diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 5.101187629638192,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.9119,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 5.480307785813243,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.9083,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 5.012807712445904,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8471,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 5.409912036468613,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.8618,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.163576321239627,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.9193,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 4.683818117327167,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.8629,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 3.7956109616774745,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.8307,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 3.755778104339143,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.8883,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.4278525095119052,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7984,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.188475441105332,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7807,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 1.8477003374996563,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7288,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.112880331703768,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.7517,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.2769613395096164,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7612,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.5162752106373003,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.808,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.27812301790997,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.7723,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.9301613479597743,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.7386,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.51821772016618,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.7364,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 1.9884869761059056,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6883,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.625469805409786,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.6691,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.5356993305245876,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6716,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.362995098159882,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.6904,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.7386871217978872,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.6625,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.5507322073070882,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.6757,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.3379957257521407,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.6471,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.1836329204638547,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7095,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.0686888820695997,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.6286,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.068030296426337,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.6131,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 1.3255709176580492,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.6436,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 1.0071230798275725,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.6299,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.7565828487228352,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.6454,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.941585568967656,
+      "learning_rate": 1e-05,
+      "loss": 0.6086,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.8599707162372204,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5828,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.8271263841508276,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.616,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.6833770635245192,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.6063,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.8001405096729182,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5738,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.8242842210915463,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.5655,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6768932809766068,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5885,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6921469639051533,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5629,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.6899172975118257,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.5799,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.7460355919148276,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5709,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6350157656002162,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.6152,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6436225149417872,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.6091,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.626241483209601,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.5965,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.6413898925087788,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5839,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.6680337237077099,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.5546,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.6175980363383253,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6021,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6944381086542009,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.5755,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6502764883649751,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.5589,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5691189617999871,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.586,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.6656706793602494,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.5989,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 2.4121662720842245,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.5541,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.7263494203826785,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6064,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.708311500809212,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.5813,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.668179471256152,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.5115,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6266599709826962,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6014,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.7356625672621931,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.5769,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.663469224904909,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.5808,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.6898472779015915,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.5615,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.6914293815088469,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.54,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.6782332544065784,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.5744,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.6222206102979762,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6141,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.6464448381413793,
+      "learning_rate": 2e-05,
+      "loss": 0.5697,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.6931028638696662,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.5858,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6037030439316055,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.5661,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.6703602997692155,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.5288,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.8583834543664235,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.5906,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.6616896943206243,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.5753,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.7540505953947924,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.5632,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.987598928073662,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.5881,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.7094524691660827,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.5365,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.7838242700834965,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.5879,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.7038380007715208,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.5572,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.6581188559525496,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.5638,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.7566126741199174,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.5597,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.711517994723536,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.5869,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.7479136593942446,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.5567,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.7480663410329599,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.5603,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.6935728850580072,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.531,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.7255602870873896,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.5608,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.7069540672907481,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.5589,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.712793715054933,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.5415,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.6880790436800625,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.5279,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.6042890346272992,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.5604,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.7797110642390426,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.5728,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.6790313462940362,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.5645,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.7719060669773435,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.5901,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.6366841234099511,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.5572,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.7060327489320907,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.5558,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.7442530535594719,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.5616,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.8095067535324489,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6334,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.6658826164928786,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6037,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.7832795838224306,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.5673,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.7052236185560826,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5523,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.7439323496689378,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.5255,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.7447927915965278,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.5595,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.7537520759955376,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.561,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.840550709584267,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.5759,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.8744664040530535,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.5794,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.7251841305397388,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.5323,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.9376172710677466,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.5646,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.7759003121808495,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.544,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.9583600762380247,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.5462,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.6538174762895367,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.5358,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.8967200250889944,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.614,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.766970004783781,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.5918,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.5786483169786726,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.5007,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.6615416474630557,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.5384,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7491726431797121,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.5619,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.6067135683917354,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.5139,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.762667685421317,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.5484,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.6779088084678363,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.5335,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.6653800973167735,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.5654,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.762438649275861,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.5258,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.6459167796684671,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.5468,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.8118605384907647,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6016,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.6685751317177683,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.5893,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.6500679782925579,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.5042,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.6670559621542543,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.5922,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.6484308548343136,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.5285,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.6056393079531084,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.5408,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.6140855939231213,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.5483,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.6527657931771281,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6144,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.5676385503925789,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.7581,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.6118527477649961,
+      "learning_rate": 4e-05,
+      "loss": 0.5446,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.5695415623573012,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.5494,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.6631800065569651,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.5733,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.6137930379624456,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.5302,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.6928965963649811,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.5687,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.6733021088627311,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.5569,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.6488435113616035,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.5478,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.6981689733692931,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.5664,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.6571609292104544,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.5322,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.6290077587095274,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.5724,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.8084799783300967,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.5776,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.5326878464698889,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.5601,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.7665514588081452,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.5436,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.6179055396032975,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.5278,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.7693080742927265,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.5393,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.646358553614155,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.5784,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.6912499752160559,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.5516,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.6841949038592663,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.5464,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.6637779506308262,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.5082,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6406791053946291,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.5962,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.8254119970494566,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.5566,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.7679275693413233,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.5601,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.6773818035247516,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.5523,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.7942264008204825,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.5865,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.6580311667485585,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.5685,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6455021973004359,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.5511,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.6054181052149884,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.5432,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.6778763405217887,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.5637,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.677135044203078,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.5766,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5998106648025874,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.5257,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6762224615784155,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.5283,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5795963969135711,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.527,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6635630909858249,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.5269,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.700824078985827,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.5355,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.7348379730496107,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.5326,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.7280737764561843,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.524,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.7293514606521146,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.5558,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.7364363779841959,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.553,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.5709054902145102,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.5153,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.6625181946511295,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6025,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6062463390472479,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.571,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.5530148139493792,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.507,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6156979684127432,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.5632,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.6143830121551923,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.544,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5808274477241734,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.5613,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.6208850112398347,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.5604,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.5386789773638343,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.5783,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6143226928042035,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.544,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5818370347869375,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.5669,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.5550151333330802,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.5836,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.5979836299209085,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.5891,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5708031587311684,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.5631,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.4883123937628846,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.5689,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5777186955335691,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.5608,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.4824040944712554,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.5436,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.5117260531767479,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.5437,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.4831845954424775,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.5241,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5892534722211418,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.5308,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.4766432383027191,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.5539,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.6360125943933402,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.5764,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.4852385730723131,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.5207,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5234530823453668,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.5494,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5311300753975025,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.5258,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5746727648248549,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.5838,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5404263983022687,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.5669,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5747223708543007,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.5496,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.5783352541169341,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.5846,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.6372131379274535,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.5655,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5956941498573118,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.5682,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5383771710957815,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.5318,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5566877528323518,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.5196,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5393232646248701,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.5214,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6380477083002719,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.5624,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.5574860917178234,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6035,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.55365014226044,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.5119,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5284019430164218,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.5341,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.5209867520685834,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.5436,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5907805890779676,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.5698,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5921357117050527,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.5815,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.552269409203369,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.5586,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5133084406871975,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.5353,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6241007814144196,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.5729,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.504002940226415,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.5388,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5058967885204597,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.5393,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5855646883561032,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6044,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.5461746997003701,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.553,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.5248302618751698,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.5396,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5659636758303008,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.5491,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5929290641710135,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.5353,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5427633747148011,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.5278,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6264462192139747,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.5588,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.6025361704857332,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.5409,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5424057194125151,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.5463,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5465601892346681,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.5458,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.6266931835819582,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.5515,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.5221030291751929,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.6271,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.5577503979253242,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.5518,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5367991607286429,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6731,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5330745820144142,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.653,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.5286353180423006,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.5599,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5167783619630978,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.5455,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.619339829167579,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5243,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5663174753226218,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.5556,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.9022738445401142,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.5661,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.5659355194439609,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.5275,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.60048283810232,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.5492,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5671666972997026,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6081,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.579055492825574,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.6149,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.49809668291405973,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.5311,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.4976096166179712,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.5448,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.48325651103902795,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.5354,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.4977246417020161,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.5834,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.5482105483651482,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5821,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.5655302639338089,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.5494,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.6368278752118361,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5743,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5427219505809963,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5453,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.5336131210398637,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.5301,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5481761248574069,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.5324,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.5625554040330071,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.5467,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.6174756323566272,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.7143,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5553506375128154,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.5202,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.5417819367748766,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.5149,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5467611746707036,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.5822,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5190008065670285,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.5394,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5440497632786402,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.4597,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.4729656822740093,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.4763,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.5099899542458664,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.4241,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5613839382488569,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.4407,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.6200654799362902,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.4607,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5070696391957435,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.465,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5901950403188219,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.4396,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5488751289965994,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.463,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.556160576141801,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.4476,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.5979692245259443,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5021,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.519922371407779,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.4237,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5258440480443787,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.437,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5313936492081838,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.4184,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5811314900451036,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.4375,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.5359745949340696,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.466,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.49042998292856366,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.4614,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.560401853224201,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.4176,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.48124537894763897,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.4199,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.590397729622423,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.4729,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5335808600649573,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.42,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.5706015583946603,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.462,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5345156480243248,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.4314,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5288275867981239,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.454,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.49528316753314233,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.4409,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.534911390264527,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5077,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.45636006426953074,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.4168,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5661579334740344,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.4463,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.4481449414253138,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3781,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.560250011174283,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.4202,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.519361572811071,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.4109,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5468600140413913,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.4084,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5463657170896388,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.4526,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.5854842266237736,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.4557,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5257307393335355,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.4546,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5123110615988998,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.4157,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.52639097034319,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.4378,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5034196665821664,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.4398,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5240811794292316,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.4357,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.49452318925539557,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.4384,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5201624390817344,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.4467,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5429080291708731,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.437,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.5255744150857099,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.4484,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5681227627720461,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.4616,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5165260693281303,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.4427,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5311096987061209,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.4316,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5637872900006309,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.4871,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5344886793846687,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.4708,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.5353097984000121,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.4019,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.4357646288593864,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.4345,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.5587625983412444,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.476,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.4587169744962717,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.4327,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5620731209482562,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.4432,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.4803027239512539,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.4497,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.559989253280443,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.4397,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.543680429416339,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.4372,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.5143105876322698,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.4648,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.43839693811339353,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.4515,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.5964731722640697,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.4423,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.6626994508073296,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.4631,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5333526000847993,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.4724,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5872735046344769,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.4425,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4937236268107708,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.4538,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.5774494229264644,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.4591,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.5937090070907962,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.4541,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.5921551948436022,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.4296,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.5526763485321696,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.4634,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.6205474860323085,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.4449,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5493372890026617,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.4379,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.6260662451774148,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.4722,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.4937683525856568,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.4197,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5880249294359458,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.4649,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.5533486941844367,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.4426,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.5466419124498051,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.4493,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.5470461644326817,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.4571,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.509222279383723,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.4221,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.5284095397546729,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.4627,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5095824194848539,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.4473,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5766653440237212,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.4254,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.529629660939448,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.4118,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.5872708956279433,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.4843,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.5061179674790431,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.4441,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.48418343248108914,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.4191,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.551972319848578,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.4611,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.5935900731919582,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.437,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.41495868056733937,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.41,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.5407851366218634,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.4408,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5391173909978446,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.4267,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4665710744556268,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.4335,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.6651891634803456,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.498,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.45624269285556224,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.4608,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.5657253734281229,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.4458,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.6084413710579432,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.4241,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.5521716444335524,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.4753,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5265678432862689,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.4057,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.5140674407208854,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.4597,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.4523170544409755,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.4824,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5232033149231547,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.4322,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.5668640032688712,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.4587,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.542100773178159,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.4285,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.49677957374566767,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.4348,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.7124600438237555,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.4597,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5677882321929885,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.4473,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.5484805528117771,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4449,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5253655890023703,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.4425,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.4402071827073809,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.4675,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.4672008599844658,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.4445,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.42699101647092724,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4381,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.4922022564124576,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.4683,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.41715528578258265,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.421,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.5368396665837039,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.4684,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.5224085205214192,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.4353,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.49595718238187164,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.4268,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.47811858149194936,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.4696,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.48042386980040247,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.4338,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.5789637587236054,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.6436,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.47822677509429373,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.4245,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.5433828537390856,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.4786,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.43571186884948593,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.4397,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.48533661583219595,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.4638,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.5146025806017951,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.4356,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.447068217139121,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.4372,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.45917738179630374,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.4474,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.43143453087218864,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.4658,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.4650761220124989,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.4657,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4111484417674356,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.4367,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.45647527631204043,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.4173,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.4211253754631651,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.4506,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.45723859762129576,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.4511,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.5416754351784283,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.4652,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.57691517894684,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.4235,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.47834908307132096,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.4273,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.5392035725994715,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5078,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.5603902812279634,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.4733,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5267955992796316,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.4422,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.47089911190026046,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.4235,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.4640144509527867,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.4543,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4375116917666556,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.4456,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.4073169132154579,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.4035,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4666897792061538,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.4467,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5237483052264494,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.425,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.49806231114601096,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.4618,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.49764227762805385,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.4472,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.47319382001859867,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.451,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.4246651714255558,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.4487,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.5469242985885995,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.4566,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4353280272165594,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.427,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.5242748491614174,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.4621,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.5672379617660569,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.4415,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.5081803854913015,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.4562,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.5442217900423847,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.4527,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4627017611862095,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.4405,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.561788029960994,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.4572,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.47934094725548887,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5623,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.5103878965317021,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.4351,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.508387496977771,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.426,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.4708148101129489,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.4304,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.5250325655343953,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.4171,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.5074835344465612,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.4606,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.6076339054517019,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.4829,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.48590758834633363,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.4256,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.5197177769433801,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.4633,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.5030676954835416,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.4784,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.4634938239803247,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.4534,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.5800196539263762,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.4651,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4858678761299365,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.4407,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.5515134648872646,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.4595,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4546387882683665,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.4369,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.4465345192126883,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.4746,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5253388421658903,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.4859,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.43717136011799657,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.4319,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.4465914299844491,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.4684,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.4596220962436299,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.4687,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.5315779487950792,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.4466,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.39689319585625554,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.4125,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.5254703526221163,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.4464,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4695609609523173,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.428,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.46616607515648706,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.4529,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.5011313603554297,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.4484,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.45577118847669557,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5155,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.6134494198490631,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.4573,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.46580577432579506,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.4627,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4569779439982075,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.4248,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.46852653889411955,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.4536,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.5141498688917732,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.4406,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.5210365750267894,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.4413,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.4449606173073473,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.4473,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.520810270940984,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.4475,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.48800742182615625,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.4514,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.4464442370511691,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.4499,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4460959858729854,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.442,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4081058673087995,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.4287,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5001310048007548,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.4541,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.47555999545851324,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.4261,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.46184589315505703,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.4533,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.5141336011588034,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.4766,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.5542424008178706,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.4171,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.42489478802340147,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.4283,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.6550070375767546,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.4726,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.48220430304395384,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.4688,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.43347339721291145,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.4512,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.5191061680935967,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.4618,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.4734188637258068,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.4549,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4875709153293266,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.4574,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4965509976918136,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.4357,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4457114889046888,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.4455,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.5722991983935789,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.4391,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.5408959909762165,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.4395,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.5043216588584911,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.4157,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.4686650095277997,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.4545,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.4700436332685083,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.423,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.45273327519090273,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.4681,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.5478779841657703,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.4323,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.5671105478125007,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.4254,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.4533075882968814,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.4243,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.5312329310547187,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.4256,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.5197512130580111,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.4718,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.4593094232238045,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.4279,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.43746879604365807,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.449,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.5681390367280915,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.4622,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.5170673821514317,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.4339,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.49835220927095203,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.4795,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.41285859226734295,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.4334,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.48848352309968573,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.4431,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.3896514384861204,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.4117,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.48037743657654675,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.4421,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4258990805182417,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.4497,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.5737901184163744,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.4591,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.4342504804871862,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.4454,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.43041900979980513,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.4342,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.5159962252958964,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.4198,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.44116124072664104,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.4438,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.45293192177815444,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.429,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.41689035375976174,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.4614,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.4970986889375652,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.4536,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.36820739545940534,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.4101,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.4903475637592479,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.4367,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.5153418839071345,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.4258,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.4391279197650838,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.4689,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.5414282698661074,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.4421,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.4629223124142633,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5903,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.4385303314416403,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.4128,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.4304284239516942,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.4523,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.5318386720465119,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.4226,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.4894012117462827,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.459,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.45217859521318493,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.448,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.5163989951099908,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.474,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5071580302146492,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.4394,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5003187895851682,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.4348,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.9105755184078933,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3479,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5007129815384022,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3134,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.5688894242985764,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3547,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.5245326623656764,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3374,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.4508384659718969,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3224,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.5720297668078177,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3215,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.522690643891135,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3317,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.4747420805358719,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3148,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.5362677105392062,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3293,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.4475666560101273,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3356,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.5618439450484333,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3453,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.49080581102270476,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.331,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.4982157519991453,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3576,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.5338352474665323,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3237,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.5726065036108638,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3285,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4245628946600939,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3159,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.501913688620042,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4108,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4687932468269026,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4788,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.47515002029861725,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3004,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.48411370963566597,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3215,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.45881861950818215,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3145,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.49185587864850194,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.4232,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.4932329774223246,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3127,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.5571568908729553,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4404,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.4142339702694579,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3206,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.4863324636340336,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3429,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.444366844487894,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.2968,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.46705214300486786,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3238,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4997033445758238,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3074,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.4043928535365291,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.2904,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.5335228425116164,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3352,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4497731797008325,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3169,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.5542235516282984,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3252,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.46135644162215583,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.311,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.5037519146368804,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.318,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.5930453875491395,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3307,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.4198331039194153,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.319,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.4644818469985103,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.2971,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.4855082527363367,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.332,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.524647944934238,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3573,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.5361650277450857,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3499,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.5685376837655163,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3269,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.4693337943973303,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3031,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.43526640249805915,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3255,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.45867919583668587,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.2976,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.43391125492087135,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.42295059470118607,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3504,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.47198256785709813,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3451,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.43902286368864135,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.317,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.4213635463143576,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3299,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.4938709672558501,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3624,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.3812626189497815,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3227,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.42875865775959304,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3435,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.4379377438950085,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3326,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4540264245443818,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3072,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.5018681412661488,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.361,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.4336584612332583,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3412,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.41467164248481614,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3281,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.4242925357985655,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3358,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.422501886090224,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3348,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.5240913233832719,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3069,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.48126912167347907,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3283,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.528939507566257,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.2996,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.5144399228211889,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3189,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.46689491784174936,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3296,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.5407858625088549,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3465,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.453497193370516,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.338,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.4302887764412078,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3241,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.5739473024261663,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.5591,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.49859402232124816,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3534,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.4580153379846647,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3257,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.41237971426831355,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3338,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.42299734968550645,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.304,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.4960327899179331,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.2952,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.43324077913250575,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3505,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.46492324835204546,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4015,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.39234575882142797,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3541,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.4173744846815931,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3347,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4133250827062887,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3362,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.4135003277482448,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3226,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.8137492980379814,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3328,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.5130218444483808,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3602,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.43689339188498755,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3002,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.4721473126399638,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3191,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.4772664209247985,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3345,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.501696316846425,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3229,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.44272750370278546,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3167,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.4361055507846564,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3406,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.7980117271469821,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.334,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3654248225811533,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3299,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.40431553941680193,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3221,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.43806117610113937,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3265,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.4913601509005351,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3438,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.605508459796476,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3485,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.4661321528321261,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3768,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.45127356252285256,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3326,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.4383336826854905,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3577,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3626250690665683,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3297,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.49955393724490554,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3381,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.4460919853978422,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3166,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3982505689349963,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3083,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.5340900221424131,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3429,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.5144720558317543,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.308,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.4752405332204803,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3464,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.41087640513817836,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3175,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.46358450838129805,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3247,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.4375909079432726,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3219,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.42148818265019217,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3303,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.47509561931176664,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3318,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.4457677158160172,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3378,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.405923262277584,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3366,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.4896530938713224,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3344,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.4910965488736677,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3093,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.43457421231430565,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3065,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.43028050468386597,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3252,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.444403244836217,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3339,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.41659535494644934,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3162,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.44402091417778755,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3343,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.38118310332040817,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3276,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.41955906760478373,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3435,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.47029088411217085,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3359,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.38472761050224547,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3314,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.4328406617513775,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.2884,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.45247493930902505,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3294,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.4067477579460405,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.333,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.4419253653208117,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3199,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.42679695113303445,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3328,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.4411265151404532,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3427,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.4653349609651511,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3228,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.4695857541556067,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3196,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.49142860537859023,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3208,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.48958500253076925,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3096,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.4263964778608344,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3403,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.42558481353653277,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3192,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.46971943245235115,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.31,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.44079951061543726,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3106,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.4511840936997099,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3436,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.4755667088960868,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3303,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.45090601717860274,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3546,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.485312091430238,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3635,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.52124452431562,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3504,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.48182352701059944,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3678,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.40147772974986606,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3426,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.47534119294216925,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.321,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.42877909581221035,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.314,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.4407319072029567,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3158,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.6138388668046959,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3045,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.4227520362280612,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3367,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3798274326028109,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.2996,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.47826625196074657,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3195,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.48350620736564937,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3103,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.5311047316915476,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3078,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.47926844445910655,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3276,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.38128891962069966,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3247,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3922833844983746,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.31,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.4627166975482907,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3138,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.44328049944971193,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3516,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.47370533570232015,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3265,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.397368315783819,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3343,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.4487883838121622,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3217,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.4403696851043197,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3384,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.417225378797329,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3518,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.4483553082013632,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3308,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.4368555698720753,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3524,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.39274796255411065,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3664,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.43907529935629314,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3536,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.47245667358233956,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3355,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.35527797024210117,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3578,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.4025540488586111,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.323,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.4053126764290083,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.4079,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.4337143476665814,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3222,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.5146447673810957,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3278,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.4718769521449096,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3375,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.45855404435194963,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3174,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.49143280400756506,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3199,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.45760471518392754,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3174,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.4392735255520274,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3268,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3921611834986734,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3383,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.4243882264862642,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3289,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.44035919258946615,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3458,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.40021127123078043,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3083,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.5117591225982498,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3538,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.4452104106048746,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3313,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.4791207171531045,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3063,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.5705347586165593,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3272,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.438284905350451,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3443,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.4295137094541398,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3293,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.43617915292174975,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3077,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.4680490362330408,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.315,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.5273374068175771,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.308,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3691175031663134,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3139,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.4617116930718978,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3357,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.405732194646844,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3237,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.4197204702269574,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3211,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.4270261966595454,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.324,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3783212174901159,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3367,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.4735648861793698,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3386,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.5260698466665327,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3672,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.438707086872259,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3218,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.4078297754812302,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3288,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.45390419937727194,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.5036,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.43383359337430955,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3207,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.4776470136374787,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.315,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.38496076790134426,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.34,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.7813140889779929,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3312,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.4279294002143851,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3416,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.36154110328702044,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3545,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.40022000677869474,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3158,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.4627122541324605,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3459,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.4997113613488328,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3568,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.4721999595453288,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3196,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.3965387526266225,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3346,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.4683326461889589,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3215,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.45377556447696316,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3517,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.38258312043353715,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3327,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.44298601551026207,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3436,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.3832319308330068,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3381,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.45473860079319617,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3545,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.40548615404280713,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3241,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.4049707542009368,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3182,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.47517412323045755,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3262,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.420354224969957,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3277,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.4418835722608143,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3614,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.41111152698954195,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3179,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.3943780890876943,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3511,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.4838594989399546,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.322,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.40934535703997343,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3516,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.43483963068012615,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.309,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.3410797543163726,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.326,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.42655188081030604,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3336,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.431908054336489,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3115,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.4171294867278995,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3466,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.4242080993125614,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3418,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.40638259478861904,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3325,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.4374339532813079,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3117,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.4191527454066604,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3188,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.39523164992717164,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3732,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.4508780703807368,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3224,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.5148477185583605,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3311,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.43045325869557005,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3261,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.4112965790549314,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3226,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.40487594383151015,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3279,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.49345573268201176,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.2958,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.39956731527800243,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3112,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.4280339746678114,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3242,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.45711972488186603,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3229,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5062735141292095,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2106,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3886502959939385,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2575,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.5829599957138042,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2394,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.4647443074356374,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2641,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.44862627385725545,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2643,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.5389359905458385,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2555,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.4732832115922081,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2598,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.45712444180864337,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2531,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.45703578767999714,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2333,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.4361882106539107,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2337,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3703293367756557,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2272,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3880353540623094,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.241,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.4280459202387578,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.1844,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3932402137286119,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2611,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.423138229577508,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2292,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.406052073656191,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2269,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.4512046908006928,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2578,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.4489830364832828,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2207,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.40875657072160776,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2495,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3803346652354699,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.1977,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.4457198714242928,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2243,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.39352759418572303,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2136,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.39973750155501947,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2256,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.41444235322503026,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2424,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3873862681548239,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2191,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3957465781004137,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2332,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3791791339430285,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2177,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.43489444797776494,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2143,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3646441595317313,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.242,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.42631981421062276,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2168,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.3768373872975192,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2502,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.394950692029929,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2211,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.42224756335985686,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2097,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.48797083459927276,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2213,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.41582252450073753,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2149,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.3803743501828939,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.253,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3927735348777565,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2452,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.4141704493439624,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2373,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.4440559206953959,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2041,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3912889091119431,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2237,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.4314121974957813,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2288,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.40015680399661985,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2455,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.38319590305644596,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.241,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.375774053452859,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2398,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.4094330825217414,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.233,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.40573066394390317,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2261,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.40492828865576824,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2319,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.36955650645706084,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2399,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.44428899058852683,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2018,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3680411675035013,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2341,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.37316141235435507,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.223,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.42848600465527503,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.223,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.3403009463085529,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2438,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.4121035040650398,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.22,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3989144100726755,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2106,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.39566859216113937,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2407,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.4314297499020856,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2424,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.4195157939787936,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2312,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.4441176532392023,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2168,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.42862680427934036,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2597,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.39223679500381486,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2207,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.39346734786808774,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2377,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.41919478337159766,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2318,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.3667920505632845,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2507,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.42024621829544434,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.234,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.40327295917031325,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2354,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.3995966458531143,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2802,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.38930941368801436,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2842,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.39819793420765454,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2164,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.4036841151339865,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.1973,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.40318691826362535,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2599,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.37902552340847123,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.244,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.4205790504930336,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2658,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.3736728159925616,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2175,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.4618021050244697,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.241,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.4174461945701866,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2237,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.36853612814712,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2238,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.378732084299307,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2406,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.41469937287021147,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.1982,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.43746314520544083,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2453,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.33728668250776206,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2603,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.3773687343888753,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2566,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.41271261050820074,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2433,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.44531937830687457,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2905,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.4089162748189101,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2136,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.39494717980776034,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2267,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.4299403690845539,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2617,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.3526282865764544,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2658,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.3507584721889251,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2306,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.4489207803911998,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2571,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.4384869237237623,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2584,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.3877668848661181,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.1937,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3973205169026797,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2339,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.4011213410852412,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.229,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.3890245675308267,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.217,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.39462788250097247,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2382,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.4041095068196608,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2273,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.39176183323834396,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2407,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.37182016883380997,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2112,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.4340170194179001,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2269,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.37249214307569617,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2412,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.39108076156745947,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2124,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3846465406934914,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.278,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.38144480097334355,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2267,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.380195698491585,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2296,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.35146154643170713,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2376,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.3726077163252541,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.246,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.36803924885646594,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2393,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.4263771632908619,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.212,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.39061771885314056,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2246,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.3953277701038039,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2411,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.47492494428105164,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2098,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.4381467834265376,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2336,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.44951381318308264,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2257,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3787342770874258,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2183,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.36276501678614004,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2288,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.3787888043920065,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2427,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.3654559032497016,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2346,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.36393307022795585,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.233,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.36100150356319877,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2082,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.353618760867119,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2225,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.38693605789908997,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2448,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.35658901501103724,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2274,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.378586500313506,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2509,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.37730910423415204,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2452,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.4766422928338597,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2385,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.41144736506560164,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2387,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.35061496677819814,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2376,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.3662182247210059,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2298,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.39031954341127173,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2445,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.3651762418092345,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2747,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.4192409155497121,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2121,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.3830370674346145,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2178,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.4116965086059163,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.3557,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.41586314591198875,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2555,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.37947994484105724,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2042,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.3812418509723405,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2162,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.3881154208156075,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2537,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.3668874474859814,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2777,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.4193875389429049,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2517,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.42352986052369235,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2311,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.38041524069910276,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2137,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.3382235880191095,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.252,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.43141456689553376,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2236,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.4135407095340727,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2318,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.3898713112549109,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2233,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.3959871954234848,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2074,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.432520326375738,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2414,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.41530650128180713,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.227,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.3835629260614596,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2438,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.3693389343746867,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2406,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3651438712782808,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2298,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.349884975197535,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2787,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.3220501945835714,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2149,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.4033429651554876,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2419,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.37641551445426813,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2902,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.35004408405099824,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2467,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.37678432879794593,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.247,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.38110113188779265,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2513,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.3824283173373508,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2387,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.3849215275592779,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2342,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.36700432781514086,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2355,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.36258055831407493,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2466,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.36618857784935965,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.3254,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.33233512136907073,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.244,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.4174597563487877,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.1866,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.416185096033035,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2284,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.3826066119419693,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2371,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.3774411907065305,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2448,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.41818959747299184,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2234,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.4086188719647015,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2535,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.3624056222251327,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2084,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.4049197250042866,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2247,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.42900312436421945,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2231,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.4320607331166067,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2199,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.4033170873329158,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2492,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3504030681693513,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.209,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.406836820431888,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2135,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.37350894158479153,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2304,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.3796353665481668,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2584,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.3875176968814285,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2366,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.39249139645783354,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2612,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.39736774916601947,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2158,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.39248513816529484,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2338,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.3479190574830739,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2423,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.360436700468449,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.233,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.38178530298250085,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2387,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.35826695681650383,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2337,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.362536165546286,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2381,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.37335700177245873,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.218,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.3306880869208319,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2358,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.39333130352102313,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2315,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.3540403183244876,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2863,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.35636789094003263,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2327,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.3420916359807384,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2369,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.3372219997538222,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2375,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.35836024094121244,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.228,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.33477768090665805,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2361,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.3750070984532522,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2131,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.40592807407799464,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2115,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.33051485575724776,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2201,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.3336608092887691,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.251,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.3818347578243733,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2471,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.3482683708452448,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2407,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.34170941134890126,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2235,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.4452763583130281,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2411,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.3690664935047482,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.226,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.35377619469299715,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2078,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.36424797437245127,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2522,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.3928135865780688,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2344,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.37790418646674173,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2501,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.37445876556345803,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2611,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.3309529295756687,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2426,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.3550956912058478,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2153,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.3645392626567163,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2135,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.37282464702746115,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.1933,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.3834892490318246,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.276,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.4299111114195623,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2391,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.3768990922238224,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2475,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.38175603216364173,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2008,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.35349254479351544,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2206,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.3702611547907671,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.1932,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.3485269839492436,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2121,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.3585261874815189,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2573,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.35589019107286357,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2916,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.3672082648461547,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2335,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.39494542771997987,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2186,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.34224536259321164,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.201,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.4120237455359025,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2075,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.31759404280721104,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2484,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.3475326540714425,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2344,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.33073738792777513,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2402,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.3864677389385627,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2146,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.39453658963247845,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2426,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.3697557130715084,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2519,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.32404528181152925,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.207,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.3961335441608937,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2484,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.3896037538275849,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2102,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.4134100164356006,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.1954,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.41721899063496704,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2248,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.36011575117105704,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2247,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.4278682293052621,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2258,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.3691887569153573,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2343,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.3748104442806183,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2369,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.38938319127810567,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2466,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.34349437123192533,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2296,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.4008777559845559,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2066,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.4461379227894572,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1806,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.4013829861598417,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1689,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.33589217439928293,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1619,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3576239315580399,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1556,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.35973158645059133,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1504,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.43414507051963797,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1672,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.4137949658150491,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1616,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.3739494907883889,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1721,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.3570617137747847,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1564,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.36254657346992875,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1683,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.3887091047174778,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1607,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.4055864295246942,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.16,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.3953005927727533,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1442,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.37098533143083373,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1418,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.3156093739217388,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1424,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.3377322829001082,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1806,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.3921572021696728,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1709,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.3802218892712652,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1864,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.38689578010294184,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.2038,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.35061373260946516,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1668,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2972083119682277,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1906,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.3182589356337099,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1843,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.31897948465928705,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1784,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.3196800670407168,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.176,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.35636986898915074,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1638,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.3214320726048193,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1531,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.32277487634923985,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1751,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.30998573269531177,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.1875,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.35421133842108693,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1611,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.35435034828942985,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1879,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.3330574397742846,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1767,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.35428539281067517,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1819,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.33360723382925983,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1677,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.38545469090659923,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1697,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.322028062359536,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1799,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.37674777498397727,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1502,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.34418979893759344,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1611,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.3188972467151948,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1636,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.33728611039794787,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.204,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.3470935796052107,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.203,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.40304323110622636,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1574,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.3217566825919494,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1644,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.33163488034471705,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1468,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.3335221181717988,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1913,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.3136090235218182,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1703,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.37195586262244273,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1581,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.3007162156175051,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1688,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.327200945068625,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2018,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.30006192332082354,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.199,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.32823814087263287,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1817,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.3262948329303315,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1722,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.347328481242527,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1705,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.35787527799312213,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2133,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.32464960262736664,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1456,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.3525447075739603,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1957,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.3233103014522786,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.2129,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.35794491453916755,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1647,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.3232253157342703,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.2103,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.3230131697193597,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1928,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.3223468317576275,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1649,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.33765025240476215,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1896,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.3138349903832304,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1433,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.36744170443291296,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.152,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.38737288448740537,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1784,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.3271325525002661,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1694,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.33456795718716986,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1504,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.3226307105171009,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1741,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2961810767030733,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1777,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.38279126843407374,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1668,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.386994032253301,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.3442,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.33745941543617863,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1633,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.35026793949061064,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1785,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.33200219003624143,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.181,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.36072262134864747,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1784,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.3518734894766553,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1605,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.32068920551548286,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1671,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.3336189983240211,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1764,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.3194338523749552,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1552,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.32108268925130806,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.2089,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.30953722942792117,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1406,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.3522011537549927,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.174,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.3596237205018245,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1746,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.402653511559958,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1551,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.3281308806533522,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.194,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.35896481267965574,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1753,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.3423364234155965,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1846,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.3423749018039515,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1637,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.37647844828781174,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1418,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.3572528255391414,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1778,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.337744098738676,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1771,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.3218201849385497,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1839,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.31787731429722244,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.157,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.4011139318260464,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1605,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.3002709702136994,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1926,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.3536141037530628,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1672,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.31896396118655623,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1728,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.29934577951166325,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1766,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.3973549043402336,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.2622,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.3204052271162904,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1423,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.31529446167764463,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1802,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.36153313327710207,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1424,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.3511399724176934,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1652,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.40121579991642065,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1646,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.3624021745068757,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1706,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.29783758199411564,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1589,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.3272654216171757,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1578,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.3338351878878201,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1819,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.35153818016653854,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1535,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.34188789484779863,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1949,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.3163871733604761,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1752,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.3438345496761507,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1982,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.3447264557681321,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1549,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.31057714792815483,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.2065,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.32626383787945984,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1869,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.3064712820735257,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1417,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.33803401583482234,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.2002,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.3277403162558388,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1593,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.3100599937843426,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1822,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.3192576270675051,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1988,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.3520674270540693,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1468,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.3064539869737425,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1507,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.3067056627170464,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1824,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.3279455108002219,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1573,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.3860491771832849,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1531,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.31996289228177377,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1625,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.35567093196609834,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1721,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.32625351120877416,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1425,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.33786353403945907,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2165,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.3255036026071535,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.2238,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.3385728229497371,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1707,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.33816342436940794,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1499,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.3487222659936048,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1593,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.3112688569617832,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1711,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.35774758008947755,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.165,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.29795867193952574,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.171,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.33178140617250856,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1974,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.34330251767981584,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1646,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.31413374290876417,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.147,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.32874280546965495,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1586,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.34750942201336604,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1382,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.36075559755537706,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.2422,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.3373000326409476,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1947,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.3230978302975328,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1489,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.31842261891098145,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1778,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.3473053770619781,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1725,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.39793014810096755,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.2044,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.32916773323238474,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1547,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.31522121590308516,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1509,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.29426429841047863,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1772,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.35150496476687654,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1875,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.3068050625698932,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1737,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.3211870308263566,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1632,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.3150772830448807,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1831,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.311306134524483,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.152,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.3667541380841685,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1714,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.29220610298688193,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.2064,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.40928626946808827,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1421,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.3249755691257104,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.2265,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.3405223308592626,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.2066,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.3429047181753989,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1757,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.3152780480661016,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1636,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.32022212780171594,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1765,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.3704136180429885,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1779,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.3560687509208485,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1548,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.38619167801745485,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1953,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.3057531203125204,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.179,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.34512202491155836,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1845,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.31598848781440947,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1816,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.33460462091115206,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1603,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.315432530433319,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1861,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.38744840524270663,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1711,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.3297440779333071,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1636,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.38571228264603497,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1568,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.3732295063059201,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1719,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.31413868491361985,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.2016,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.346453616777785,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1589,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.3155283796421409,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1854,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.3141269736875177,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1464,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.3331362874577962,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1557,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.32453203775737016,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1748,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.31488301027125953,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1913,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.3179844979671578,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.157,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.36284301605235136,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1688,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.3001990002709996,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1649,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.32049028811915176,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1655,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.315895454595922,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.183,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.3389869215457553,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1767,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.31655275120424436,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1515,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.33597842911488474,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.155,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.33583241920728946,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1571,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.3452268703247631,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1713,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.315396930553523,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1666,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.3170372419791775,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1606,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.3081232340746034,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.205,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.3355434305328918,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1357,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.3207385066690026,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1777,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.30108048644724117,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1667,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.3636233295648346,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1655,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.35265602738312524,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1586,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.42010333917176435,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1738,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.3006690645042131,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1822,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.3155001623186242,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1791,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.32459092604503403,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.2022,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.37000663036914044,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1552,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.31731221975264523,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.154,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.3114999605002439,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.2049,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.33940495292978023,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1662,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2942617113139344,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1801,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.31844058113666057,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1618,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.2906223596588683,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1764,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.33675489750411963,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.3214,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.3476136508909935,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1684,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.33738793565112923,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1521,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.32810305843238075,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1988,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.336303764037132,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2038,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.3231616014562831,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1779,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.3219303745403928,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1567,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.35696188939061274,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1524,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.3353652205332025,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1639,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.3250286124743562,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1599,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.33425074687901773,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1416,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.3152554125021866,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1584,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.3438829591252694,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1653,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.29435790453134364,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1572,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.31658767991050424,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1633,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.3016899017487661,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1627,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.32433900697908746,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1766,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.3172187755953414,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1971,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.33399379182791494,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1724,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.3371348892670988,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1623,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.3579861156673766,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.2304,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.30876860474292933,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1557,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.30514956428908013,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1823,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.35572944786310196,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1818,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.32437107936109705,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1709,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.32373176291636707,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1512,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.3393166033911238,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1721,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.36285404492490325,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1769,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.3266253460611715,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1613,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.40186491910668753,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1763,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.3415862276573771,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1838,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.3364809564672168,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1738,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2955236217018174,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1582,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.34925982245636905,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1778,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.3503001981533381,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1714,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.3283204174266531,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1624,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.3203489354324029,
+      "learning_rate": 0.0,
+      "loss": 0.156,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1.428173626883965e+18,
+      "train_loss": 0.35492887093953274,
+      "train_runtime": 24806.4226,
+      "train_samples_per_second": 6.369,
+      "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": 1.428173626883965e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}