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": 6.556274302380661,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.8802,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 7.099925976326479,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.9148,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.582244498207503,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.886,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.608353019790543,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.899,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.378800100518705,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.8732,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.384354366717922,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.8657,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 5.18867599438634,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.8373,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 5.021655728528782,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.8484,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.565997888454773,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7709,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.4461917262397903,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7633,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.1482178730403145,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7779,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.8397954960807645,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.7661,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.12915601808594,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7825,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.9142646308099898,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.7442,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.859876474162646,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.7304,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.134844778848195,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.6967,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.9091231589713518,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.6548,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.3589028511036854,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6436,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.819308615521742,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.6339,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.612679055536336,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6417,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.4803281532876702,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.612,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.7281336382007642,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.6304,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.7314196888442932,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.6199,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.3117492831367177,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.6075,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.8826794504204815,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5972,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.0158453104273923,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.6447,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.9523574065390978,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.5883,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7474490170317815,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.5977,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.6833896917340865,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.6089,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.7221243959663343,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5899,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.8729224936609835,
+      "learning_rate": 1e-05,
+      "loss": 0.5892,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6933824433960507,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5613,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6327588206167358,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5579,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7966800263343176,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5781,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.7394354027491982,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5445,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.6340584336541913,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.6083,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6570649126056638,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5453,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6479509529487099,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5552,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5836812042781374,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.5315,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.5556632564422724,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5451,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5725323275784325,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.5596,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5492420084609584,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.55,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.536285520507715,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.5366,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.5546759190655415,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5414,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.672299903626552,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.5436,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.6396174000676711,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.5541,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6304544329044366,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.5503,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6055013640101096,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.5411,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5932033805907122,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.5373,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.5685846000515822,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.5274,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6121334689981252,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.5265,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5589229392994763,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.484,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.504035009621965,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.5048,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.564991810566453,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.5244,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.517465959861378,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.5281,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.4942832666784935,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.5082,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.5174614998744484,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.5053,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.5363039236613291,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.5318,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.48604400977454176,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.5153,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5266695404493132,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.532,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.47699144259715953,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.5105,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.5883897992049375,
+      "learning_rate": 2e-05,
+      "loss": 0.5085,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.5265513462288146,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.5087,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6161465009397633,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.5311,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.6655047514473712,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.5116,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.6204280318180871,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.5126,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5193635896217719,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.5079,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6097244428884012,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.5028,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.5572450049369314,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.511,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.5574635621449044,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.5309,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.5984562483162593,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4968,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.7061393108242683,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.4967,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.6670557387041655,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.5487,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.6158990790834089,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.5108,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.5515598231567637,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.5135,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.5949227398502409,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.5097,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.5541632073734026,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.4893,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.750876414351505,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.5148,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.5659248296369143,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.4898,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.6361528799851963,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.522,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.5749835445206146,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.5179,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.5779660257306047,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.5132,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.7615354160607605,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.5282,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.679005870936874,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.4794,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.678950401771645,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.5066,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.737272495323125,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.5125,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.7984758103211171,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.5152,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.7384114895490471,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.5111,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.7714169675017556,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.5081,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.7851384630641379,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.5271,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.7895168745825927,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.4924,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.8552471805380664,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.5133,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.8994023159331521,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5101,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.871957290788016,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.4986,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.7339064036204248,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.5061,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.8979811636251567,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.4892,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.6947534106192653,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.5077,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.8047800429056541,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.5071,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.6797368268683477,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4953,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.4080382701977625,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.5528,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.7152199830806081,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.5343,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.8510066654070667,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.5217,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.061631957326944,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.5027,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.8974115164218561,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4911,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.8588343522046761,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.5169,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.0740613300473958,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.5201,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.5554090096942136,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.4914,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7935779304837829,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.5205,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.7509490835481274,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.5125,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.797834667838227,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.5087,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.8438624739383283,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.482,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.7756937779636404,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.4885,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.8128677169012504,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4794,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.7502675464917193,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.5243,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.5965420094451622,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.5206,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.6428696080133028,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.5252,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.7425960403665488,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.4947,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.5617263577287404,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.5143,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.5648064587287597,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.5066,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.7107676429852857,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4873,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.6094626024785169,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.5071,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.7297190130538801,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.4847,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.9072861753423752,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.5136,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.6369482484693051,
+      "learning_rate": 4e-05,
+      "loss": 0.5138,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.6266941849316542,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.5082,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.8824561260976835,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4845,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.643622790471516,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4898,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.7393088306532676,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.5054,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.7850192410398735,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.5334,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.519483259869964,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4912,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.7170387685315575,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.5033,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.8369222916966772,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.5017,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.5980979208523914,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.499,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.7519282331357625,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.4694,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.6235608857842048,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4995,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.6153366576453728,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4806,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.5788351687120209,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.4698,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.6280056135115947,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.5165,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.6383423679095268,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4875,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.5966876336513959,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.499,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.6029938435071709,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.5164,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.5634934048294375,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.5114,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6122624194903992,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4828,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.5270353754583734,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.4882,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.5238255218155325,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.5177,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.4944291322451637,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4848,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.5248620756237187,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.5028,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.49343439189532595,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4995,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6487221394660517,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.5047,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.48172549138812565,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4768,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.7097035103854626,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.5024,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.5228040233799739,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.5173,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5909728455184118,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.5028,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6005497279018323,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4862,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5535209298808069,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4903,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.521688007530726,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4905,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.5082511721548374,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4902,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.46853633121144667,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4796,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5812945620238508,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.5244,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.4413831545783975,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.475,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.5187360222090703,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4798,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.5020158623350619,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4978,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.4905239563410164,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.4929,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.48595031379219467,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4775,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.5446053866612691,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.5216,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.49925260304045055,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4897,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.5472929997493156,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.5069,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5758188606999005,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4759,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.5514106321539053,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4858,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.49950594197609816,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4943,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5423695802338585,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4952,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.41541116671294664,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4909,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.5360662060701472,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.515,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.4383346673625947,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.507,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5134741458048531,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4994,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5056984073598935,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.5313,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5407946990200136,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.5164,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.42100719703362,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.5022,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.524088931515803,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4872,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.49073356865612405,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.5044,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.46486872067755053,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4748,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.4453568260946404,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.5072,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5890803874666826,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.496,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.4991732058782883,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.5022,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5666719276741827,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4907,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5086607842116152,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4897,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5610623847989382,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4731,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5166344519290431,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.51,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5566670434140166,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4705,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.46372233893374004,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.489,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.5242091734501708,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.475,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.46590960499870354,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4829,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.46364031257212146,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4656,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.4410090490609485,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.5059,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.4538463471447881,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.5056,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.45804734567226973,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4856,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.4233769149666748,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4785,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.4225741311906946,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.4767,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.44453474035040724,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4794,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.47130443728674093,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4973,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.4694255218419074,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4792,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.4864960124065077,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4802,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.41889860816459373,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4841,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5120081889844778,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.5107,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5127006516098399,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.5047,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.47540364103216315,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4705,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.6489522873928512,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.4963,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.49117767585185124,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4917,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.5905228517183303,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.4945,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.7288776962839623,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4674,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.6547959226509573,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4966,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.4996390061558336,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.5159,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.7076543127181204,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.5059,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.5919032154578742,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4826,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.4653198082590639,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4709,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5754763008977315,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4735,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.49990067420249645,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4785,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5290422452228254,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4953,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.5335082565811666,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.5086,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.42144468237243143,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.4836,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.6036029734179247,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.501,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.4480743031320677,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4878,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.539680418631316,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4869,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5448430040963609,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.5016,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.4728756599507083,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5012,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.544556488756892,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4712,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5866496209980814,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4814,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.5474289973536101,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.5137,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.46312138587303936,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.4531,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.523897661826743,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4614,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.4231472182694884,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4752,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5231241510037498,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.4805,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.39500548075411446,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.4994,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.4909484806574135,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.4928,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.4835529094697343,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.5005,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.4401728752737332,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4708,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.48694484884035716,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4949,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.4705987953789656,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5081,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.453695756561949,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4822,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.4631735748039865,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4924,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.428077588767107,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4874,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.43853241261588766,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4916,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.42192242602995667,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4655,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5068813408981486,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.4948,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.38651559322926216,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.5005,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5093714035783689,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.4757,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.46395478732247586,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.4829,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.6059146805864685,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.4176,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.46955096870734137,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.4036,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.5269481185425379,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3957,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6290522320605374,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.4317,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.5053184922883985,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.4211,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.4946498596534427,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3681,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.46183253927655293,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.4284,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5639672450073587,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.4301,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5856518780057097,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.4138,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.4850161146426228,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.4033,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.47956049892840036,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.4216,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.48071676236646543,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.4096,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.44891116743934545,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3859,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5386641810605257,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3813,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.523855902852812,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3775,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.4617863577686081,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.4102,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.46393022674556367,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.4026,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.507458725580501,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.4122,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.4274007426944779,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.4001,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.6807401308821458,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3885,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.4511390135723267,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.4115,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5522518832622753,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.4066,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5496830909525228,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.407,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5394389899969798,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.4018,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.48078914669921496,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3828,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.4863423813553496,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3926,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5479754790200554,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.3956,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.48936994960310115,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3988,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.4889645053787898,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3772,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.5894136722337002,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3814,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5070184960676499,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.4287,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5706954339856809,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.4144,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.4828979231861648,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.4179,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.4509891874687832,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.4118,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5384526718766505,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.4078,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.4368412843270201,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.4051,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.611725903370901,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.4119,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.44877742359006945,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3777,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5913531859143691,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3999,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5016974311093372,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.4043,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5272369481591119,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3653,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.457414704050737,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.3964,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.4715594534487697,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3789,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.4060267961036177,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.4129,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.4965344427098742,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.4001,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.45970196981521344,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.403,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.45646944206537965,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.4005,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.43080530851182514,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.4219,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.6194021535705901,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.4052,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.4424641700736448,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.4127,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5539688872085105,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.4053,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.4607133849739246,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3926,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.42894753601977187,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.4265,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.4811087483487087,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.383,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.3798618086405649,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3921,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.5477232614605599,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3995,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.3907240513894972,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.4001,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.5029942559216053,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.4049,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.45147499286319787,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.3669,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.4314604484676248,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.4013,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.505517162189193,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3931,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4590685036306828,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.4161,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.48613193387516884,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.3719,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.5047993971382795,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.4064,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.48922862336849476,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.422,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.4852646521544164,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3885,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.49043120758224096,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.4171,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.4512093134545266,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.4061,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.45631150115509733,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.4141,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.3984784817149205,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.4105,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.504073355577155,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.4159,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.38280001317318524,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.4283,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.4834673831260887,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.4051,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.42846435906905217,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3873,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.4328507608330132,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.3899,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.4710367156310434,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.4035,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.41425805571370594,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.4252,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5101031232960807,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.4103,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.41892467031082703,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3883,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.5321394286726747,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3897,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.46186090361379817,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.4063,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.5852906333538979,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.4253,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.4182082288121783,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3824,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.48205173393755363,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.4064,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.4175986614624853,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3984,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.40463263988854165,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3977,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.4351293365693614,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.4121,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.46628991248296964,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.4157,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.4646900304926367,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3988,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.4701042785962829,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.411,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.46667022112234324,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.4109,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.44229232132747603,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3578,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.4790201358516701,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.4068,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.38806825683909624,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.428,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.4562214944877759,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.4247,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.45568409245779723,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.4127,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.48209379347374437,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.4051,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.4644618311769956,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.4127,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.5449485321460462,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3842,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.5217031377981908,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.4201,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 1.0903204502458665,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.4129,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.4764208848992023,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.4068,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.6196161492801281,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4002,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5143232633738862,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.4408,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5095473462076531,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.4118,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.46493570650045085,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.4084,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.5201574460313998,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4001,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.38774344395105864,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.4063,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4974195948449409,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.4074,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.37781648603050394,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3878,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.5097004707287387,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.3941,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4543292365781642,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.4426,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.45439346937057024,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.4225,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.43703869698351977,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.4067,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.4187517915041932,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.3921,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.42214159783760874,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.3913,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4521851999911768,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.4136,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.3897310289471666,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.3826,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.4135622707378669,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3816,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.4022931448421802,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3927,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.4057637022202229,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3924,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.4254618743253842,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.4018,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.4258718045291669,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.4168,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.4882165154469844,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.4037,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4891231687173529,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.4225,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4746444843859338,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.4319,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.42200086299056294,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.4252,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.44753934476488005,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.4019,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.43109245022805165,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.4044,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4118986597510915,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3875,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.4141669981884563,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.3938,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.505018451706969,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.4069,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.37367587297311583,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.3891,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.4959393138348356,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.3895,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4225688732939184,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3974,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.43917958501495324,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.403,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4833340185496365,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.4111,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.48114619720402163,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.4092,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.472306518821016,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.4271,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.48134360282902433,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.4078,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.45884611634673556,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.4164,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.5076069056437178,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.4176,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.4122807133658982,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.403,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.46023462651373553,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.4102,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.39286926175310016,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.3943,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4164443467279868,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.4035,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.4612855163203143,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3888,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.39404749080181206,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.4044,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.4386508245481883,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.4166,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.4116843313236591,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.4453,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4280096307105853,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.4233,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.4406577323137038,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.4011,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.41361920073339814,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.397,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.4371392601248481,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.4227,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.35156942720868584,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.407,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.4168433781221945,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.3995,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.3739199831229741,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.4014,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.39348913619933745,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.438,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.42940551658185755,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.4122,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.390215191521809,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.4054,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.45454284745713225,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.4203,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.3765208547684692,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3974,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.43690786438171914,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3793,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.41953460185411817,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.4098,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.5125846292822954,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.3817,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.4485330249253853,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3875,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.44359744646867805,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.4172,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.45087077834663003,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.4208,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.4591836631057004,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.4313,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4113885963855489,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.4075,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.509012818905189,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.4261,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.4287705567968438,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.4035,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.4451244857264928,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3557,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.3692810498461831,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.424,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.4564668231876424,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.4036,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4087054929624889,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.408,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.38415917386812204,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.4131,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.41519239367067573,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.3832,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4203271455189271,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.4095,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.3831819276638983,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3981,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.39990624449174805,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.4085,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.42918984262574683,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3936,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.41751039139326773,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.4291,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.4209162169824343,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3948,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.3453264621227897,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3796,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.393421154442882,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.4011,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.3853470832432421,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.4256,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.3781712470348502,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.3954,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.38392388699746993,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.4133,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4007451691416192,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.4174,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.3760205236450616,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.4387,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.3630413993542896,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.3849,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.3883326827115642,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.4023,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.37176926075635186,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3818,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.35154670243320396,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.4042,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.39082171900450513,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.4065,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3893269291692787,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.398,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.43652740257127365,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.3944,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.404250389577489,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.4066,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.45892649935164764,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3987,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.4143343138551275,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.3798,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.42602888679862794,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3677,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4220956877589648,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.4054,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.40233658382347814,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.4125,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4292234799194469,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.397,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.4074395290308018,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3955,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.4008967423178474,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.4131,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.38541214119538314,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.4121,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.3995208331977738,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.3929,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.35315137731958346,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3894,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.35799288612545616,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.3954,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.41937745086227884,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3812,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.37244568107534726,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.3947,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.5178728045097625,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3986,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.3630130390217793,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.3904,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4126347875016539,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.4083,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3671504993259414,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3763,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.4115703504245329,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.4083,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.429048994705203,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.3832,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.38484957697946776,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3813,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.4690636029485988,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.4247,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.45008293580159053,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.4086,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.48038293648042946,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3863,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.42449686533325576,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3972,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.42592198778584356,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.4173,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4289215751149791,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3974,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.37933331701099005,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.4158,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.39003471495475633,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.4104,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.37835077289058366,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.4036,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.34085667952714616,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.4054,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.393015383323638,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.4119,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.3470255874196555,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3782,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.4104158265310309,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3985,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.3625095995604447,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.404,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.45611573302641945,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3973,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.3590623383934537,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.395,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.43396680587271674,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3793,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.4042181152711686,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.4154,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.4223059611908839,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.3843,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.4710956641024398,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.3929,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.42915786804028533,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3931,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.4578606572540288,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.3985,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.3715427279576852,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.4004,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.3929918238512957,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.4012,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4095397978244464,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.416,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.36491598711837736,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.3833,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.37777990121645155,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.3843,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.499404421264139,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3125,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.5171521119021955,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3045,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5141528819972043,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3044,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.46549830948961535,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3139,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.43062238470732583,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.2898,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.4152789703353153,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.2894,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.4492500487053409,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3091,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4545143692121376,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3063,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.3941770237919938,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3054,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.39102356784016606,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3053,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.36992613870626795,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3121,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.43599681448535826,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3022,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.3782251703333513,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.2945,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.40268483718229114,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3013,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.3960864530767455,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.2978,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.43569046253509813,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.2943,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.36605806501446364,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3002,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.45399594901380486,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.276,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.3721739297961333,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.2865,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.43183014983785034,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3164,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.36293541463949136,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.2953,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4175049367553891,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.2701,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.4096407205596863,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3025,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.35644858420354847,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.2965,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.47853128180298293,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3054,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.3577680659328215,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.318,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.5057437222327896,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3098,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.39899838276625993,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.2925,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.4566770403145852,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.2972,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.38175433099242884,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3049,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3945567118988323,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.2815,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3854271560702956,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.2983,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.40101633533288006,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.2725,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4036762524700297,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.2668,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.3749404626484285,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3043,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.43572680144151216,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.2836,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.36293885741381554,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3171,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3877132552464175,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.2928,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.3796919274217298,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.2671,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.37482801204078947,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.2828,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.4190303624319354,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3071,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.3837044140924232,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.2786,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.38206560337336626,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3012,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.39696510533069623,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.2879,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.3644513757370934,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.2936,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.39190244304444394,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3101,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3836823946835769,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3058,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.37864842632118256,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.2882,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.37498712416822505,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3007,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.445001811929001,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3144,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.38661807556540156,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.2823,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.39949051348762654,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.2957,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.42800676832731205,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3019,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.40744320309332654,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3132,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.37945628264102793,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.2885,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.36853202051915585,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.2949,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.35908200250102246,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.2845,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3586565545788762,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3074,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.38188403262590254,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.2969,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.33154859048473473,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3123,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.3640793178911397,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3042,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.3753667405177556,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.2978,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.34850012103457445,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.2923,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3960629427348628,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3032,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.35856229301966264,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.2985,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.4155151189078132,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.2933,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.37575796216629953,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.2764,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.3594457735784178,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3019,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.3453815935278019,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3212,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.38245388082217985,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3187,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.33795605948986657,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3013,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.3429127821095667,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3071,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.3913053197265062,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3031,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3323032734090706,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.2865,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.38841751239201106,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3008,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.36406446434031964,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3117,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.39610926129177165,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3088,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.3546700794733638,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3059,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.38048052924372405,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.2891,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.3651973521308871,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3159,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.36771172991437906,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3025,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.3510464067025229,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3035,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.4047527108381044,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3198,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3491059064642846,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3091,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.3758065624055233,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.2741,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.37075684459791197,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.2867,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.37315173396299617,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3109,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3387807608598699,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3127,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3589941781766907,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.2876,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3774729323968865,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3206,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3399807905853028,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.2833,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.39814946671695156,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.304,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.34354933605334637,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.2971,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.3894450254570818,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.2845,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3451119900840926,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.2859,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.4034592843560877,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3133,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.3615085731448513,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.2921,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3475472495061407,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3142,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.37695773745996114,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3328,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3723882359319822,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.2934,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3508795465497479,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.2866,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.37328911636136514,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3069,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3670434625721506,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3113,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3955935236512198,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3052,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3406985330055972,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.2997,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.38226262480648454,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.2997,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.351508156943178,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3026,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.3502574367573885,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3054,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3673896478060117,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3105,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3453298623645241,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3024,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.34973239065707407,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.2963,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.37686181832916404,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3012,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3522881028104781,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.2864,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.37744097731418474,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.31,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3412179337586249,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3019,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.36536576004444454,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3065,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.36389565477657687,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3118,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.3570110192397364,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3097,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.35245979590983173,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.2762,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.3786230303066388,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.315,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3867791274583866,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.2884,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.35198100434132656,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.2953,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.3750888546973525,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3159,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.3566634719768836,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.2876,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.357270005954124,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.2956,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.34754920897949865,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.3084,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.37745710445473335,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.2931,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.339871969259998,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3019,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3546474372440999,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3036,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3287408133755845,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.2945,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.34861676611129727,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.2944,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.35139484605884796,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3048,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.3448602006387764,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3088,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.33638592760843405,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3075,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3385601048902238,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.2801,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.35378747625672646,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.2842,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.3370871271342426,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3034,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3440588523216146,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3082,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3274525203059856,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.301,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3202123747774165,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3093,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.339288441502347,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.2996,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.34546293510617204,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.2938,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3338472538413915,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.277,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3296409179195078,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3017,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.3963272674018254,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3118,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3273759975890379,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3053,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.34583619439215113,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3242,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.32317659444304075,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.2972,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.34847378718131444,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.2762,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3429672678972233,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3268,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3224970702959143,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3048,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.35401645705864415,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3171,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3473508262681078,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.2896,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.33935041808266914,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3094,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.36080128472855416,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.2968,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3201125761232248,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.2833,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3422587763128137,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.2967,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3186023200323988,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3109,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3396626384578393,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.2828,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.31912706156775383,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.2802,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.33858832754637075,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.2818,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.36443295036272044,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.2962,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3533052039757116,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.281,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.3253862006546905,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3057,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.3841640758827502,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3176,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3661400962025709,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3003,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.33458618145491353,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.2997,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.4330429180296713,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.295,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.3244099122287424,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.2924,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.399632026573939,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3288,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.33998352117138003,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.2932,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.33439809654623076,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3196,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.35770839932843296,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3034,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.3443962925184109,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.2984,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.35980871603118764,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.2905,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.311746278928668,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3144,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3244144266625734,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3121,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.3285954052012885,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.2902,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3177432599502493,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3219,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.37052815421573976,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3147,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.34141479308250516,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.283,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.3311106891766888,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.2908,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.31874174811481043,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3021,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3388360567103506,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3203,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.33465659822662647,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3031,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.34812700163632454,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3265,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.3621335994425166,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.2843,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.3315716557498456,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.2874,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3863343203224552,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.2847,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.29912347466661177,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.2892,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3267984205152978,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3047,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3922904436240387,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3111,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.31147436674716705,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.2832,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.39000702881803845,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3257,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.3555344802568603,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.291,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.3117520108890136,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3009,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.36624903731032826,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3152,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.3589927914405929,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3031,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 1.946648387402122,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3557,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3498193351277658,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3076,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.3672065930923676,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.298,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.33132361251049475,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.2599,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.36479254822179047,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3022,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.3734361718775305,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3065,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.36063296742218764,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3106,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.34801965410698343,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.2885,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.35634065451333746,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.2859,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.34409305350162966,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.2882,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3382591563084022,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.2936,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.3357812594727726,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.2971,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.34289047104658676,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3029,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.36096667622211387,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.2963,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.34046382403153985,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.2911,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.35573893792701883,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.2678,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.34623525845256276,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.2887,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3327672527621622,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3136,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.35393063387651863,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3091,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.34693445340274726,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.2866,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.3428746048843223,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3015,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3720116752200264,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3049,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.3234517515077685,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.2988,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3433405382212526,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.321,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3646253108003175,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3021,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.31500923208189,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.309,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.3285415282644879,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.2865,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.35791150516412357,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3166,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.32456195197588644,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3059,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3538414596885373,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.2768,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3480937146551145,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3156,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.3470189988666378,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3064,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.35247841314422834,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3029,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.32644174886981564,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.284,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.34749949862608753,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3127,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3480459675656848,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3082,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.32040559547877956,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3106,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.3191584897283473,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.2751,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.35791549651432775,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.2657,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3586993443890776,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.2992,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.3145650449423859,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.2966,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.3336447594217652,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3256,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.335512254990166,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.286,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.30990820616039166,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3464,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.37069812539531627,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3266,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.3087541058289086,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.2987,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3196668161183661,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.2802,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.3311797678269953,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3168,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2995403678633752,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3002,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.4769436891243227,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2637,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.32830037915377175,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2346,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.3915364127205246,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2179,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.5505936780930261,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2234,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.3822378603760669,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2307,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4459089097738581,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2169,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.4249717129228513,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2199,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.34843048176052266,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2124,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.3456386588064554,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2406,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.4111137000922857,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2374,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3843145273076816,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2205,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3715333211347172,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2375,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3826440109004691,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2242,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3392939698691791,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2285,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3302351192158396,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2121,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.3507587719979627,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2443,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.2979336000162868,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2152,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.32382387915392297,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2066,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.34547616854029634,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2185,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.2822770023587261,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2045,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.31804237505141747,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2173,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.30023570675497785,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.1986,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.2957545945390916,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2022,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.32577292127785057,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2267,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.2976940960941164,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2362,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3129368818763839,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2048,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.32022258746677906,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2253,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.29672856445479306,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2003,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.30998223236014943,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.213,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.2883823608207196,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2092,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.3052733302747884,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2259,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3029719317371634,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2331,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.29337324084311794,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2255,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.2980236580054929,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2284,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.30782761719637364,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2273,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.31561209677439894,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2108,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.2881081696703155,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2113,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.3064051115142653,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2029,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.28558214735128157,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2198,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.28836094434137516,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2156,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.30840366799593233,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2299,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.29918323569922695,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.205,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.2923027323081819,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.226,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.29885899464134413,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.1958,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.32290763736102424,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2407,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.2978390028056818,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.1973,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.28962020777223974,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2195,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.32617637350314876,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2155,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.29059073236878796,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2022,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.28293277709395065,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2333,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.307400148248732,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2174,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.29996452006532753,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2093,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.3096471024251485,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2071,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.283422995371418,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2212,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3105089188008118,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2313,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.30271436415722636,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.1897,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.27290186646556425,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2171,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.29512330754187105,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2152,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.2788907687210559,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2001,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.3029907232141767,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2107,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2945128796911709,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2026,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2849019738474327,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2122,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.28477348384883105,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2071,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.28101012619945454,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2374,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.2924268359762821,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2087,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.27988786472013244,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2061,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2834073177821463,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2172,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.29503038866634573,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2449,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.27817589777041074,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2101,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.28531471378438134,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.1934,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.28238930392741496,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2264,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.2819782446580212,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2288,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.28001065583213874,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2327,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.2957697473546588,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.1913,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.2849145830381724,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2102,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.28376461694575605,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.1936,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.2799434110339242,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2092,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.27263748387981007,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.206,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.28305862878309657,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.1995,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.28152964425867877,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2201,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.29451189212575174,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2205,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.28257049054945654,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2069,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.277775381577879,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2053,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2891515332303084,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.1993,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.30962716092921955,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2153,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2779953053369515,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2111,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2865056113372746,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2166,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.2909219811786107,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2199,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.2675653772530739,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.1929,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.29968925433046234,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.1934,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.29143742690162866,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2246,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.29843470212047224,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.204,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.2833381411067846,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2078,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.2756887846843509,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2208,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.28407433024358664,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2059,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2907904485032671,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2038,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.27950087286514397,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2329,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.2843995061742193,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2295,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.28719118830377716,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.222,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.2750426668153478,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2267,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.27569432924207116,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2134,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.282079240501513,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.249,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.286052848008598,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2095,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.2752977590574467,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2303,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.2792669898263883,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2128,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.2967928168868847,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.207,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.28663009453925475,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2276,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2754132369203718,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2049,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.27875387175877175,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.1913,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.2722338050574281,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2004,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.2900917896961266,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.1978,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.28670937603987334,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2281,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.28285533625178144,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2243,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.29556718944537774,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2546,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3038999654157776,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2136,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.28901725929007077,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.1846,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.28490038010828206,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.229,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.28621792477721775,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2112,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.28659930898777125,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.22,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.27762980871766246,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.224,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.2771015209751526,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2174,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.27461679293191243,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2177,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.27734698972337474,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2511,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.27716084183888856,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2385,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.2982446314972509,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.1989,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.2763950600577214,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2134,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.2777419730719413,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2083,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.27647543722568607,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2126,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.27327357128445623,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.222,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.30479070133457403,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2079,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.2983359925846448,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2391,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.27684629646652664,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.1854,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.27367558992210017,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2048,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2700822871573334,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2026,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.2813716521794458,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.234,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.27130357978997366,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.1892,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 1.8588612963212543,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2566,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2789741153986464,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2323,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.27647071037718607,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.1932,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.2920792346826258,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.1734,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.29049451950121064,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2266,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.2715983987657835,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2183,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.27774291876387186,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2114,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2825048279640416,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.1967,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.29180013053558856,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.198,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.27696815029197686,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2029,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2847268575849375,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.1991,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2872943103719591,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2099,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2859152274650336,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2202,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2746725004902577,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2286,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.28260939485350595,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2206,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.2865443464344928,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.188,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2753719074303366,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2149,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.2696271025601353,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2204,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.275524923162696,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2319,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2730151693013441,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2482,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.2966180892721013,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.223,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.2691959740907546,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2091,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.27537090886354076,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.198,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.28323433053891106,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2174,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.2805358625688899,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2149,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.2808449542331432,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2211,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.27913503498253056,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2269,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.28338858315854115,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2261,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.2643644845737761,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2311,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.277405719114505,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2143,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.27090147204746257,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2076,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.27560811206812563,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2059,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.28549829066690957,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2226,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2753826505790694,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.1945,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2727522625549744,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2098,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.28411670102922726,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2229,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.30638435126028163,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2114,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2805469319921186,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.1868,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.28316339666819756,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2067,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.2697104254302244,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.1966,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.2868451677273567,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.207,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.31734205445709074,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2005,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.2769135002165194,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2282,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.27403175147144576,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.1971,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.27638945957061906,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.284166196784753,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.1996,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.292651194994975,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2042,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.28468046048487516,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2321,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.26885010287334393,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2197,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.2731564557252474,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2253,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.28741015221984445,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.1955,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2637299754172713,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2232,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.27673071483512446,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2061,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.2728899191212912,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2058,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.26636645584410223,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2242,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2682995006089791,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2008,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.2835851579966613,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2006,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2759638689155958,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.1987,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2868348712850842,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.1919,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.2705755948223476,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.1921,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.28078058043023424,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.1894,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.28439913586768384,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.1946,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.270914709834595,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2164,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.2823382012689793,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2038,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.28079884584321957,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2335,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.2833394160109509,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2085,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.2853403266271392,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2106,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.27769584439281,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2157,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.27888056697474467,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2012,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.28898330460368504,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2003,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.26692371383405683,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2026,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.26706675545729325,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2021,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.281636290649697,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2301,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.2679444250098464,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2212,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2671256534035338,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2122,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.27002297947598947,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2209,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.26406023439968934,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2206,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2750193813199199,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2251,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.26607930620052744,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.2625660937042746,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2223,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.26391789314443415,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.1855,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.2736190605636637,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.193,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.27891249602737295,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.27789099816863255,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2272,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.27083159058050704,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.1975,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.26118640977699176,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2018,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2655485179086376,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2249,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.26985400708145396,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2001,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.26923219744639365,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2217,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.26790687938741037,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.1929,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.2658618756395732,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2188,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2722066563346569,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2393,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.26549319973341756,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.239,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.26586194656964374,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2208,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.2668268920372592,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2377,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.2692636564336352,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2067,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.26479359944309855,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.203,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2663267802290668,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.1998,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.27555741386270594,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2176,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.27141001285181704,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.234,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.26798990695941916,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2195,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.26333319366173114,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2171,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.2874276141030996,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.1952,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.26175512651364446,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2285,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2726712433109233,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2236,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2632710407904636,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2119,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2673022842918487,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2243,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.26834649257212634,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2096,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2635727764829732,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2158,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.2677117556980114,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.223,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.25655174113500084,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.1985,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3987131440463667,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1886,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.3371641377143464,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1454,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.2368474220304081,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1662,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.2526428995712956,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1704,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3591641811987878,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1793,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.3852809062263988,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1802,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.34967682540267575,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1767,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.25464802673909,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1687,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.26321952706300994,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1546,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.3406708226832227,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1719,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.3292239346324217,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.176,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2619879542215014,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1655,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.22783054155172708,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.162,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2575137495337787,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.197,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.2878515773268283,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1576,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2986577122886881,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1776,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2523033074580569,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1529,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.23556635316658683,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1662,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2280627555703852,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1447,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.23774470243686516,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1719,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.24746001462909695,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1769,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.23893662619962705,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1493,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.24227639076320373,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1728,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.222596127787999,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1701,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2341410154424406,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1687,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.23193032428651295,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1641,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.24010972123799576,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1813,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.22211327218059193,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.1691,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.21608918708663852,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1321,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.22055531300616385,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1475,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.2169339487985144,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1543,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2344607272808242,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1576,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.21417513767661395,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1462,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2304611285177387,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1505,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.24022485753926237,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1998,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.23666287249068463,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.165,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.2291849910079479,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1552,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.2291107987627164,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1684,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.22022455429485827,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1754,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.23209062192442892,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1734,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.2247923666637479,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1714,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.23644224898072772,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1426,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.2235008222320198,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1486,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.21882538802956278,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1549,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.22546354672032637,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1805,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.22085387867637887,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1725,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.22336135343945177,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1724,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.22266444019061193,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1929,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.21767182141027605,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1678,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2122838250701992,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1591,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.22089703561836022,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1957,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2219663044674336,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1702,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.21614364308703585,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1311,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.21615744348322685,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1518,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.20560689862880324,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1769,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.22904493905549558,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1759,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.20794468168143776,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1399,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.22883908424021715,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1931,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.2295003715964355,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1695,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.2246923602284337,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1609,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.22666219396684817,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.151,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.22162567963601085,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1487,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.22348892548821644,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1497,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2148662661704708,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.161,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.22318822589178097,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1573,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.22565139512871973,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1623,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.21342973645206606,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1734,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.21528852450953762,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1421,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.23022154216640114,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.168,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.21731014444452973,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.1547,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.21191229167874578,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1718,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.21876473986096112,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1927,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.21166768088401225,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1756,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.21847474203408343,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1747,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.21173911472522264,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1395,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.22090488700008182,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1736,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.21833183366166856,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1565,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.2346085389917697,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1585,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.21192343848926307,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1544,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2263980445155554,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1811,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.219060816793228,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1782,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.21975099011112786,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1512,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.2160275415745471,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1572,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.21738489808582784,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1742,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2130546517273821,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1656,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.20887118446611827,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1401,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.21919447876477466,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1534,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2208850928362448,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1632,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.21419657122260516,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1754,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.22023076688316345,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1642,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2252559139909725,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.164,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.2174233481778527,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.2015,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.21396635355650362,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.173,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.21674016009208574,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1791,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.212499783684834,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1647,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.22951051128884278,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1845,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.2126086114034368,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1635,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.21769889901552367,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1654,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.22124166763685255,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1654,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.2831960061647459,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1784,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.22109717381346708,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1442,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.21314700073296391,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1856,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.21496242734222493,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1592,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.21710819065544928,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1513,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.21602987442990773,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.158,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.21603887490014095,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1856,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.22748984866571292,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1648,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.22302228685705086,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1696,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.21813899615483381,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1581,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.2143673891302977,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1842,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2158205576913346,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.17,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.21737694595817417,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1621,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.20695480286743476,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1486,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.21671383891188278,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1728,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.21661005279944806,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1768,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.20559269599357505,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1448,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.20828065204133334,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1795,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.22725382016583018,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1846,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.2114152979457488,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1414,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.207565201457486,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1424,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.2082183273981369,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1712,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.21411473384292248,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1897,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.21625880101035663,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1412,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.2291828787053114,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1614,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.20561313917858284,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1479,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.22763783548326147,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.149,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.2197798480406835,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1571,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.20892056233581288,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1696,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.21236147235909392,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1599,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2129768646390587,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1706,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.21013020758998557,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1578,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.22784036564756957,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1606,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.23496468585547683,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1564,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.21522964100932018,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1749,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.8392480469054394,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1732,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.2138089309845691,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1573,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.21434425443933916,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1526,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2156969614274668,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1437,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2112028784001703,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1657,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.20928393473770274,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1653,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2153294127380816,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1914,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.2172782455697935,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1533,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.2124742832470036,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1701,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.22167317770435938,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1774,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.22848006845664334,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1616,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.22700105095644205,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1429,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.21918707334795515,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1715,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.22067601120251654,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1583,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.21470881613106516,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1662,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.2237839231864565,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1537,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.2104126724353705,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1672,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.21447640352868166,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1666,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.21061434625729775,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1631,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.22561312505895384,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1924,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.212942932410705,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1533,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.22019206448776835,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1489,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.2149611904856874,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1641,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.2234382802049065,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1509,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.22962462135017467,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1534,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.21214194383953167,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1912,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.21893542055853962,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1555,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.22154976144686053,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1661,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.2310118145239414,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1773,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.21835183266418423,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1622,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.22585585924313342,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1654,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.21047040762468155,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1532,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.22697239870681976,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1565,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.21925062128816947,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.152,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.21184077214680247,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1546,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.21741849945238254,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1373,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.22668744664764895,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1923,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.20720319908828871,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1464,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.21376935623769325,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1565,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.21506112223891755,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1772,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.21625856179729627,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1444,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.2128585171852745,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1585,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.2253549889843707,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1641,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.22302847507964035,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1526,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.22218168824821588,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1616,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.21096841506695524,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1746,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.22419885228058836,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1639,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.25159853489722267,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1481,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.21547673810898457,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1763,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.21922612056423335,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1519,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.21042740819417446,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1509,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.21928935436188912,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1568,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.20606751715148783,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1654,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.22640984199568132,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1623,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.21722636570800882,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1568,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.22462845786722618,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1647,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.2145797477944711,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1774,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.21651872100490174,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1714,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.2259981718084829,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1909,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.22166937520614247,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1769,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2255951153734374,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1456,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.2226120013537009,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1734,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.2219413150207592,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1723,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.22298495770591756,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1736,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.21574063859552547,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1484,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.20975432430527013,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1812,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.21615091370949394,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1488,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.2203496251288906,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1659,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.20852232374338373,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1578,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.2156181617056093,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1615,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.22854313937141454,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.147,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.209087515593992,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1682,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.21757258547300573,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1515,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.21294428324540135,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1808,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.21616063199437355,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1552,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.2199142250704388,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1632,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.215691542495429,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1543,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.2144772766175735,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1734,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.20907575571372458,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1453,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.2179762556066765,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1776,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.2256282640104185,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1791,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.21591854453735873,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1698,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.22860597005504935,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1754,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.20990355176000683,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1739,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.20774372684078524,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1897,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.22136310160235848,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1549,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2240287211785961,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1573,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.21324853166878208,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1693,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.21255359575704627,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1663,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.21123887787374165,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1625,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.20786217354808942,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1752,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.21876813903486553,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1714,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.21028314229214046,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.15,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.2087976965982052,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1781,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.21679161820896187,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1819,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.21168215274738658,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1884,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.21214312646599467,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1747,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.22879579878382011,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1901,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.2144863806817987,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1482,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.22111815733951687,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1564,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.2023976151438614,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1616,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.22033142638233474,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1551,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.21993292453109373,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1662,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.22182239670645953,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1591,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.22401411542515226,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1691,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.21884823946244572,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1492,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.2101065183067928,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1436,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2076401791684069,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1591,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.21358221989770504,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1614,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.22812228039217886,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.165,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.22439442263445364,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1423,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.210835625974454,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1541,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.206810626305833,
+      "learning_rate": 0.0,
+      "loss": 0.1716,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 2.190159325803053e+18,
+      "train_loss": 0.32256396708459506,
+      "train_runtime": 24829.5428,
+      "train_samples_per_second": 6.363,
+      "train_steps_per_second": 0.05
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.190159325803053e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}