.21815021917188296,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.352,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.20742601084111006,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.3659,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.20364450868436057,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.3631,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.20013952602390128,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.3678,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.2111285522321456,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.3535,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.20413909852877116,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.3701,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2023597010287136,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.3648,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.2008481185236974,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.3689,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.6747612185061025,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.4454,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.20320641878401316,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.3547,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.206607542635274,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.3595,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.20643727711012502,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.3635,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2055263880748834,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.3613,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.2161042034429474,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.361,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.20821481358326452,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.3596,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.20819018062629577,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.3759,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.21543091006628357,
+      "learning_rate": 0.0,
+      "loss": 0.3632,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 4.1022953318053315e+18,
+      "train_loss": 0.48579915404802393,
+      "train_runtime": 32208.9394,
+      "train_samples_per_second": 4.905,
+      "train_steps_per_second": 0.038
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.1022953318053315e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}